nonlinear design of 3-axes autopilot for … · short range skid-to-turn surface-to-surface homing...
TRANSCRIPT
NONLINEAR DESIGN OF 3-AXES AUTOPILOT FOR
SHORT RANGE SKID-TO-TURN
SURFACE-TO-SURFACE HOMING MISSILES
By
Abhijit Das
SUBMITTED IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE (BY RESEARCH)
IN
ELECTRICAL ENGINEERING
AT
INDIAN INSTITUTE OF TECHNOLOGY
KHARAGPUR
MAY 2006
Certificate
This is to certify that the thesis entitled “Nonlinear Design of 3-axes Autopilot for
Short Range Skid-to-Turn Surface-to-Surface Homing Missiles” submitted by
Abhijit Das for the award of the degree of Master of Science (by research) is a
record of bonafide research work carried out by him under our guidance and supervision
during the period 2003-2006. The results embodied in this thesis have not been submitted
to any other University or Institute for the award of any degree or diploma.
IIT, Kharagpur
1st May, 2006
Siddhartha Mukhopadhyay
Professor,
Department of EE
Indian Institute of Technology
Kharagpur -721 302, INDIA
Amit Patra
Professor,
Department of EE
Indian Institute of Technology
Kharagpur -721 302, INDIA
INDIAN INSTITUTE OF TECHNOLOGY
Date: May 2006
Author: Abhijit Das
Title: Nonlinear Design of 3-Axes Autopilot for Short
Range Skid-to-Turn Surface-to-Surface Homing
Missiles
Department: Electrical Engineering
Degree: M.S. Convocation: May Year: 2007
Permission is herewith granted to Indian Institute of Technology tocirculate and to have copied for non-commercial purposes, at its discretion, theabove title upon the request of individuals or institutions.
Signature of Author
THE AUTHOR RESERVES OTHER PUBLICATION RIGHTS, ANDNEITHER THE THESIS NOR EXTENSIVE EXTRACTS FROM IT MAYBE PRINTED OR OTHERWISE REPRODUCED WITHOUT THE AUTHOR’SWRITTEN PERMISSION.
THE AUTHOR ATTESTS THAT PERMISSION HAS BEEN OBTAINEDFOR THE USE OF ANY COPYRIGHTED MATERIAL APPEARING IN THISTHESIS (OTHER THAN BRIEF EXCERPTS REQUIRING ONLY PROPERACKNOWLEDGEMENT IN SCHOLARLY WRITING) AND THAT ALL SUCH USEIS CLEARLY ACKNOWLEDGED.
iii
To the loving memory of ’MA’.
From the album of my memory, I remember those early days of my life, a
naughty boy always tried to get rid of his mother’s domination. Now I
can understand that without Her domination, I may not be able to see
these days.
To ’BABA’, who had the arduous task of rising that incorrigible boy.
Acknowledgments
I would like to thank Prof. Siddhartha Mukhopadhyay and Prof. Amit Patra, my su-
pervisors, for their many suggestions and constant support during this research. I am
also thankful to Mr. Ranajit Das, Sc. ”C”, DRDL Hyderabad and Mr. Sourav Patra,
research scholar, Electrical Engineering, for their help through the early years of chaos
and confusion.
Abhijit Das
Systems and Information Lab
Dept. of Electrical Engineering
Indian Institute of Technology
Kharagpur-721302
Abstract
Traditionally, missile autopilots have been designed using linear control approaches with
gain scheduling. Autopilot design is typically carried out in the frequency domain and
the plant is linearized around various operating points on the trajectory. Moreover, three
single axis autopilots are usually designed without considering the interaction among the
motion axes, i.e., the autopilots in each of the three axes are designed independently of
each other. Such designs can not handle the coupling among pitch-yaw-roll channels, es-
pecially under high angles of attack occurring in high maneuver zones. In the last decade,
design of missile autopilots has been extensively studied using modern control design
paradigms such as, robust control, feedback linearization, sliding mode control, singular
perturbation etc. But in most of these studies, realistic factors like fin saturation and lim-
itation of gimbal freedom have not been considered. One therefore cannot really evaluate
the performance and relative merits of these methods in practical applications. This work
presents a nonlinear multivariable approach to the design of an autopilot for a realistic
missile that overcomes these difficulties. At first, exact input-output (IO) feedback lin-
earization and decoupling have been carried out for the dynamic IO characteristics of the
inner rate loop of the pitch and yaw channels. In the process, the missile dynamics also
becomes largely independent of flight conditions such as missile velocity, air density etc.
This enables the design of scalar linear controllers for the inner rate loops. In this work
the superiority of the new nonlinear multi-input multi-output (MIMO) autopilot over a
traditional autopilot has been demonstrated through realistic simulation results in pres-
ence of closed loop guidance and seeker. However performance deteriorates when the plant
model is perturbed, due to aerodynamic uncertainties, from the nominal model. A robust
IO linearization technique is therefore needed to tackle the aerodynamic uncertainties in
the system. In this study H∞ and sliding mode techniques have been applied to design
a robust controller for the feedback linearized plant. A nonlinear Luenberger observer
for the missile airframe dynamics has also been designed for estimating the unmeasured
states that are required for feedback linearization. All the design and simulations have
been carried out in a realistic environment and thus the results presented in this thesis
may be useful for designing more accurate missile system in future.
Contents
1 Introduction 1
1.1 Missile basics: an overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Missile coordinate system . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Angle of attack(AoA), side-slip angle and aero-phi . . . . . . . . . . 2
1.1.3 Missile model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.4 Aerodynamic data used in 6−dof model . . . . . . . . . . . . . . . 6
1.1.5 Skid-to-turn steering law . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Missile subsystems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2 Guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2.1 Guidance sensors . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.2.2 Types of guidance . . . . . . . . . . . . . . . . . . . . . . 13
1.2.2.3 Main features of PN guidance . . . . . . . . . . . . . . . . 14
1.2.2.4 Design of trajectory/ guidance scheme . . . . . . . . . . . 14
1.2.2.5 Turning radius capability . . . . . . . . . . . . . . . . . . 16
1.2.2.6 Switching sight line range at PN start . . . . . . . . . . . 16
1.2.2.7 On board velocity computation . . . . . . . . . . . . . . . 16
1.2.3 Autopilot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.2.3.1 Lateral autopilot . . . . . . . . . . . . . . . . . . . . . . . 18
1.2.3.2 Roll autopilot . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.2.4 Missile control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
CONTENTS ii
1.2.4.1 Aerodynamic . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.2.4.2 Jet reaction forces . . . . . . . . . . . . . . . . . . . . . . 21
1.3 Typical problems with classical autopilots for highly maneuvering missiles . 22
1.3.1 Sources of roll disturbance . . . . . . . . . . . . . . . . . . . . . . . 22
1.3.2 Lateral and Roll aerodynamic Characteristics . . . . . . . . . . . . 24
1.3.3 Problems associated with roll rate . . . . . . . . . . . . . . . . . . . 25
1.4 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.5 Literature survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.5.1 Feedback linearization based controller . . . . . . . . . . . . . . . . 29
1.5.2 Modern gain scheduling based controller . . . . . . . . . . . . . . . 30
1.5.3 Sliding mode controller . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.5.4 Linear and nonlinear robust controller . . . . . . . . . . . . . . . . 33
1.5.5 Model based adaptive controller . . . . . . . . . . . . . . . . . . . . 34
1.5.6 Nonlinear observer . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.5.7 Literature related to other control approaches . . . . . . . . . . . . 36
1.6 Contribution of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.7 Organization of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2 Autopilot Design with Input-Output Linearization by Feedback 39
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2 Brief theory and application to an nonlinear missile model . . . . . . . . . 40
2.2.1 Case I : Input-output linearization with q, r, p as outputs . . . . . . 42
2.2.1.1 Zero dynamics analysis . . . . . . . . . . . . . . . . . . . . 46
2.2.1.2 Simulation results . . . . . . . . . . . . . . . . . . . . . . 47
2.2.1.3 Decoupling of three axes motion . . . . . . . . . . . . . . 49
2.2.1.4 Input-Output Linearization . . . . . . . . . . . . . . . . . 49
2.2.2 Case II: Input-output linearization with w,v and p as outputs . . . 62
2.2.2.1 Formulation of the problem for the STT missile . . . . . . 62
2.2.2.2 Simulation results . . . . . . . . . . . . . . . . . . . . . . 65
2.3 Nonlinear Luenberger observer design . . . . . . . . . . . . . . . . . . . . . 68
CONTENTS iii
2.3.1 Introducion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.3.2 Observer construction . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.3.3 Application to a realistic STT missile model . . . . . . . . . . . . . 72
2.3.3.1 Gain ’K’ calculation . . . . . . . . . . . . . . . . . . . . . 73
2.3.4 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.4 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3 H∞ Control of Feedback Linearized Inner Rate Loop Dynamics 80
3.1 Introducion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.2 Problem formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.2.1 A linear matrix inequality approach to H∞ control for designing K
in pitch plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.2.2 Design Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.2.3 Application to the Nonlinear Missile . . . . . . . . . . . . . . . . . 91
3.2.3.1 Uncertainty and disturbance modeling between y and v . . 91
3.2.3.2 Uncertainty and disturbance model for nonlinear missile . 94
3.2.3.3 Robust controller formulation . . . . . . . . . . . . . . . . 96
3.2.3.4 Simulation results . . . . . . . . . . . . . . . . . . . . . . 98
3.3 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4 Sliding Mode control after Feedback Linearization 106
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.2 Formulation of sliding mode controller . . . . . . . . . . . . . . . . . . . . 108
4.2.1 Step II: Formulation of sliding mode control law for designing v . . 108
4.2.1.1 Specifying sliding surfaces . . . . . . . . . . . . . . . . . . 109
4.2.1.2 Achieving sliding condition . . . . . . . . . . . . . . . . . 110
4.2.2 Application to the STT missile model . . . . . . . . . . . . . . . . . 111
4.2.2.1 Simulation Results . . . . . . . . . . . . . . . . . . . . . . 113
4.3 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5 Conclusions 128
CONTENTS iv
A A Brief Theory of Feedback Linearization 131
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
A.1.1 Input-output linearisation . . . . . . . . . . . . . . . . . . . . . . . 134
A.2 Multi input multi output systems . . . . . . . . . . . . . . . . . . . . . . . 139
A.2.1 Feedback Linearization of MIMO Systems . . . . . . . . . . . . . . 139
A.2.2 Zero-dynamics and control design . . . . . . . . . . . . . . . . . . . 144
B LMI Approach to H∞ Control 146
B.1 The Theory of H∞ Control based on LMI Approach . . . . . . . . . . . . . 146
B.1.1 Singular value decomposition . . . . . . . . . . . . . . . . . . . . . 146
B.1.2 Norms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
B.1.3 Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
B.1.4 Basis Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
B.1.5 L2 space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
B.1.6 L∞ space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
B.1.7 H2 space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
B.1.8 H∞ space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
B.1.9 Packed Matrix Notation . . . . . . . . . . . . . . . . . . . . . . . . 151
B.1.10 Robust Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
B.2 A Linear Matrix Inequality Approach to H∞ Control . . . . . . . . . . . . 157
B.2.1 Brief theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
B.2.2 Advantages of LMI . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
B.2.3 Basic Idea about LMI . . . . . . . . . . . . . . . . . . . . . . . . . 159
B.2.4 Matrices as variable . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
B.2.5 Lyapunov’s Inequality . . . . . . . . . . . . . . . . . . . . . . . . . 161
B.3 Stabilizing Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
B.3.1 System Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
B.3.2 Stabilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
B.4 H∞ Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
B.4.1 Two important matrix inequalities . . . . . . . . . . . . . . . . . . 171
CONTENTS v
B.4.2 The KYP Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
B.4.3 Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
B.4.4 Controller reconstruction and connections . . . . . . . . . . . . . . 190
List of Figures
1.1 Anti Tank Missile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Fixed body frame (x, y, z) and definitions . . . . . . . . . . . . . . . . . . . 3
1.3 Sideslip Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Maneuver Plane Roll Orientation; Aerophi(φ) . . . . . . . . . . . . . . . . 5
1.5 Skid to turn Steering law . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Definitions of positive pitch, yaw and roll Control . . . . . . . . . . . . . . 10
1.7 Basic Steering / Roll Control . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.8 Autopilot functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.9 Missile seeker showing angular geometry . . . . . . . . . . . . . . . . . . . 13
1.10 PN guidance trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.11 General trajectory for a surface-to-surface STT missile . . . . . . . . . . . 15
1.12 Comparison of missile velocity and estimated velocity . . . . . . . . . . . . 17
1.13 Block diagram lateral autopilot . . . . . . . . . . . . . . . . . . . . . . . . 19
1.14 Block diagram of a roll autopilot . . . . . . . . . . . . . . . . . . . . . . . 21
1.15 Rolling moment coefficient characteristic . . . . . . . . . . . . . . . . . . . 24
1.16 Side force coefficient characteristic . . . . . . . . . . . . . . . . . . . . . . . 25
1.17 Yawing moment coefficient characteristic . . . . . . . . . . . . . . . . . . . 26
1.18 Rolling moment coefficient characteristic during PN maneuver . . . . . . . 27
1.19 Normal Force coefficient characteristics . . . . . . . . . . . . . . . . . . . . 28
2.1 Block diagram of the system with FBLC for outputs q, r and p . . . . . . . 43
2.2 Boundedness of u,v and w in zero dynamic condition . . . . . . . . . . . . 48
LIST OF FIGURES vii
2.3 Autopilot response with step command . . . . . . . . . . . . . . . . . . . . 50
2.4 Decoupling in the three channels . . . . . . . . . . . . . . . . . . . . . . . 51
2.5 Evidence of linearization in pitch channel . . . . . . . . . . . . . . . . . . . 52
2.6 Evidence of linearization in yaw channel . . . . . . . . . . . . . . . . . . . 53
2.7 Evidence of linearization in roll channel . . . . . . . . . . . . . . . . . . . . 54
2.8 Pitch and yaw latax profiles . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.9 Effective pitch yaw roll deflection . . . . . . . . . . . . . . . . . . . . . . . 56
2.10 Fin deflections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.11 Fin deflection rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.12 Gimbal angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.13 Alpha and beta profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.14 Force and moment coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.15 Block diagram of the system with output w, v and p . . . . . . . . . . . . 63
2.16 Autopilot response with step command . . . . . . . . . . . . . . . . . . . . 66
2.17 Autopilot response with guidance command . . . . . . . . . . . . . . . . . 67
2.18 Effective Pitch yaw roll deflection . . . . . . . . . . . . . . . . . . . . . . . 68
2.19 Estimated and true α . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.20 Estimated and true β . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.21 Estimated and true α for full flight time . . . . . . . . . . . . . . . . . . . 77
2.22 Estimated and true β for full flight time . . . . . . . . . . . . . . . . . . . 78
3.1 Block diagram of the system representing robust feedback linearization for
outputs q, r and p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.2 Structure of robust control K in pitch plane . . . . . . . . . . . . . . . . . 83
3.3 General feedback arrangement . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.4 Mixed sensitivity configuration . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.5 Complete model structure of the system and noisy signal . . . . . . . . . . 93
3.6 Characteristics of ∆pitch . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.7 Characteristics of ∆yaw . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.8 Characteristics of ∆roll . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
LIST OF FIGURES viii
3.9 Comparison of FBLC and robust controller in pitch plane . . . . . . . . . . 100
3.10 Comparison of FBLC and robust controller in yaw plane . . . . . . . . . . 101
3.11 Comparison of FBLC and robust controller in roll plane . . . . . . . . . . . 102
3.12 Control deflection comparison . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.13 Control deflection rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4.1 Block diagram of the system representing robust feedback linearization for
outputs q, r and p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.2 Linearization in pitch channel . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.3 Linearization in yaw channel . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.4 Linearization in roll channel . . . . . . . . . . . . . . . . . . . . . . . . . . 117
4.5 Control deflection comparison . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.6 Pitch, yaw and roll rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
4.7 α and β . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4.8 Gimbal angle profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.9 Pitch and yaw latax . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.10 Pitch,yaw and roll deflections . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.11 Fin demands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.12 Fin deflection rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
4.13 Force and moment Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 126
B.1 System interconnections: (a) Series connection, (b) Inversion, (c) Parallel
connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
B.2 Closed loop with loop transfer H . . . . . . . . . . . . . . . . . . . . . . . 152
B.3 Small gain stability in Nyquist space . . . . . . . . . . . . . . . . . . . . . 153
B.4 Baby small gain theorem for additive model error . . . . . . . . . . . . . . 154
B.5 Control sensitivity guards stability robustness for additive model error . . . 154
B.6 Baby small gain theorem for multiplicative model error . . . . . . . . . . . 155
B.7 Complementary sensitivity guards stability robustness for multiplicative
model error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
B.8 General feedback arrangement . . . . . . . . . . . . . . . . . . . . . . . . . 162
LIST OF FIGURES ix
B.9 Input-output stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
List of Tables
1.1 Calculations of aero-coefficients from αR and φ . . . . . . . . . . . . . . . . 7
1.2 Calculations of aero-coefficients on the basis of Table 1.1 variables . . . . . 8
1.3 Calculations of input-coefficients . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Calculations of aero-derivatives . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1 Variation in aerodynamic coefficients and thrusts in x− y − z directions . . 99
4.1 Variation in aerodynamic coefficients and thrusts in x− y − z directions . . 113
List of Symbols
(gx)f Gravity component in fin frame along x direction in m/s2
(gy)f Gravity component in fin frame along y direction in m/s2
(gz)f Gravity component in fin frame along z direction in m/s2
α Angle of attack in rad
β Side slip angle in rad
δω1 to δω4 Wing deflection due to misalignment in rad
δc1 to δc4 Fin deflection due to misalignment in rad
∆Xf , ∆Yf , ∆Zf Lateral C.G shift w.r.t nose of the missile in m
δp, δy, δr Control Deflection in pitch, yaw and roll respectively in rad
η1, η3 Control deflection for fins 1 and 3 in rad
φ Maneuver plane roll orientation in rad
ξ2, ξ4 Control deflection for fins 2 and 4 in rad
CL Rolling moment coefficient
CN Normal force coefficient
LIST OF TABLES xii
CS Side force coefficient
CDO Force Coefficients due to drag
CLp Rolling damping coefficient
Clzetaw Control moment coefficient for wing
Clzeta Control moment coefficient for control surface or fin
Cmq Damping moment coefficient due to pitch rate
Cmr Damping moment coefficient due to yaw rate
CmαDamping moment coefficient due to change in angle of attack
CmβDamping moment coefficient due to change in side slip
Cmpcg Pitching moment coefficient w.r.t C.G
Cnδ Control effectiveness
Cnαw Control force effectiveness for wing
Cnycg Yawing moment coefficient w.r.t C.G
CYr Damping force coefficient due to yaw rate
CYβDamping force coefficient due to β
Czq Damping force coefficient due to pitch rate
CzαDamping force coefficient due to change in angle of attack
D Diameter of the missile in m
fz, fy Lateral accelerations (latax) along z and y axes respectively m/s2
IXX Moment of inertia about x in kgm2
LIST OF TABLES xiii
IY Y Moment of inertia about y in kgm2
IZZ Moment of inertia about z in kgm2
m Mass of the missile
p, q, r Angular velocities round the x, y, z-axes respectively in rad/sec
Q Dynamic pressure of the missile in Pascal
S Surface area of the missile in m2
TX Thrust component in x direction in Newton
TX Thrust component in x direction in Newton
TY Thrust component in y direction in Newton
TZ Thrust component in z direction in Newton
Tmx Thrust misalignment along x in Newton
Tmy Thrust misalignment along y in Newton
Tmz Thrust misalignment along z in Newton
U, v, w Linear velocities along x, y and z axes respectively in m/sec
Vm Resultant velocity along trajectory in m/sec
x, y, z Body frame coordinates
Xcg C.G of missile w.r.t nose in m
Xcpωp C.P of wing w.r.t nose of the missile in m
Xcpc C.P of fin w.r.t nose of the missile in m
XLTP , XLTY Distance of hinge line from nose in elevation and azimuth in m
LIST OF TABLES xiv
CG Center of gravity
FBLCL FBLC with linear rate loop controller
FBLCR FBLC with robust rate loop controller
FBLCSM FBLC with sliding mode rate loop controller
FBLC Feedback linearizing controller
PN Proportional Navigation
LOS Line of sight
The units of aerodynamic coefficients are commensurate with the units of other
variables
Chapter 1
Introduction
A missile is a projectile that is, something thrown or otherwise propelled. The earliest
form of a missile was probably a stone that, when thrown forcefully through the air, would
follow a ballistic path. Adding gunpowder to a projectile, resulted in the rocket, the first
powered, but as yet unguided, missile [34]. Rockets were first invented in medieval China
(Circa 1044 AD) but its first practical use for serious purpose other then entertainment
took place in 1232 AD, by the Chinese against the Mongols at the siege of Kai-Feng-Fue.
Thereafter from 1750 AD to 1799 AD Haider Ali and Tipu Sultan (Sultan of Mysore, in
south India) perfected the rocket’s use for military purposes, very effectively using it in
war against British colonial armies. It was not until the early 1900s that guided missile
development was begun. In sofar as the missile as we know it today is concerned, the
impetus came primarily from World War II and in particular from German scientists.
Immediately after the war there was a rapid growth in missile activity through out the
world. Although this work is concerned with a particular type of short range missile
autopilot design.
Chapter 1. Introduction 2
1.1 Missile basics: an overview
The missile that is considered in this thesis is a short range, surface to surface, Skid-to-
Turn, homing missile and for an example some relevant parts of that kind of a missile are
shown in Figure 1.1.
Figure 1.1: Anti Tank Missile
1.1.1 Missile coordinate system
A fixed body frame coordinate system (x, y, z) is introduced and shown in Figure 1.2. Its
origin is placed at the center of gravity. The movement and position of the missile are
described in this coordinate system. The resultant speed of the missile, Vm, is given by
Vm =√
U2 + v2 + w2 where U is usually the major contributor as the x axis is generally
aligned with the thrust direction, or nearly so.
1.1.2 Angle of attack(AoA), side-slip angle and aero-phi
Angle of attack (α) is a term used in aerodynamics to describe the angle between the
missile’s body x axis and the direction of airflow wind, effectively the direction in which
Chapter 1. Introduction 3
Figure 1.2: Fixed body frame (x, y, z) and definitions
the Center of Gravity (C.G) of the missile is currently moving, as shown in this schematic
1.2. From the Figure the mathematical expression of α can be derived as α = tan−1(
wU
).
The sideslip angle(β) is defined by the angle between the velocity vector of the vehicle and
the longitudinal axis also shown in Figure 1.3 and from Figure 1.2, β = tan−1(
vU
). The
orientation of the tail fins with respect to the body axes for a cross or ’X’ configuration
and their conventional numbering are shown in Figure 1.4. The plane defined by the
velocity vector V and the body x- axis is called the maneuver plane. This is the plane
along which the resultant force of air acts on the missile body, in the direction opposite to
the velocity vector Vm. The term aerophi(φ) denotes orientation of the maneuver plane
w.r.t the fin axis f2-f4, the angle being measured from f2. The fin axis system is the body
axis system rotated 45 around the x−axis so that the y and z body axes align with the
fin planes. Figure 1.4 shows the aerophi and side force CS as well as yawing moment Cny
are defined 90 anticlockwise w.r.t maneuver plane (αR).
1.1.3 Missile model
The standard nonlinear differential equations relating the forces and moments acting on
a missile with its motion components are given by,
Chapter 1. Introduction 4
Figure 1.3: Sideslip Angle
Force equations
U = rv − qw + 1m
[TX −QSCDO] + (gx)f
v = pw − rU + QSm
Cy(α, β, Vm
Vs, δy) + TY
m+ (gy)f
w = qU − pv + QSm
Cz(α, β, Vm
Vs, δp) + TZ
m+ (gz)f
(1.1)
Moment equations
p = 1IXX
QSDCl(α, β, Vm
Vs, δr) + Tmx
IXX
q = 1IY Y
QSDCm(α, β, Vm
Vs, δp) + IZZ−IXX
IY Ypr + Tmy
IY Y
r = 1IZZ
QSDCn(α, β, Vm
Vs, δy) + IXX−IY Y
IZZpq + Tmz
IZZ
(1.2)
where, Cy and Cz are force coefficients and Cl, Cm and Cn are the moment coefficients
including aerodynamic variables and inputs. From Equation 1.1, we can subdivide the
force Euler equations in the following way: Force = coriolis component+ thrust compo-
Chapter 1. Introduction 5
Figure 1.4: Maneuver Plane Roll Orientation; Aerophi(φ)
nent + aerodynamic component + gravity component. For example, if we consider the
expression in Equation 1.1 of v, then, coriolis component=pw− rU , thrust component =TY
m, aerodynamic component = QS
mCy(α, β, Vm
Vs, δy), gravity component = (gy)f . Similarly
for moment equations we can write, moment = aerodynamic component + thrust com-
ponent + centrifugal component. For example, if we consider the expression in Equation
1.2 of q, aerodynamic component = 1IY Y
QSDCm(α, β, Vm
Vs, δp), thrust component = Tmy
IY Y,
centrifugal component = IZZ−IXX
IY Ypr. More detailed nonlinear 6-DOF equations which is
used for realistic simulations are given in [3], [29], [30], [13], [11], [12] and are expressed
as follows:
Chapter 1. Introduction 6
Force equations
U = rv − qw + 1m
[TX −QSCDO] + (gx)f
v = pw − rU + 1m
[TY + QSCNBn + CNBs + Clzie + D2Vm
(−CYββ + CYrr)]
+(gy)f + QSm
[Cnαw(δω2 − δω4) + Cnδ(δc2 − δc4)/2]
w = qU − pv + 1m
[TZ + QSCNAn + CNAs + Cleta + D2Vm
(−Czqq − Czαα)]
+(gz)f − QSm
[Cnαw(δω1 − δω3) + Cnδ(δc1 − δc3)/2]
(1.3)
Moment equations
p = 1IXX
[−IXXp + TmX + QSD D2Vm
CLpp− Clzeta
2δR + CL]
− QSD2IXX
[Clzetaw(δω1 + δω2 + δω3 + δω4) + Clzeta(δc1 + δc2 + δc3 + δc4)]
+ 1IXX
[(TY + QS(CNBn + CNBs + CLzie + D2Vm
(−CYββ + CYrr)))∆Zf
−(TZ + QS(CNAn + CNAs + Cleta + D2Vm
(−Czqq − Czαα)))∆Yf ]
q = 1IY Y
[(QSD[−Cmeta + CMAn + CMAs + D2Vm
(Cmqq + Cmαα)] + TmY )
−(IXX − IZZ)pr − IY Y q]− QSIY Y
[Cnαw(δω1 − δω3)(Xcpωp −Xcg)
+Cnδ
2(δc1 − δc3)(Xcpc −Xcg)] +
∆Zf
IY Y[−TX − (−QSCDO)]
r = 1IZZ
[(QSD[Cmzie + CMBn + CMBs + D2Vm
(Cmrr − Cmββ)]+
TmZ)− (IY Y − IXX)pq − IZZr]− QSIZZ
[Cnαw(δω2 − δω4)(Xcpωp −Xcg)
+Cnδ
2(δc2 − δc4)(Xcpc −Xcg)] +
∆Yf
IZZ[TX + (−QSCDO)]
(1.4)
1.1.4 Aerodynamic data used in 6−dof model
The aerodynamic coefficients and data used in the above six degree of freedom (i.e. a
set of six independent displacements that specify completely the displaced or deformed
Chapter 1. Introduction 7
position of the body or system) equations are mostly functions of angle of attack, side
slip, aero-phi and Mach number. From the wind tunnel test data, it is possible to express
these aerodynamic coefficients as polynomials in α, β etc. Some of these coefficients such
as CN , CS etc can be directly computed as a polynomial in αR and φ as shown in Table
1.1.
Aero-coefficient (force) CalculationsCN fn(φ, αR) = A2α
2R + A1αR
CS fn(φ, αR) =m∑
i=0
diαiR
Aero-coefficient (moment) CalculationsCmp fn(φ, αR) = B2α
2R + B1αR
CL fn(φ, αR) =p∑
i=0
CiαiR
Cny fn(φ, αR) =n∑
i=0
eiαiR
Table 1.1: Calculations of aero-coefficients from αR and φ
Some other aerodynamic coefficients are computed in fin frame given in Table 1.2 from
the aero-coefficients listed in Table 1.1.
Some of the coefficients which are associated with the inputs are calculated as given
in Table 1.3. These coefficients are mainly functions of Mach number.
The coefficients and aerodynamic derivatives given in Table 1.4 are taken to be con-
stant for simulations carried out in this thesis. Although these are also function of Mach
number. Since Vm is assumed constant, these are also assumed constants.
1.1.5 Skid-to-turn steering law
Two principal steering policies have been used in the design of tactical missile systems,
namely, skid-to-turn (STT) control and bank-to-turn (BTT) control. The former has
been the customary choice for most missiles in the past and the same steering law has
been used for the missile that we have considered in this thesis.
Chapter 1. Introduction 8
Aero-coefficient(force) in fin frame CalculationsCNAn −CN × cos(φ)CNAs −CS × sin(φ)CNBn −CN × sin(φ)CNBs CS × cos(φ)Aero-coefficient(moment) in fin frame Calculations w.r.t c.g.
Cmpcg Cmp + CN × Xcg
D
Cnycg Cny + CS × Xcg
D
CMAn Cmpcg × cos(φ)CMBn −Cmpcg × sin(φ)CMAs Cnycg × sin(φ)CMBs Cnycg × cos(φ)
Table 1.2: Calculations of aero-coefficients on the basis of Table 1.1 variables
Input-coefficient in fin frame CalculationsCnδ fn (Mach no), value=2.43Cleta Cnδ × δp
Clzie Cnδ × δy
CmetaCleta×(XLTP−XCG)
D
Cmzie−Clzie×(XLTY −XCG)
D
Table 1.3: Calculations of input-coefficients
Aerodynamic derivatives and Clzeta ValueClzeta fn (Mach no), value=1.0Czq 85.0Czα
52.0Cmq −420.0Cmα
−220.0CLp −10.0
Table 1.4: Calculations of aero-derivatives
Chapter 1. Introduction 9
Figure 1.5: Skid to turn Steering law
For an STT system as shown in Figure 1.5, the missile is roll stabilized in space and
performs aerodynamic maneuvers in each of two orthogonal planes, pitch and yaw, to
achieve a resultant maneuver in any desired planes. Generally these types of missile are
made to be roll stabilized and its four control surfaces are placed in a cruciform pattern as
shown in Figure 1.6. In this case, all of the four control surfaces are used simultaneously
for pitch, yaw and roll. If the surface deflections are considered to be positive when
the panel is rotated clockwise as one looks down the hinge line toward the body, then
pitch,yaw and roll control for the two cases can be defined in terms of individual surface
deflections by the formulas for δp = −δ1+δ2+δ3−δ44
, δy = δ1+δ2−δ3−δ44
and δr = δ1+δ2+δ3+δ44
Chapter 1. Introduction 10
Figure 1.6: Definitions of positive pitch, yaw and roll Control
1.2 Missile subsystems
A missile consists of several subsystems. The most closely related subsystems are shown
in the block diagrams of Figures 1.7 and 1.8, which describe the principal function of each
subsystem and list the principal elements of each. A short description of each subsystem
follows
Figure 1.7: Basic Steering / Roll Control
Chapter 1. Introduction 11
1.2.1 Navigation
When the missile is launched, its position, attitude, speed, acceleration and rotation are
to be known. The navigation subsystem updates these variables during the flight. This
is done by using sensor data and strap down navigation algorithms. The variables are
supplied to the guidance subsystems.
1.2.2 Guidance
Although we have not designed the guidance law for the missile, but for the sake of com-
pleteness, in the following sections, some relevant terms are described. The guidance
subsystem computes the error between the missile’s actual and desired courses, computes
the corrections necessary to reduce or nullify the error according to a chosen guidance law,
and gives commands to the autopilot to activate the controls to achieve the corrections.
These commands may be for lateral accelerations, angular rates, etc. The navigation
system contains sensors that provide information on the missile’s actual and desired tra-
jectories, noise filters and a computer in guidance system to process the information into
the commands to the autopilot.
Figure 1.8: Autopilot functions
Chapter 1. Introduction 12
1.2.2.1 Guidance sensors
The guidance system operates based on the relative target-pursuer kinematics, which are
defined by the following variables:
1. Line of sight (LOS) angles in azimuth and elevation
2. LOS rates
3. Range between the target and the interceptor
4. Range rate etc
These variables can be measured by different sensors based on indirect measurements.
One of the important sensors is seeker. This sensor is mounted on the radome of an
interceptor (in case of homing engagement) and detects the relative position vector of the
target with respect to the platform where it is mounted.
The line of sight(LOS), λ, is defined as the angle between a line from the center of the
seeker antenna to the target, and some arbitrary non-rotating(e.g. inertial) reference line.
LOS is generally defined for both azimuth and elevation planes. Figure 1.9 shows the
LOS in elevation plane only. Commonly, it is convenient to select this reference equal to
the LOS position at the beginning of the homing guidance phase. Consequently, λ(t) at
time t is the total change in the angular position of the LOS relative to the initial LOS.
Here θm is defined as the pitch angle in elevation. Furthermore the angular position of
the antenna centerline is defined by the gimbal angle θh in elevation. Therefore, the LOS
angle in elevation (λ) is given by
λ = θm + θh + ε
where, ε is the true boresight error, that is, the error between antenna centerline and line
of sight to the target. Same treatment can also be applied for azimuth plane.
Chapter 1. Introduction 13
Figure 1.9: Missile seeker showing angular geometry
1.2.2.2 Types of guidance
Many different guidance laws have been developed and used depending upon the capa-
bilities of the interceptor and the target and the engagement scenario. The examples of
some typical guidance laws are as follows:
1. LOS angle guidance
2. LOS rate guidance
3. Pursuit guidance
4. Attitude Pursuit guidance
5. Proportional navigation Guidance
6. Optimal guidance
Chapter 1. Introduction 14
Here, in this thesis we are concerned about the Proportional Navigation (PN) guidance
only as it is used for surface-to-surface STT missile that we have considered here.
1.2.2.3 Main features of PN guidance
1. Sightline angle (and rate) vary at the start of engagement
2. Later the sightline moves parallel to itself so the sightline rate → 0 and a shrinking
interception triangle is produced
Figure 1.10: PN guidance trajectory
1.2.2.4 Design of trajectory/ guidance scheme
For a typical short range surface-to-surface missile trajectory, two distinct trajectory/guidance
policies may be evolved based on target range. Up to a certain small target range; im-
pact angle optimized trajectory/ guidance policy has been considered, which maximizes
impact angle in the presence of seeker gimbal angle limit, airframe turning capability and
Chapter 1. Introduction 15
subsystem’s lags and is known as attitude hold phase. For target range above that
specified range, maximum height constraint on the trajectory is to be satisfied through
a ’Height limited trajectory/guidance policy’. However, during initial phase, to optimize
the impact angle, missile trajectory is made to pitch up as sharply as seeker gimbal angle
limit allows, through a gimbal angle hold(GAH) phase. The trajectory in elevation
plane is depicted in Figure 1.11. Switching from gimbal angle hold phase to PN guid-
ance phase is done as soon as the required sight line range or equivalently sight line rate
is reached. As the switching sight line range RSW is obtained, the missile jumps into
unsaturated PN phase from saturated PN phase correcting the heading error (gimbal
angle at switching) between sight line and flight path with minimum radius of turn ca-
pability. Required homing distance (dH) is available for settling the errors/transients in
Figure 1.11: General trajectory for a surface-to-surface STT missile
the final homing phase. Minimum requirement of homing distance is established with
normalized simulation of seeker based homing guidance loop for achieving the required
miss distance. The required RSW is derived from the known value of maximum gimbal
angle freedom θ, minimum radius of turn of the configuration R and the required hom-
Chapter 1. Introduction 16
ing distance dH. Therefore,Rsw is obtained as a fundamental parameter independent of
target range, prospective impact angle and velocity profile. In general for a surface-to-
surface STT missile, R is constrained by a tight limit on configuration size arising out of
tube launching requirement. For example, the finally achieved design value corresponding
to a configuration of a typical short range surface-to-surface STT missile may be raised
up to 5.5g maximum latax for a base velocity of 200m/s in one plane which is taken as
elevation and the demand in azimuth may be limited to 2g during maximum elevation
demand period.
1.2.2.5 Turning radius capability
The minimum turning capability of missile configuration R which is a function of Cnα,
maximum angle of attack permissible and air density is obtained for different range and
temperature. The maximum latax limit is calculated based on αmax(≈ 18) and low-
est velocity condition for each range and temperature and stored on OBC into a two
dimensional look up table as a function of range and temperature. The turning radius
R(R = U2m
latax) during SPN phase is obtained for each range and temperature.
1.2.2.6 Switching sight line range at PN start
A range independent relation for switching sight line range Rsw at PN start is obtained
from geometry so that the heading error θg is corrected with turning radius R, keeping
homing distance dH for settling of errors. Rsw is a function of turning radius R, dH and
gimbal angle at PN switching θg.
1.2.2.7 On board velocity computation
From cost and size consideration, a typical short range surface-to-surface homing missile
has no velocity sensor on board. Based on 6-DOF simulations, considering thrust versus
time profiles for each temperature and PN event, empirical formula based velocity algo-
rithm is developed on board for each range and temperature. The average acceleration
Chapter 1. Introduction 17
during booster and sustainer phase is stored on OBC into 2−D look up table as a func-
tion of range and temperature. The estimated velocity is matched with missile nominal
velocity at each range and temperature in Figure 1.12. This estimated velocity closely
matches the missile forward velocity u and is assumed to be known during flight.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
40
60
80
100
120
140
160
180
200
Normalised time
Res
ulta
nt v
eloc
ity (
Vm
)and
Est
imat
ed v
eloc
ity (
Ues
t)
Vm
Uest
Typical surface−to−surface missile
Figure 1.12: Comparison of missile velocity and estimated velocity
1.2.3 Autopilot
The autopilot receives commands from the guidance computer and processes them into
commands to the controls such as deflections or rates of deflection of control surfaces or
jet controls through action of servomechanisms. To provide the deflection at a desired
rate, the servomechanism motors must contend with the inertia of the control device
and the torque about its shaft. Since the autopilot will convert guidance commands of
acceleration or angular rate into control commands, it must have a way to determine
Chapter 1. Introduction 18
if the accelerations and angular rates provided by deflecting the controls are meeting
the guidance commands. Thus this subsystem will have accelerometers for measuring
the achieved accelerations and gyroscopes for measuring the angular positions or rates.
Depending on where these instruments are placed, the autopilot may have to provide
corrections to the instrument readings to obtain the true accelerations of the missile’s
center of gravity (CG) and true angular rates about its principal axes. Classically, the
missile autopilot comprises three independent autopilots, one for each lateral direction,
namely pitch and yaw, and one for roll.
1.2.3.1 Lateral autopilot
The missile autopilot controls the acceleration normal to the missile body. In this case
study, the autopilot structure is a three-loop design using measurements from an ac-
celerometer located ahead of the missile’s center of gravity and from a rate gyro to
provide additional damping. Figure 1.13 shows the classical configuration of an autopi-
lot. The controller gains are scheduled based on Mach number and tuned for robust
performance at an altitude of around 4000 meters (10000 feet) in general for small range
surface-to-surface STT missiles. According to Figure 1.13, there are three feedback
loops present. Since control of acceleration is required, the outermost loop is closed
by an accelerometer. This outer accelerometer loop has the lowest bandwidth of the
three loops. The innermost rate-damping loop is required to damp the response of
the bare airframe, which has an under-damped resonance in the stable case. In addi-
tion, the innermost rate-damping loop has a wide bandwidth for damping the poles
of airframe. The last one, known as synthetic stability loop, improves the high frequency
poles of the autopilot if the airframe is stable, and enables the autopilot to tolerate some
instability of the airframe. Furthermore the synthetic stability loop in Figure 1.13 effec-
tively feeds incremental pitch angle back to the fin servos, thereby moving the autopilot
closed-loop poles, corresponding to the bare airframe poles. Three important parameters
of the 3-Loop Autopilot (Figure 1.13) i.e., system damping, time constant of desired Latax
demand transfer function fz
fyand open-loop cross over frequency of innermost loop need
to be tightly controlled. Controlling system damping ensures that guidance system is not
Chapter 1. Introduction 19
Figure 1.13: Block diagram lateral autopilot
sensitive to body rate coupling. Selecting and controlling system time constant to the
specified value as per guidance loop requirement tightly means that adequate perfor-
mance in terms of miss distance can be achieved. Finally, controlling open-loop cross
over frequency means that we will have a robust design, which is not overly sensitive
to un-modeled high frequency dynamics.
1.2.3.2 Roll autopilot
The basic function of the roll autopilot is to make the missile roll stabilized, that
is, to provide missile stabilization of roll attitude about the longitudinal axis. This is
accomplished by sensing roll rate, and using the fins (or wings) defection by an amount
sufficient to counteract roll disturbances. Moreover, the response of the system must
be sufficiently fast to prevent the accumulation of significant roll angles. A block diagram
of the roll autopilot is shown in Figure 1.14. One common type of roll autopilot utilizes
a spring-restrained rate gyroscope for measurement of roll rate, in conjunction with
proportional-plus-integral (PI) compensation in the autopilot amplifier, in order to give
Chapter 1. Introduction 20
the approximate equivalent of roll-rate plus roll-angle feedback. Other roll autopilot de-
signs utilize a free vertical gyroscope as an attitude reference. That is, in order to maintain
a desired roll angle, an attitude reference must be used. A simple PI controller is used
for controlling the missile roll dynamics. To overcome the uncertainties during high angle
of attack, a roll autopilot design concept based on on-board disturbance estimation
has been formulated. The roll autopilot (Figure 1.14) estimates the disturbance torque on
board based on flight measurements and generates the required feed-forward command to
the actuator for correcting the effect of disturbances. The expression for disturbance
torque (Td) is obtained from the equation of motion in roll as:
Td = IXX p + 2QSDCLζδfb
where, δfb = effective fin deflection for roll obtained from control surface deflections
measured. Q = dynamic pressure. IXX = moment of inertia (roll axis). The required
feed-forward command δff to cancel the disturbance torque is obtained from the torque
balance equation as: Feed-forward control torque= disturbance torque. Or,
2QSDCLζδff = IXX p + 2QSDCLζδfb
∴ δff = δfb + IXX
2QSDCLζ× p
, (1.5)
Therefore, the required feed-forward command is obtained from effective fin deflection for
roll δfb measured in flight and roll angular acceleration measured from flight measure-
ment of p. In order to cater for estimation/ measurement error, 80% of the feed-forward
command as obtained in (1.14), is applied in the proposed design .
1.2.4 Missile control
The choice of control system will depend on the use of the missile, flight path and the
height at which it will operate. There are two main types of control system used:
Chapter 1. Introduction 21
Figure 1.14: Block diagram of a roll autopilot
1.2.4.1 Aerodynamic
Aerodynamic change of flight path is obtained by using a wing at incidence to the direction
of flight. The wing is set at incidence in one of two ways:
• By using a rotatable wing.
• By using a fixed wing, but a movable control surface to set the whole missile at
incidence. The control surfaces may be at the front or rear of the missile.
1.2.4.2 Jet reaction forces
Change of flight path by jet reaction forces is achieved in one of two ways:
• By altering the direction of thrust of the propulsion unit, either by swivelling the
whole unit, or by deflecting the gas stream by vanes or similar devices.
Chapter 1. Introduction 22
• By the use of separate auxiliary jet reaction units.
1.3 Typical problems with classical autopilots for highly
maneuvering missiles
The classical autopilots suffer from the problems of instability due to their highly nonlin-
ear and uncertain aerodynamic characteristics. At large angles of attack (AoA) or at high
maneuvering zones, missile flight dynamics become highly nonlinear, due to significant
amount of cross-coupling between three axes. Also, almost all missiles have significant
nonlinearities, associated with limitations in the movement of aerodynamic control sur-
faces. The other forms of uncertainties are namely as model variations in mass, inertia
and center of gravity positions, aerodynamic tolerances, air data system tolerances, struc-
tural modes, failure cases, etc. Due to the issues outlined above, all flight control laws
are required to undergo a rigorous certification (or clearance) process before being evalu-
ated in flight tests. The search for worst-case control inputs is an important part of this
process.
Generally, missiles with clean configuration (no protrusions) exhibit smooth aerodynamic
characteristics. However, in many missiles due to functional requirements, some ad-
ditional objects such as wire tunnels, lifting lugs, nozzles, antenna are required for a
complete configuration. The existence of these devices modifies the flow field around the
lifting surfaces and in turn longitudinal and lateral aerodynamic characteristics of the
configuration. That is why the wind tunnel data shows undesired lateral aerodynamic
characteristics (side force, yawing moment and rolling moment) for a small surface to
surface missile configuration particularly at higher angles of attack.
1.3.1 Sources of roll disturbance
For low moment of inertia about roll axis of the missile, the effect of aerodynamic asym-
metry on performance is highly appreciable due to proportionately higher cross coupling
forces / moments with respect to control forces/ moments. With simultaneous pitch and
Chapter 1. Introduction 23
yaw maneuvers present, appreciable cross coupling roll torque at around lateral frequency
occurs mainly due to aerodynamic asymmetry. Several flight trial data show high roll rate
oscillation during both the maneuver phases. This roll motion is detrimental to the mis-
sion of the missile. Hence, it is imperative to control this undesired rolling motion within
the permissible limits. Flight roll rate recorded during SPN phase shows the recorded roll
rate profile of a flight test data during saturated PN phase where the angles of attack
is quite high. It shows that, during sustained ’PN’ phase; roll oscillation builds up from
lower value to a large value as 100 /sec at end. The FFT of this roll rate shows that the
frequencies of oscillation are mainly ≈ 3Hz (Roll-lateral frequency) and 5/6 Hz (roll fre-
quency). Yaw latax and rate also oscillate at frequency ≈ 3Hz. The oscillation gradually
builds up from lower value to higher value and has been seen in flight roll rate and yaw
dynamics profiles during the SPN phase. From the flight test data, it can be observed
that with desired pitch down maneuver, angles of attack builds up to higher value. Due
to uncertainty of rolling moment at higher angles of attack, these roll disturbance torque
can not be controlled accurately by the roll autopilot as the autopilot BW is limited by
the hardware. During SPN phase, roll rate increases. As fy = v + Ur − pw, therefore,
fy increases as w increases in maneuver to a high value. Hence, fy increases with high
fz leads to increase in missile maneuver plane roll orientation φ from desired orientation.
Aerodynamic characteristic shows that rolling moment coefficient CL increases with roll
orientation variation from desired values. This higher value of CL leads to higher roll
rate. Therefore, it is a vicious cycle of cross coupling of p ↑→ fy ↑→ φ ↑→ p ↑. Hence it
can be remarked that high frequency roll disturbance torque (3 to 7Hz) is found during
maneuver phases of several flight records. The roll autopilot design can not cater to this
disturbance torque as the autopilot stiffness at that frequency can not be improved due
to low stability. This disturbance torque leads to high roll rate oscillation during the
maneuver phases. Secondly, the onboard roll disturbance torque estimation scheme gives
inaccuracy due to high frequency of disturbance torque and actuator/computation delay.
This leads to inaccuracy in control torque generated. The unbalanced torque leads to
high roll oscillation in flight.
Chapter 1. Introduction 24
1.3.2 Lateral and Roll aerodynamic Characteristics
Aerodynamic forces and moments are important inputs for the controller design. The
wind tunnel test data obtained (used in 6 − DOF simulation) are thoroughly analyzed
and presented here for different roll orientations and resultant angles of attack. Figure
1.15 represents the rolling moment coefficient for different angles of attack during up
and down maneuver with roll orientation variation [52]. It is noticed that roll moment
uncertainty increases for angles of attack more than 15 . Also rolling moment value is
more for negative angles of attack (pitch down maneuver).
Figure 1.15: Rolling moment coefficient characteristic
Figure 1.16 represents the side force coefficient and Figure 1.17 represents the yawing
moment coefficient for different angles of attack during up and down maneuver with
±5 roll orientation variation. Side force and Yawing moment uncertainty with small
roll orientation variations increase with higher angle of attack. Therefore control force
requirement to cancel the disturbances will be more for higher angle of attack.
Chapter 1. Introduction 25
Figure 1.16: Side force coefficient characteristic
Figure 1.18 represents the rolling moment coefficient during pitch down maneuver for
different angle of attack. It is noticed that the rolling moment variation with respect
to small roll orientation is large for a high angle of attack. This is the reason for high
frequency roll disturbance at high angle of attack. The magnitude of roll disturbance
increases with the increase of angle of attack. Figure 1.19 shows the characteristics of
the normal force coefficients. The normal force coefficient increases almost proportionally
with the angle of attack specially in high angle of attack zones.
1.3.3 Problems associated with roll rate
For a high maneuvering missile with a small moment of inertia about roll axis, the roll
rate needs to be well controlled with in the specified limit. The problems associated with
unwanted roll disturbances are listed out after detailed investigations through 6 −DOF
Chapter 1. Introduction 26
Figure 1.17: Yawing moment coefficient characteristic
simulations. These are :
• The roll-pitch-yaw cross coupling effect. The roll induced yaw rate helps to build
up azimuth gimbal angle ( gimbal angle freedom in azimuth plane is only ±12 ).
• The lateral and roll aerodynamics force and moment characteristics show that with
roll rate, roll orientation changes and side force disturbances increase. To control
the yaw disturbances, yaw control requirement increases. And hence, with desired
pitch requirement, control goes to saturation level (mainly one pair of control fin)
with high yaw and roll disturbances.
• The roll rate and roll induced yaw rate are coupled to guidance demand through the
seeker body rate coupling. This causes high miss distance as guidance commands
are erroneous.
Chapter 1. Introduction 27
Figure 1.18: Rolling moment coefficient characteristic during PN maneuver
• The turning radius during SPN phase is limited with the limit of angle of attack. It
is already mentioned that with high angle of attack, the uncertainty of roll distur-
bances increases. The early switching and shallow turning deteriorates the terminal
performances in terms of impact angle and miss distance. Hence minimum range
increases in order to keep the angle of attack within permissible low value.
1.4 Motivation
In summary, the traditional three-loop autopilot which is being widely used in many
aerospace industry for most missile autopilot designs, suffers from the following problems.
• Coupling between the pitch-roll axes tends to cause high roll rates whenever a sharp
pitch maneuver is commanded.
Chapter 1. Introduction 28
Figure 1.19: Normal Force coefficient characteristics
• Due to high roll rate an induced yaw rate is generated as a result of roll-yaw coupling.
• To correct for the undesirable yaw rate, control deflections are commanded by the
yaw autopilot. This results in saturation of the deflection of one or more fins in the
presence of a simultaneous high pitch and roll channel input.
• Further, increased yaw rate causes the side slip angle to increase, causing gimbal
angle saturation in the azimuth plane. This can result in track loss of the missile if
the target goes out of the view of the seeker.
• Aerodynamic coefficients, especially side force coefficients (CS) and yawing moment
coefficient (Cny), increase in value, which in turn effectively increase the values of
yaw rate and yaw deflections. Thus the effect feeds on the cause and an instability
mechanism is set up.
The main motivation of this work is to achieve superior performance in terms of impact
angle, roll rates and miss distance over the traditional three-loop autopilot. In addition
to this some other performance criteria are to be achieved. For examples, roll angle has
Chapter 1. Introduction 29
to be kept within specified bounds due to asymmetric look angle freedom of the seeker in
elevation plane and image rotation vis-a-vis pixel addresses.
1.5 Literature survey
During the last decade, a significant research effort has been contributed in the area of
nonlinear missile autopilot design. The nonlinear design approaches found during survey
are briefly classified into following methods.
1. Feedback linearization based controller
2. Modern gain scheduling based controller
3. Sliding mode controller
4. Robust controller
5. Model based adaptive controller.
6. Nonlinear observer
7. Other approaches
1.5.1 Feedback linearization based controller
In [53] a new nonlinear controller has been proposed considering the coupling effect on a
more or less generalized missile model. In the paper it has also been sshown that through
a kind of partial linearization along with singular perturbation techniques, it is possible
to transform the non-minimum phase missile dynamics into an approximate input-output
feedback linearizable system and thereby achieving a linear input-output dynamics, which
is decoupled as well as independent of flight conditions. An almost similar methodology
has been developed in [52]. However, some aspects have not been considered in these two
Chapter 1. Introduction 30
papers like actuator saturation and robust stability of the system. Another related paper
[62] demonstrates some nonlinear methodology like feedback linearization approach along
with a linear controller on a short range air-to-air missile. Here the superiority of the non-
linear autopilot over the traditional autopilot based on gain scheduling technique has also
been shown through simulation studies. The example of robust feedback linearization has
been given in the [35]. In this paper a HAVE DASH missile system is being linearized with
input-output linearization technique and then a sliding mode control has been designed
for robustness. This paper presents a novel approach to design robust feedback lineariza-
tion with sliding mode control. The states are assumed to be known. In [89] input-output
approximate linearization of a non-linear sixth order system has been studied. The order
of the zero dynamics has been made forcefully zero by increasing the relative degree of
the system. The design procedure may not be applicable for complicated missile systems
where the aerodynamic coefficients can not be represented as a function of state variables
of the system. The paper [84], presents a novel systematic approach for the autopilot
design of STT missiles. First, the nonlinear model of a STT missile is partially linearized
via functional inversion techniques and then, the additional set-point tracking controller
is designed by the well-known LMI approach. The stabilization conditions are given in
terms of LMI’s. Feedback Linearization is the way of converting a nonlinear system into
a linear one and thus may be considered as the best choice for designing the nonlinear
missile autopilot. The above review revealed that the problem of decoupling has not yet
been well addressed with all practical considerations like fin saturation, exceeding of az-
imuth gimbal angle limitations etc. The other performance criteria like minimizing miss
distance, maximizing the impact angle are also not taken into account. Moreover in most
of the cases the missile or system model taken is not a realistic one.
1.5.2 Modern gain scheduling based controller
The paper [90] deals with gain scheduled control system synthesis, applied to a missile
autopilot design problem. Here low order linear autopilots designed at discrete operating
points using classical control techniques are gain scheduled. Also, the Mach number,
Chapter 1. Introduction 31
modeled as a measured, time-varying exogenous signal, is viewed as a disturbance to be
rejected in the linear design phase, leading to improved steady-state tracking performance.
[73] presents two nonlinear controller designs for a bank-to-turn, air-to-air missile. The
first controller is a gain-scheduled H∞ design and the second is a nonlinear dynamic
inversion design using a two time scale separation. Comparisons in simulation results
for the two methods are also given. Different linear to nonlinear approaches on missile
autopilot design are reported in [16] and the comparisons among the simulation results
of the various controller has been also been presented. Another paper [71] presented
sequentially the tuning procedures for two schemes of lateral acceleration autopilots and
for one roll autopilot. They also presented the design relationship between plant and
controller. [55] has designed the missile autopilot based on gain scheduling with sequential
tuning for lateral as well as roll autopilot. [62] presents linear and nonlinear procedures
for designing the missile autopilot via feedback linearization as well as gain scheduling
techniques. [60] shows how the time-scale separation helps to improve the robustness
of feedback linearized autopilots by simplifying the feedback linearization maps, and by
permitting the design of low-order controllers. This paper presents the development of
three distinct time-scale separation schemes for the design of feedback linearized missile
autopilots. Gain scheduling based controller will be more effective for the systems where
the order of nonlinearity and uncertainties is not that much of concern. And so the
direct application of this method specially to a aerodynamic missile may not provide
effective results in practice. In those cases adaptive gain scheduling methods may be
useful depending upon complexity of aerodynamic uncertainties. In this thesis we have
not discussed about this method.
1.5.3 Sliding mode controller
A comparison via Monte-Carlo simulations is made of Sliding Mode Observers versus
Kalman Filter in the homing missile guidance system using different guidance laws has
been given in [80]. It has been shown that Sliding mode observer shows more accurate
results in terms of miss distance than the Kalman Filter due to its noise. The paper [74]
Chapter 1. Introduction 32
presents an optimal sliding mode control design method for a nonlinear system with a
cascade or two-loop structure.The control design method is demonstrated on a nonlinear
model of an F-16 aircraft. But the aircraft model has been taken to be much simpler
than any other practical model. [82], [33], [81], [59], [50] and [43] are the examples of
time varying sliding mode control, traditional high order sliding modes, dynamic sliding
modes, terminal sliding modes and sliding mode observer with gain adaptation. Simu-
lation results have been presented along with a reusable launch vehicle model, satellite
formation as well as integrated guidance and control application of aerodynamic missiles.
A sliding mode controller is shown in [78] for an integrated missile autopilot and guidance
loop. Motivated by a differential game formulation of the guidance problem, a single slid-
ing surface, defined as the zero-effort miss distance, is used. [96] presents a good example
of second order sliding mode control along with backstepping approach application on a
aerodynamic missile. Another three papers [42], [94] and [36] give the example of applica-
tion of Fuzzy-Neural and adaptive fuzzy sliding-mode control and traditional sliding mode
approach to a BTT missile. The example of robust feedback linearization has been given
in [35]. In this paper a HAVE DASH missile system is being linearized with input-output
linearization technique and then a sliding mode control has been designed for robustness.
This paper presents a novel approach to design robust feedback linearization with sliding
mode control. The states are assumed to be known. [4] shows the robust smooth MIMO
sliding mode controller with finite reaching time. A sliding mode estimator is also pro-
vided to eliminate the effects of unmodeled, bounded disturbances and uncertainties. In
[79], an integrated two-loop guidance and flight control system is designed to incorporate
a variety of guidance strategies and robustly enforce them regardless of target maneuvers,
atmospheric disturbances, and dynamic uncertainty of airframe actuator. Backstepping
approach and sliding mode approach have been introduced simultaneously for designing
the controller. Aerodynamic coefficients of most of the missiles in general express their
uncertain characteristics during flight in atmosphere. The application of sliding mode con-
trol may be one of the solutions to get rid of this problem. The literature survey shows
that the only problem regarding the implementation of sliding mode control is chattering.
In this thesis we have used this method, to cater to aerodynamic uncertainties. [35] shows
Chapter 1. Introduction 33
the application of sliding mode control for a HAVE DASH II missile and here we have
tested in a more realistic skid-to-turn homing missile model.
1.5.4 Linear and nonlinear robust controller
[57], [31] and [24] present a robust multivariable autopilot design for missile system. In the
first case, a canonical robust control design formulation is introduced followed by three
robust autopilot designs. The next one shows the gain scheduled control performance
along with H∞ control application. The third and last one deals with the H∞ as well as
µ synthesis approaches successfully. The simulation shows that the designs achieve good
response against significant kinematic and inertia couplings and aerodynamic parameter
variations. In both cases it is assumed that a linear model for the nonlinear missile can
be found properly. In [37] a design method is proposed namely, dynamic robust recursive
control, in order to obtain output tracking performance for a general class of nonlinear
systems. Application of this method was then demonstrated for the design of a robust
autopilot controller for a nonlinear missile model with structured uncertainty. Simulation
results for this system demonstrated that the dynamic robust recursive design was able
to achieve significant performance improvements over a more conventional I/O linearized
controller design. But all the simulation results has been given using a second order
example. [68] applies a result on output feedback guaranteed cost control of stochastic
uncertain systems to the problem of designing a missile autopilot. Missile autopilot model
has been taken as a simple one. A total least squares approach is used to fit the data
to a norm bounded uncertain system model. [56] presents a robust H2 and H∞ control
design for a HAVE DASH II missile system using a generalized Hamiltonian formulation.
The design endures significant kinematic and inertia couplings, aerodynamic parameter
variations and high frequency flexible effects. [64] presents a roll-yaw autopilot for a non-
axis symmetric missile model to robustly decouple roll tracking from yaw regulation. The
controller structure exhibits considerable inherent robustness and decoupling capability
without high actuator activity, providing a useful framework for dealing with truly NLTV
problems. The time-varying closed-loop PD-spectrum allows real-time adjustment of
Chapter 1. Introduction 34
bandwidth, thereby achieving in flight tradeoffs among performance, energy consumption,
robustness and other operation concerns. [65] also deals with the robustness analysis
of the missile autopilot. A different approach has been pointed out for designing the
nonlinear robust controller although simulation results has been given on the basis of a
second order nonlinear missile model. In the paper [57] robust multivariable autopilot
designs are examined through a canonical robust control design formulation. Although
the simulation results are given for HAVE DASH missile system, coupling effect has not
been taken fully into account. Another example of application of state feedback control
has been given in [87]. In this paper two different control laws for roll rate and acceleration
of a air-to-air missile are shown along with simulation results. It is also shown that the
results are applicable for both STT and BTT missile models with time varying flight
conditions. The estimator for α and β have been left for future study. Robust H∞ control
is one of the most popular approaches from 1992 for tackling the uncertainties. But as
the application of this method directly to a nonlinear system is not as easy as for a linear
one. Literature survey shows that direct application of H∞ control to a realistic short
range missile model considering different types of uncertainties and disturbances in full
amount is rare. We have applied the same method in this thesis to a feedback linearized
missile model.
1.5.5 Model based adaptive controller
Applications of Adaptive control and Neural Network in missile autopilot design can be
found in the literature surveys [72]. Some research work is also found in the field of opti-
mal/classical approach of missile autopilot design [66]. Model based adaptive controller is
effective for system with uncertainties. We have not considered this method in our design.
1.5.6 Nonlinear observer
[22] presents a nonlinear approach of observer design to Inverted Pendulum model in the
form of a extended Kalman Filter. But the observer gains are restricted for stability rea-
sons. [5] gives a high-gain observer for a large class of nonlinear systems. This observer is
Chapter 1. Introduction 35
designed to work for systems with bounded state. [10] presents the traditional approach of
designing the nonlinear state observer. Stability of the observer is also proved along with
the drift-observability property shown to be a sufficient condition for existence of the expo-
nential observer. A new approach of designing the nonlinear observer has been proposed
in [15]. Carleman linearization technique has been used to get exact input-output lin-
earization and the simulation results have been given as an example of good performance
with a simple nonlinear system. [88], [17], [45] and [77] give the idea of designing non-
linear observer using backstepping, high gain extended Kalman Filter, and constant gain
exponential approach. [86] is a geometric study of finding general exponential observers
for nonlinear systems. Using center manifold theory, it derives necessary and sufficient
conditions for general exponential observers for Lyapunov stable nonlinear systems. [44]
presents the nonlinear state observer based on input-output linearization. Assumption
has been made that the internal dynamics of the system is globally stable. [49] describes
a linearizing feedback adaptive control structure which guarantees high quality regulation
of the output error in the face of unknown parameters. The simulation results have been
given based on a CSTR without considering the jacket dynamics. [47] presents a nonlinear
reduced order Luenberger observer for estimating the states of a nonlinear system. [39]
and [40] present a reduced order nonlinear observer with application to a permanent mag-
net synchronous motor. The authors show through simulation results that the proposed
observer can estimate the rotor position and velocity very accurately. Similar papers [25],
[85] apply the same reduced order observer with different formulation on chaotic systems
as well as reactors. [23], [58] and [2] give a different observer model for a large class of
MIMO nonlinear systems. The gain of the observer can be tuned with a single parameter
variation. Observer has been also tested in continuous tank reactors and bio-reactors. In
[69], the problem of robust state observation is tackled. A high-gain observer is employed
to carry out the state estimation of a continuous time uncertain nonlinear system subject
to external perturbations of stochastic nature. The papers [93] and [18] give two different
types of observer design. The first one with sliding mode observer and the second one
gives a large overview with Extended Luenberger and Extended Kalman filter approach
and along with their application to a bio-chemical process. Most of the linear and non-
Chapter 1. Introduction 36
linear feedback control requires state feedback for computing the plant input. Therefore
if all of the system states are not measurable then a state observer is required. The liter-
ature survey shows that implementation of nonlinear observer is common to most of the
nonlinear systems. But, often the measured outputs of the system are such that they can
not be easily expressed in terms of system states although they are directly or indirectly
functions of states. Observer design for these types of system is rare. In this thesis we
have designed a nonlinear Luenberger observer for these types of systems.
1.5.7 Literature related to other control approaches
Some more approaches to designing the missile autopilots have been reported in [75]
another different approach has been taken for autopilot deign using Linear Parameter
Varying (LPV) techniques. Here firstly the missile dynamics has been brought into an
LPV form via a state transformation rather than the usual coordinate transformation.
And finally an inner-outer loop decomposition is applied. This paper presents a good
method for outer loop design of the missile autopilot keeping the robustness of the inner
loop. But only 10% variation in aerodynamic coefficients has been taken into account
which in practical cases can be almost 100%. [97] and [98] present a missile autopilot
design using extended mean assignment (EMA) control technique for linear time vary-
ing (LTV) systems. The EMA control technique is based on a new series D-eigenvalue
(SD eigenvalue) concept in a way similar to the conventional pole placement design for
linear time invariant (LTI) systems. The nonlinear dynamics of the missile is rendered
into a linear one that is tractable by the EMA control technique via the classical lin-
earization along a nominal normal acceleration profile, followed by a linear coordinate
transformation. Simulation results are also presented for the zero input stabilization.
But the classical linearization technique may not be applicable for all complex nonlinear
missile model. [67] presents plant input mapping digital redesign methods for the dig-
ital implementation of missile autopilots. The authors show along with the simulation
results that the new controller response is good enough. But the coupling effect and the
model uncertainties have not been considered in full. The paper [7] presents a novel ap-
Chapter 1. Introduction 37
proach of designing nonlinear autopilot design for an air-to-air missile. The authors have
considered actuator saturations and show good tracking performance through simulation
results. [76] presents an example of system inversion and manifold invariance applied for
designing missile autopilot. It has been shown that the classical notions of invariance and
inversion, together with tools from the nonlinear regulator theory can be used to design
globally stabilizing control laws for general nonlinear systems. An approach to integrated
guidance/autopilot design for missiles is proposed in [95]. Integrated guidance/autopilot
design has been formulated based on variable structure techniques and simulation results
have been given for an anti-vessel missile.
1.6 Contribution of the thesis
The thesis has proposed a nonlinear 3-axes control structure that uses robust linear and
nonlinear control designs. The controller has been tested both within an ideal simulation
environment built using MATLAB, as well as a highly detailed and realistic simulation
environment coded in FORTRAN. The performance of the proposed controller has the
following features:
1. Good decoupling between the three axes during pitch maneuver.
2. Minimized roll rate, yaw deflection, rate of change of aerodynamic coefficients even
during high angle of attack maneuver.
3. Estimation of the unmeasured states of the system using a nonlinear observer re-
quired for the computation of nonlinear feedback.
4. Design of a robust H∞ control law that retains performance even with aerodynamic
uncertainties.
5. Design of a robust sliding mode controller to tolerate the uncertainties caused due
to aerodynamic coefficients.
Chapter 1. Introduction 38
1.7 Organization of the thesis
This thesis has been organized into five chapters.
Chapter 1 provides an introduction to guidance and control system of a missile. It also
presents typical trajectories, guidance schemes and structure of autopilots. Finally,
this section presents the limitation of the existing autopilot and thereby motivates
the work done here. This chapter also carries a literature survey for autopilot design
methods.
Chapter 2 presents input-output feedback linearization technique for designing rate loop
control law.
Chapter 3 presents robust feedback linearizing control law with H∞ technique for de-
signing the inner rate loop of the autopilot.
Chapter 4 presents the design of inner rate loop of the autopilot based on sliding mode
technique.
Chapter 5 draws the final conclusions and outlines the scope for future study.
Chapter 2
Autopilot Design with Input-Output
Linearization by Feedback
2.1 Introduction
Typical short range tactical missiles suffer from the high roll rates due to uncertain aero-
dynamic coefficients with increasing angle of attack and coupling among pitch, yaw and
roll. The effort with three loop autopilot has been seen to be inefficient particularly
when decoupling is the point of attraction as in this conventional case of three separate
decoupled and linear autopilots, a maneuver in one axis causes disturbances in the oth-
ers and effectively causes fin saturation for high lateral accelerations demand. To tackle
the decoupling problem the use of nonlinear multivariable method such as input-output
linearization technique may be effective. Feedback linearization approach to nonlinear
control design has attracted a great deal of research interest in recent years. The central
idea is to algebraically transform a nonlinear system dynamics into a (fully or partially)
linear one, so that linear control techniques can be applied. This chapter provides an
application of input-output linearization technique to a small range surface-to-surface
skid-to-turn homing missile to show the decoupling among pitch-yaw-roll axes along with
its effect on missile overall performance such as control deflections, side slip angle, azimuth
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback40
gimbal angle etc.
2.2 Brief theory and application to an nonlinear mis-
sile model
The following discussion of input/output linearization for multi-input/multi-output sys-
tems is adapted from [48],[83],[61],[32],[8] and Appendix A. Consider the square system
x = f(x) + g1(x)u1 + ..... + gm(x)um
y1 = h1(x)
· · ·ym = hm(x)
(2.1)
Where x is the state vector,uis(i = 1, ....., m) are control inputs, yjs(j = 1, ....., m) are
outputs, and , f and gis are the smooth vector fields, and hjs are smooth scalar functions.
To obtain an input-output relation of the system 2.1, we differentiate the jth output, for
m = 1,
yj =∂yj
∂xx =
∂yj
∂x[f(x) + g(x)u] = Lfh1 + Lgh1u (2.2)
and for MIMO case,
yj = Lfhj +m∑
i=1
(Lgihj)ui, (2.3)
Where, Lfhj is defined as Lie derivative of hj w.r.t f(x) and is equal to∂hj
∂xf(x). If
Lgihj(x) = 0 ∀i ,then the inputs do not appear and we have to differentiate again. Assume
that rj is the smallest integer such that at least one of the inputs appear in y(rj)j , then
y(rj)j = L
rj
f hj +m∑
i=1
LgiL
rj−1f hjui, (2.4)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback41
With LgiL
rj−1f hj(x) 6= 0 for at least one i. If we perform the above procedure for each
output yj, we can obtain a total of m equations in the above form, which can be written
compactly as
y(r1)1
· · ·· · ·y
(rm)m
=
Lr1f h1
· · ·· · ·Lrm
f hm
+ E(x)
u1
· · ·· · ·um
(2.5)
where the m×m matrix E is defined as
E(x) =
Lg1Lr1−1f h1 · · · · · · LgmLr1−1
f h1
.... . .
......
. . ....
Lg1Lrm−1f hm · · · · · · LgmLrm−1
f hm
(2.6)
The matrix E(x) is called the decoupling matrix for the MIMO system. If the decoupling
matrix is non-singular around a point x0, then the input transformation
u = −E−1
Lr1f h1
· · ·· · ·Lrm
f hm
+ E−1
v1
· · ·· · ·vm
(2.7)
yields a linear differential relation between the output y and the new input v
y(r1)1
· · ·· · ·y
(rm)m
=
v1
· · ·· · ·vm
(2.8)
Note that the above input-output relation is decoupled, in addition to being linear. If
the new inputs v1, v2, v3 are designed in such a way that the control law only affects the
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback42
corresponding yjs and not the others, then such a control law of the form 2.7 is called a
decoupling control law, or non-interacting control law. As a result of the decoupling, one
can apply SISO linear control laws on each y− v channel in the above decoupled dynamics
to construct tracking or stabilization controllers. The relative degree of the system can
be computed as ρ = r1 + r2 + .... + rj + .... + rm. If ρ = n, i.e. if the relative degree
of the system is equal to the order of the system, then that system is known as a full
input-output linearizable system. On the other hand, if ρ < n, then it is possible to make
the system partially input-output linearized, as all of the states may not appear in the
linearized form. These remaining states appear as internal dynamics of the system. This
internal dynamics may not be ’visible’ from the linearized plant but the whole system is
stable only if the internal dynamics is. This aspect has been discussed later and in detail
in Appendix A.
2.2.1 Case I : Input-output linearization with q, r, p as outputs
Figure 2.1 describes the schematic block diagram of the system with feedback linearizing
controller or FBLC. fzd, fyd and φd are demanded lateral accelerations and roll angle
respectively while fz, fy, φ are the sensed ones. qd, rd and pd are the demanded rates
computed from error dynamics of lateral accelerations and roll angle. FBLC is the feed-
back linearizing controller which computes u with the help of a new input vector v. This
v is being computed from rate error dynamics. Now from the theory of feedback lin-
earization, it can be shown that the relation between v and rate dynamics could be made
linear and decoupled as shown in the dotted box. Beside that v is to be computed in
such a way, that good tracking performance in the inner rate loop as well as in outer
acceleration loop is maintained. In Figure 2.1, the block ’LC’ stands for ’Linear Con-
troller’, that is used to compute v. Let the state, output and the input vectors be defined
as, v = [v1, v2, v3]T , x = [U, v, w, p, q, r]T , u = [δp, δy, δr]
T . Now the whole system can be
represented as,
x = f(x) + g(x)u
where f(x) and g(x) are described as follows:
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback43
Figure 2.1: Block diagram of the system with FBLC for outputs q, r and p
f(x) =
(rvm−qwm+TX−QSCD0−Q11GRm)
m(√
2TY +√
2QSCNBn+√
2QSCNBs−√
2QS D2Vm
Cyβ
β+√
2QS D2Vm
rCyr+√
2pwm−√2rUm−GRmQ21
−GRmQ31+√
2FMY Q) √2m
−(−√2qUm+√
2pvm−GRmQ21+GRmQ31−√
2TZ−√
2QSCNAn−√
2QSCNAs+√
2QS D2Vm
qCZq
+√
2QS D2Vm
αCzα ) √2m
(−IXXp+Tmx+QSDCL+QSD D2Vm
CLP p)
IXX−(−QSDCMAnD−QSDCMAsD−QSD D
2VmDCmq q−QSD D
2VmDCmα α−TmyD+prDIXX−prDIZZ+IY Y qD)
D×IY Y(QSDCMBnD+QSDCMBsD+QSD D
2VmDrCnr−QSD D
2VmDCm
ββ+TmzD−qpDIY Y +qpDIXX−IZZrD)
DIZZ
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback44
g(x) =
0 0 0
0QSCnδ
m0
−QSCnδ
m0 0
0 0 −QSDClzeta
IXX(−QSDCnδ
XLTP +QSDCnδXCG)
DIY Y0 0
0 − (QSDCnδXLTY −QSDCnδ
XCG)
DIZZ0
For easy reference we can write, f(x) = [fi(x)]T
i=1→6
and g(x) = [gij(x)]i=1→6,j=1→3
.
From the theory of input-output linearization we obtain, with reference to the equa-
tions A.37, A.38 and A.39, that,
y1
y2
y3
= M(x) + E(x)
δp
δq
δr
where,
M(x) =
Lfh1(x)
Lfh2(x)
Lfh3(x)
and
E(x) =
Lg1h1 Lg2h1 Lg3h1
Lg1h2 Lg2h2 Lg3h2
Lg1h3 Lg2h3 Lg3h3
Note that, as r1 = 1, r2 = 1, r3 = 1, the relative degree of the system obtained in our case
is ρ = 3. It can be seen from the above equations that to get the desired input output
relation we need to differentiate the outputs only once and thus the E(x) matrix obtained
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback45
is non-singular for the whole flight time. In this particular case
M(x) =
f5
f6
f4
and E(x) =
g62 0 0
0 g51 0
0 0 g42
It has been seen from numerical simulations that, diagonal terms of the g(x) matrix
remain non-zero for all typical flight trajectories and over the whole flight time. Thus
the E(x) matrix is non-singular for the whole flight time. As the off-diagonal terms are
all zero, the outputs become decoupled with each others and coupled only one of the
inputs. So we see that with the outputs chosen as q, r and p, the input-output relation
of the nonlinear missile model can made input-decoupled for every instant of the flight by
providing appropriate inputs. The design of the control law will be as follows
u = −E−1M + E−1
v1
v2
v3
(2.9)
to linearize the system leading to
y1
y2
y3
=
v1
v2
v3
(2.10)
A tracking controller for this single-integrator relation can be designed using linear control
techniques. For this particular case, defining the tracking error
e =
y1 − y1d
y2 − y2d
y3 − y3d
=
q − qd
r − rd
p− pd
(2.11)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback46
and choosing the new input v1, v2 and v3 as
v1 = y1d + k1(y1d − y1) = qd + k1(qd − q)
v2 = y2d + k2(y2d − y2) = rd + k2(rd − r)
v3 = y3d + k3(y3d − y3) = pd + k3(pd − p)
(2.12)
with k1, k2 and k3 being positive constants, the tracking error dynamics of the closed loop
system is given by
e1 + k1e1 = 0
e2 + k2e2 = 0
e3 + k3e3 = 0
which is exponentially stable and e(t) converges to zero exponentially. Of course,the
stability of internal zero dynamics needs to be an internally stable closed-loop system.
2.2.1.1 Zero dynamics analysis
The basic theory of zero dynamics has also been adopted from Appendix A and [1]. The
design of linear inputs v1,v2 and v3 could be done in a straightforward manner using a
classical linear control strategy such as pole placement / optimal control (LQR) provided
the zero dynamics of the system are asymptotically stable, i.e., the system is minimum-
phase.
To compute the zero dynamic stability, first we define the internal dynamics as
U = f1
v = f2 + g2u2
w = f3 + g3u3
(2.13)
, where the functions fi and gi have been defined earlier. The stability of the internal
dynamics has been studied with the help of zero dynamic analysis through simulation.
That means, we have to investigate the characteristics of zero dynamic state variables
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback47
such as U, v, w have been studied giving an input u0 =
δp0
δy0
δr0
to the plant dynamics
such that the outputs q, r, p and their derivatives remain zero. The Figure 2.2 represents
the characteristics of U ,v and w with the application of u0. u0 can be computed as
q = f5 + g51δp0 = 0
r = f6 + g62δy0 = 0
p = f4 + g43δr0 = 0
or
δp0 = − f5
g51
δy0 = − f6
g62
δr0 = − f4
g43
Simulations have been performed in various cases of flight conditions to show zero
dynamics stability for all cases. Figure 2.2 is one of the test results for a particular flight
condition. It can be seen from Figure 2.2 that all the three states of the internal dynamics
are bounded. Same scenario can be seen for any other flight conditions also. Thus it can
be concluded that the overall system can be stabilized with a stable linear controller as
the internal dynamics are asymptotically stable.
2.2.1.2 Simulation results
A program is used to simulate the full scale 6-DOF model of the missile dynamics. A con-
ventional three loop configuration lateral autopilot (pitch and yaw channels are identical)
and a roll autopilot in PI configuration are also considered for performance comparison.
The linear controller gains are scheduled as a function of velocity throughout the flight
trajectory. The 6-DOF simulation platform includes a seeker based homing guidance
(nonlinear Proportional Navigation law) loop to present the performances of the nonlin-
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback48
Figure 2.2: Boundedness of u,v and w in zero dynamic condition
ear controller under closed loop guidance. The objective of this subsection is to bring out
the salient features of the nonlinear controller through 6-DOF simulations. The plant is
considered to be time varying. A ”boost-sustain-coast” thrust profile is considered. All
the aerodynamic parameters are used in the form of look up tables. Nonlinearities of
actuation system such as dead zone, command saturation and rate saturation are con-
sidered. All the numerical results given in this thesis have been normalized based on the
results obtained from conventional three loop configuration as the data and tables used
for simulations need to be kept confidential.
As stated in the Chapter 1, when the PN switch over has taken place, the missile tries
to maneuver at its maximum capability (depending mainly on the structure and strength
of the missile). During this PN maneuver the latax demand reaches its maximum value
and when the missile tries to achieve this demand the coupling among pitch-yaw-roll
axes, large variations in yaw deflection, side slip angle, gimbal angle and aerodynamic
coefficients can be observed. In view of these phenomenon, to compare the performances
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback49
between FBLC and conventional three loop autopilot, simulation results have been orga-
nized in the following way:
1. Decoupling of the FBLC.
2. Input-output linearization in three channels.
3. Decreased yaw deflection during PN switch over.
4. Reduced side slip, gimbal angle and CS and Cny during PN switch over.
In the normalized time axis of the simulation results, the PN switch over can be observed
during t = 0.75 to t = 0.9.In comparative simulation results we have compared the
autopilot performances between FBLC (new nonlinear autopilot) and the existing one
(three loop autopilot).
2.2.1.3 Decoupling of three axes motion
Figure 2.3 illustrates the decoupling phenomenon by applying step body rate demands in
the three channels. Step commands have been given in three different instants such as
t = 0.25, t = 0.5 and t = 0.75 respectively for pitch,yaw and roll channels. Now from the
figure it can be seen that step command in one channel does not affect the other channels.
Figure 2.4 shows the comparative autopilot performance with rate demand as per
closed loop guidance requirements. The comparative results of classical three loop au-
topilot and the feedback linearizing control (FBLC) show decoupling property of the
latter. During pitch maneuver at t = 0.46 and t = 0.67, FBLC shows almost no effect
on roll rate as well as on yaw rate. But for the conventional one, during pitch maneuver
simultaneous effects on roll rate and yaw rate can be seen.
2.2.1.4 Input-Output Linearization
Figures 2.5, 2.6 and 2.7 describe the input-output linearization in the pitch-yaw-roll chan-
nel respectively applying the guidance commands. Referring to equation 2.10, Figure 2.5,
2.6, 2.7 shows the plots of q and v1, r and v2, p and v3 respectively. It can be seen that for
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback50
Figure 2.3: Autopilot response with step command
each channel, the plotted parameters are almost superimposed to each other, revealing
that very good linearization is achieved in all the three channels. Figure 2.8 shows the
comparative performance of the autopilot and illustrates how the latax demands in pitch
and yaw planes are satisfied. Almost similar latax demand and sensed profiles can be
seen for the FBLC and the classical autopilots. For classical one the yaw latax has been
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback51
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−1
−0.5
0
0.5
Normalised Time
Nor
mal
ised
Pitc
h R
ate
Comparative Performance in 6−DOF Simulation
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−3
−2
−1
0
1
Normalised Time
Nor
mal
ised
Yaw
Rat
e
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
0
1
Normalised Time
Nor
mal
ised
Rol
l Rat
e
FBLCExisting
Figure 2.4: Decoupling in the three channels
increased due to the effect of coupling during PN switch over. As a result yaw deflection
is seen to increase in Figure 2.9. Figure 2.9 illustrates the input requirement in terms
of fin deflection in the three channels. It can be seen from the comparative simulation
results that during PN maneuver, input demand in yaw channel is greater for existing
classical controller over the FBLC. This happens due to the induced yaw rate caused by
the increased roll rate during PN switch over for the classical controller. The fin require-
ment is also shown in Figure 2.10, where we can see that the control requirements in the
three planes have been distributed to the four fins.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback52
Figure 2.5: Evidence of linearization in pitch channel
The rate of fin deflections are also shown in Figure 2.11 which indicate that the rate
is below the maximum rate limit (normalized to 1).
Figures 2.12 and 2.13 describe comparative results for gimbal angle, α and β. From
these figures it can be seen that the gimbal angle in the azimuth plane and β rises higher
value for classical controller over the FBLC. This increase is prominent during PN switch
over where coupling effect is maximum for the existing controller.
Figure 2.14 shows the variations in the aerodynamic coefficients CS, Cny and CL
during flight for the classical controller. During PN switch over, due to the effect of
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback53
Figure 2.6: Evidence of linearization in yaw channel
increased β, the values of CS and Cny increase.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback54
Figure 2.7: Evidence of linearization in roll channel
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback55
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−0.5
0
0.5
Normalised Time
Nor
mal
ised
Pitc
h la
tax
Comparative Performance in 6−DOF Simulation
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
0
1
2
Normalised Time
Nor
mal
ised
Yaw
lata
x
Comparative Performance in 6−DOF Simulation
FBLCExisting
Figure 2.8: Pitch and yaw latax profiles
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback56
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−3
−2
−1
0
1
Normalised Time
Eff.
pitc
h de
fln (
δ PB
)
Comparative Performance in 6−DOF Simulation
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−4
−2
0
2
4
Normalised Time
Eff.
yaw
def
ln (
δ YB
)
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−3
−2
−1
0
1
Normalised Time
Rol
l def
ln (
δ R )
FBLCExisting
Figure 2.9: Effective pitch yaw roll deflection
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback57
5 10 15 20
−2
−1
0
1
Normalised Time
Nor
mal
ised
Fin
−1 d
efl
n
FBLCExisting
0.2 0.4 0.6 0.8 1
−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
Nor
mal
ised
Fin
−2 d
efl
n
FBLCExisting
0.2 0.4 0.6 0.8 1−3
−2
−1
0
1
Normalised Time
Nor
mal
ised
Fin
−3 d
efl
n
FBLCExisting
0.2 0.4 0.6 0.8 1−2
−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
Nor
mal
ised
Fin
−4 d
efl
n FBLCExisting
Figure 2.10: Fin deflections
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback58
0.2 0.4 0.6 0.8 1−1
−0.5
0
0.5
Time (Normalized)
Fin
−1 d
efln
rat
e in
deg
/sec
FBLCExisting
0.2 0.4 0.6 0.8 1
−0.5
0
0.5
Time (Normalized)
Fin
−2 d
efln
rat
e in
deg
/sec
FBLCExisting
0.2 0.4 0.6 0.8 1−1
−0.5
0
0.5
Time (Normalized)
Fin
−3 d
efln
rat
e in
deg
/sec
FBLCExisting
0.2 0.4 0.6 0.8 1−1
−0.5
0
0.5
Time (Normalized)
Fin
−4 d
efln
rat
e in
deg
/sec
FBLCExisting
Figure 2.11: Fin deflection rate
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback59
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
Normalised Time
Nor
mal
ised
Gim
bal A
ngle
(El)
Comparative Performance in 6−DOF Simulation
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−4
−2
0
2
4
Normalised Time
Nor
mal
ised
Gim
bal A
ngle
(Az) FBLC
Existing
Figure 2.12: Gimbal angle
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback60
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−2
−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
Nor
mal
ised
Alp
ha (
Bod
y)
Comparative Performance in 6−DOF Simulation
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−2
−1
0
1
Normalised Time
Nor
mal
ised
Bet
a (B
ody)
Comparative Performance in 6−DOF Simulation
FBLCExisting
Figure 2.13: Alpha and beta profiles
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback61
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−0.3
−0.2
−0.1
0
0.1
A Comparative Performances in 6−DOF Simulation
Rol
ling
mom
ent c
oeffi
cien
t CL
Normalised Time
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−0.5
0
0.5
1
Sid
e fo
rce
coef
ficie
nt C
S
Normalised Time
FBLCExisting
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−8
−6
−4
−2
0
2
Yaw
ing
mom
ent c
oeffi
cien
t Cny
Normalised Time
FBLCExisting
Figure 2.14: Force and moment coefficients
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback62
2.2.2 Case II: Input-output linearization with w,v and p as out-
puts
In Section 2.2.1, feedback linearization of the inner rate loop has been discussed consid-
ering the output of the system as q, r and p. But in that case, a part of the plant has
been linearized that does not include the rate to latax dynamics.In other words only the
inner rate loop of the plant was linearized. The outer loop has been designed based on
the conventional approach. Now in this section IO linearization is applied considering the
inner and the outer loops as described in the block diagram given in Figure 2.15. As a
result the plant dynamics from wd, vd, pd to w, v, p can be made linearized and decoupled.
Another advantage is that there is no extra effort required for designing the outer loop
controller. To obtain these conditions, we have chosen the outputs as w,v and p. One
important property of the feedback linearization is that, for different choice of outputs
the relative degree of the same system would become different and at times the system
may not be linearizable for some particular sets of outputs. That is why the choice of
proper outputs is important in the case of feedback linearization. From Figure 2.15, it
can be seen that the new input v can be computed from reference inputs wd, vd and pd as
per guidance (if possible) and v to w, v and p have been linearized. Note that outer loop
controller like Figure 2.1 is absent here and this is the main advantage of linearizing the
whole plant rather than partial linearizing. The aim of this thesis is to track the missile
lateral acceleration demand in both the pitch and yaw planes as well as the roll rate in the
roll plane, using the missile fins; hence yielding a system with 3 inputs and 3 controlled
outputs.
2.2.2.1 Formulation of the problem for the STT missile
The state-space form of the non-linear system of the homing missile can now be written
in a compact parametric format, as: 1.3 and 1.4 or in matrix form as described in Section
2.2.1. State vectors and inputs are the same as considered in Case I. The only difference
is in output vector and here it is taken as y = [w, v, p]T . Now the whole system can be
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback63
Figure 2.15: Block diagram of the system with output w, v and p
represented as,
x = f(x) + g(x)u
where f(x) and g(x) are the same as in Section 2.2.1. The FBLC is similar to that in
Case I, however, since the outputs are the derivative of the states, there is no need for
further differentiation since the input-output model is the same as the state space model.
Only the output p needs to be differentiated for linearization of the roll channel [51].Thus
we have,
y1
y2
y3
= M(x) + E(x)
δp
δq
δr
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback64
where,
M(x) =
h1(x)
h2(x)
Lfh3(x)
i.e, M(x) =
f3
f2
f4
and, E(x) =
g31 0 0
0 g22 0
0 0 g43
From M(x) and E(x) matrices it can be observed that by choosing the outputs as w,v
and p, inputs are already decoupled from each other as in the previous Section 2.2.1. it
can be verified similar to the case is decoupling matrix E(x) is nonsingular for the whole
flight time so we can now derive the control law similar to Equation as given below:
u = −E−1M + E−1
v1
v2
v3
Note that the above input-output relation is, again, linear and decoupled. For this par-
ticular case, the tracking error is given as,
e =
e1
e2
e3
=
y1 − y1d
y2 − y2d
y3 − y3d
=
w − wd
v − vd
p− pd
(2.14)
Now, choosing the new input v1, v2 and v3 as
v1 = y1d
v2 = y2d
v3 = y3d + k3(y3d − y3)
(2.15)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback65
with k1, k2 and k3 being positive constants, the tracking error dynamics of the closed loop
system is reduced to,
e1 = 0
e2 = 0
e3 + k3e3 = 0
Simulations are given below to show the system performance along with this FBLC con-
troller.
2.2.2.2 Simulation results
Figure 2.16 illustrates the autopilot performance for step commands. Refereing to Figure
2.16, the step commands have been given at three different time instants in the three
channels, namely, t = 0.25, t = 0.5 and t = 0.75 in pitch, yaw and roll channels re-
spectively. No significant disturbance in the other channels are seen when each of these
channels are excited by a step command.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback66
Figure 2.16: Autopilot response with step command
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback67
Figure 2.17 describes the linearization in the pitch-yaw-roll channels, with seeker based
guidance commands. One can see from this figure that the plots v1 and q are matching
very well with each other which demonstrates linearization in pitch plane dynamics. The
same phenomenon can be seen in yaw and roll channels also. Figure 2.18 illustrates the
Figure 2.17: Autopilot response with guidance command
effective pitch, yaw and roll control fin deflections.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback68
Figure 2.18: Effective Pitch yaw roll deflection
.
2.3 Nonlinear Luenberger observer design
The control inputs computed by the FBLC are functions of state variables. Since all
the state variables of the missile can not be measured through sensors, an observer is
essential to estimate the unmeasured states of the system for control computation. This
chapter presents an explicit form of a nonlinear observer for a class of multi-input multi-
output systems. Observer construction [46] for multi-input nonlinear systems is not a
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback69
trivial extension of single-output case, especially when the global error convergence is of
interest. In this section, a nonlinear reduced order Luenberger observer for a STT short
range missile system is presented.
2.3.1 Introducion
The observer has been designed with guaranteed stability of the estimation error dynamics
for the nominal plant. For a missile, measured state variables are q, r and p, while the
unmeasured state variables which are to be estimated are U , v and w. The forward velocity
assumed to be known from predetermined thrust profile of the rocket motor. Thus, only
v and w need to be estimated. The input to the observer are the measurements of lateral
accelerations fz and fy from the accelerometers of the missile. In this work, instead of
estimating w and v, the angle of attack α and sideslip angle β have been estimated as fz
and fy are the sole functions of α and β and it will be more convenient to express those
measurement variables in terms of α and β rather than U, v, w. As a result mathematical
computations such as derivative, integration will become more easier for fz and fy if they
are expressed in terms of α and β. For example fz and fy are expressed as
fz = QSm
[CNAn + CNAs − Cleta + D2Vm
(−Czqq − Czαα)]
fy = QSm
[CNBn + CNBs + Clzie + D2Vm
(Cyrr − Czββ)](2.16)
Now as we know that aerodynamic coefficients such as CNAn, CNAs etc are directly related
to α and β where α and β are related to U , v and w as
α = tan−1(w
U
)
β = tan−1( v
U
) (2.17)
Thus one can imagine how complicated it will be if fz and fy are expressed in terms of
U, v and w.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback70
2.3.2 Observer construction
We now proceed to introduce the observer. The observer has the following properties:
• The construction of the observer, based on [6], doest not require a preliminary non-
linear change of coordinates. Thus help us to avoid differentiations and integrations
with fz and fy.
• The calculation of gain is straightforward as the observer gain is constant through
out the whole flight time and can be chosen by trial and error.
• It is computationally heavy but easily implementable.
Although the theory is applicable up to forced/autonomus multiple-input/multiple-output
nonlinear system, here we shall concentrate on a general MIMO systems of the form:
·x = f (x, u)
y = h (x, u), (2.18)
which is similar to a typical nonlinear missile model described in Section 2.2 with state
x(t) ∈ Rn, h(t) ∈ Rm . In what follows, we also denote by Q(x) the so-called observability
matrix of equation 2.18 [38] as,
Q (x) = DΦ (x) =dΦ (x)
dx(2.19)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback71
where,
Φ(x) =
h1(x)
Lfh1(x)...
Lλ1−1f h1(x)
h2(x)...
Lλ2−1f h2(x)
...
Lλm−1f hm(x)
(2.20)
where the integer numbers λ1, ···, λm are such that∑m
i=1 λi = n and that Φ(x) : X → Φ(x)
is a diffeomorphism. in a linear case, i.e., h(x) = Cx and f(x) = Ax, the matrix (2.19)
reduces to the well known observability matrix. If Q(x) has the full rank for all x ∈ Rn
then the error dynamics of the proposed observer equations
˙x(t) = f(x(t)) + [Q(x(t))]−1K[y(t)− h(x(t))]
x(0) = x ∈ Rn
(2.21)
will be stable as reported in [9] and [10]. The gain vector K, can be chosen in the following
manner:x(t)− ˙x(t) = e(t) = f(x, u)− f(x, u)−G(x, u))(y − y)
where
G(x, u) = Q−1K
or
e = (A−Q−1KC)e
where
A = ∂f∂x
∣∣x=x
andC = ∂h∂x
∣∣x=x
, (2.22)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback72
For MIMO systems
A(x(t), u) =
∂f1
∂x1
∂f1
∂x2......... ∂f1
∂xn...
... ............
∂fr
∂x1
∂fr
∂x2......... ∂fr
∂xn...
... ............
∂fn
∂x1
∂fn
∂x2......... ∂fn
∂xn
∣∣∣∣∣∣∣∣∣∣∣∣∣x=x
and
C(x(t), u) =
∂h1
∂x1
∂h1
∂x2......... ∂h1
∂xn...
... ............
∂hr
∂x1
∂hr
∂x2......... ∂hr
∂xn...
... ............
∂hm
∂x1
∂hm
∂x2......... ∂hm
∂xn
∣∣∣∣∣∣∣∣∣∣∣∣∣x=x
where
x =
x1
x2
...
xr
...
xn
, y(x(t), u) =
h1(x(t), u)
h2(x(t), u)...
hr(x(t), u)...
hm(x(t), u)
and u =
δP
δY
δR
From equation 2.22 it is clear that the value of K can easily be found out using pole
placement technique.
2.3.3 Application to a realistic STT missile model
As we already discussed that the input of the observer or h(x) functions are taken as
lateral accelerations of the missile and it is very complicated task to represent those
parameters as a function of known and unknown states. So to overcome this difficulty
we have formulated the problem with new state variables, namely, angle of attack α and
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback73
sideslip angle β. Now the parametric model of the missile for the nonlinear observer can
be written as
α =wU − Uw
U2(1 + tan2(α))= fα(α, β, δp)
β =vU − Uv
U2(1 + tan2(β))= fβ(α, β, δy)
(2.23)
and U, v, w, U , v, w are taken from the system equations 1.3. The output equations can
be written as
y1 = fz = QSm
[CNAn + CNAs − Cleta + D2Vm
(−Czqq − Czαα)]
y2 = fy = QSm
[CNBn + CNBs + Clzie + D2Vm
(Cyrr − Czββ)]
y3 = QSDIXX
CL
y4 = QSDIY Y
(CMAn + CMAs) + QSD2
2VmIY YCmαα
y5 = QSDIZZ
(CMBn + CMBs) + QSD2
2VmIZZCmββ
(2.24)
As the order of the system has been reduced to 2, the observability matrix can be con-
structed with two of the outputs as
Q(x) =dΦ(x)
dx=
(∂y1
∂α∂y2
∂α∂y1
∂β∂y2
∂β
)(2.25)
It can be shown that the Q(x) matrix will be nonsingular for the whole flight time. So
we can now proceed towards the next step of the observer formulation.
2.3.3.1 Gain ’K’ calculation
Our observer equation is given as,
(˙α˙β
)=
(fα(α(t), β(t), δp)
fβ(α(t), β(t), δy)
)+ Q−1(x(t), δp, δy)K
(y1(x(t))− y1(x(t))
y2(x(t))− y2(x(t))
)(2.26)
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback74
and the error equation becomes
e =
(e1
e2
)= (A(x(t))−Q−1(x(t))KC)
(e1
e2
)(2.27)
where,
A(x(t)) =
(∂f1
∂α∂f1
∂β∂f2
∂α∂f2
∂β
)∣∣∣∣∣α=α,β=β
and C(x(t)) =
(∂h1
∂α∂h1
∂β∂h2
∂α∂h2
∂β
)∣∣∣∣∣α=α,β=β
The value of K will be such that the eigenvalues of the system 2.27 will lie in the
negative half of s plane. In this case K matrix is of the form
K =
(k11 k12
k21 k22
)
Now scalar elements k11 = −16.0, k12 = 1.0, k21 = −60.0, k22 = 0.0 can be found trial and
error basis in view of the desired eigen values of the error dynamics lie in left half of s
plane and for this case they are placed at −10.0 and −6.0 respectively. The simulation
results in the next section show the tracking performance of the proposed observer.
2.3.4 Simulation results
Figures 2.19 and 2.20 show the observer performance for a limited time period. For both α
and β, the true initial condition is zero but in the simulation we have taken some nonzero
initial conditions of about 2− 3 . The observer performance has been highlighted in the
high maneuvering zone of the missile i.e., PN switch over period. As we know that inputs
to the observer fz and fy are functions of uncertain aerodynamic coefficients which varies
with its magnitude as well as sign abnormally during PN switch over, there is maximum
possibility of divergence in error dynamics of the observer in this period. Note that
observer has been constructed based on a nominal missile model and measurement noise
has been considered during simulation. Figure 2.19 shows good tracking performance
in high maneuver zone. Figure 2.20 also depicts the same. Note that, the gain (K) of
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback75
the observer is not adaptive one and remains fixed for whole flight time. So it may be
necessary to manipulate the gain when ever the flight condition has been changed or if
there is huge change in initial condition.
0.5 0.55 0.6 0.65 0.7 0.75
−1
−0.5
0
0.5
1
Normalised Time
No
rmal
ised
alp
& a
lph in
Bo
dy
Fra
me
alp and alphat
alph
alp
Figure 2.19: Estimated and true α
Figure 2.21 and 2.22 shows the true and estimated α and β for the full envelope. One
can see from these figures that α and β tracks α and β satisfactorily.
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback76
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1−0.1
−0.05
0
0.05
0.1
0.15
Normalised Time
No
rmal
ised
bet
a h &
bet
a in
Bo
dy
Fra
me
Normalised beta and betahat
betah
beta
Figure 2.20: Estimated and true β
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback77
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
No
rmal
ised
alp
& a
lph in
Bo
dy
Fra
me
alp and alphat
alph
alp
Figure 2.21: Estimated and true α for full flight time
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback78
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−0.1
−0.05
0
0.05
0.1
0.15
0.2
0.25
Normalised Time
No
rmal
ised
bet
a h &
bet
a in
Bo
dy
Fra
me
Normalised beta and betahat
betah
beta
Figure 2.22: Estimated and true β for full flight time
Chapter 2. Autopilot Design with Input-Output Linearization by Feedback79
2.4 Comments
The above simulation results show the linearization and decoupling in pitch yaw and roll
channels. It may be also observed that the performance of the nonlinear controller is
clearly better than the existing linear one in terms of roll rate, angle of attack and impact
angle. The main achievement of this work is to achieve the decoupling among yaw roll
channels which improves the gimble angle limitation and fin saturation. A nonlinear Lu-
enberger observer has been presented along with the estimation of α and β. It can be seen
from the plots that the proposed observer can estimate the α and β quite accurately.This
is the foundation on which subsequent chapters will be built.
Chapter 3
H∞ Control of Feedback Linearized
Inner Rate Loop Dynamics
3.1 Introducion
Although input-output linearization technique may be effective to meet the desired re-
quirements of the flight control system of a missile, it fails to handle uncertainties in the
missile model which is a natural factor during real flight conditions. Input-output lineariz-
ing controller computes plant inputs based on nominal plant and that is why it fails to
cancel the nonlinearity on a largely perturbed plant. So, in order to handle the perturbed
plant, in this chapter a robust control law has been incorporated with linearizing control
law so that the perturbed plant model can still be linearized and decoupled. Here an
H∞ controller has been synthesized based on Linear Matrix Inequality (LMI) approach
to tackle aerodynamic uncertainties along with the various disturbances. So in practical
sense this robust controller will be effective for most of uncertainties and disturbances
which are not trivial specially during high angle of attack. This chapter presents some
mathematical evidences of the fact that for a high range of uncertainty, a robust control
structure exists with which the system can be stabilized.
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics81
3.2 Problem formulation
Figure 3.1 represents the robust feedback linearization approach based on H∞ control.
One can observe that the block diagram represented in 3.1 is almost similar to Figure
2.1. The only difference is that here we are using a robust control block ’RC’ instead of
the linear control block ’LC’. That means the computation of the new inputs v has been
modified for handling the model uncertainties and other disturbances. The inclusion of
the block ’RC’ makes the FBLC control law u robust for the inner rate loop.
Figure 3.1: Block diagram of the system representing robust feedback linearization foroutputs q, r and p
Note that feedback linearization approach presented in Sections 2.2.1, 2.2.2 is exact
only for the nominal model. To retain good performance even with the plant perturbation
due to aerodynamic uncertainties, a robust controller needs to be designed around the
nominally feedback linearized plant. In this section, the design of an H∞ controller
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics82
synthesized using the Linear Matrix Inequality (LMI) approach for the linearized plant
after feedback linearization is presented. Let input-output(IO) dynamics for the nominal
plant be written as
y(ρ)n = Mn(x) + En(x)u (3.1)
where ρ is the relative degree of the system for the nth output. Now, the perturbed plant
(3.1) can be written as,
y(ρ)p = Mn(x) + ∆M(x) + (En(x) + ∆E(x))u
y(ρ)n + ∆yn = [Mn(x) + En(x)u] + [∆M(x) + ∆E(x)u]
(3.2)
where yn and yp are the nominal and perturbed plant outputs, and ∆M(x) and ∆E(x)
are the perturbations in M(x) and E(x), respectively. These perturbations occur due to
the variations in aerodynamic coefficients. For example, variations around the nominal
roll moment coefficient CL0 can be expressed as CL = CL0 + ∆CL. Similarly all other
aerodynamic coefficients can be expressed with their variations around the nominal value.
∆M(x) and ∆E(x) are then given as functions of these variations in the aerodynamic
coefficients. So the total variation in the plant is given by d = ∆M(x) + ∆E(x) and
is modeled as an exogenous input to the nominal plant. Here the term robust, means
that even in the presence of aerodynamic uncertainties and various disturbances like wind
gust, fin misalignment, wing misalignment, shifting in C.G and C.P etc. the input-output
linearization can be performed and the plant dynamics can be made decoupled. It can be
shown that it is possible to formulate a robust control law over the feedback linearized
plant to achieve the desired robustness. For example let us consider only the pitch channel.
After input-output linearization and decoupling, the new input v1 and pitch rate q will
be related theoretically as 1s. Referring the Figure 3.2, G = 1
sand let K be the robust
control law. d is modeled as exogenous input which indicates the effective uncertainties
and disturbances in the pitch plane model. In this case, for nominal plant the feedback
linearized input-output relation in each of the three channels becomes G = 1s
and for
the perturbed plant let us consider that the input-output relation becomes G. Now the
notation d indicates G−G i.e., effective deviation of feedback linearized IO relation from
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics83
Figure 3.2: Structure of robust control K in pitch plane
the nominal IO one. Here we have formulated the problem for pitch channel only. v1 is
the new input for the pitch plane generated by some robust control law. Similar control
structures can be formulated for yaw and roll plane model to compute v2 and v3. Now,
our aim is to design the robust control law v = [v1, v2, v3]T in such a way that the overall
control input to the nonlinear plant u = −E−1M + E−1
v1
v2
v3
as depicted in (2.9) will
be a robust one. In the following sections we will only discuss about computation of v1
in the pitch plane as similar computations are applicable to yaw and roll channels.
3.2.1 A linear matrix inequality approach to H∞ control for de-
signing K in pitch plane
In this section we will consider continuous H∞ control problem which is solved via ele-
mentary manipulations on linear matrix inequalities (LMI). The whole design procedure
can be divided into two major parts. The first is that of checking of solvability condi-
tions, and the second one involves an LMI based parametrization of the H∞-suboptimal
controller. The solvability condition involves Riccati inequalities rather than the usual in-
definite Riccati equations. Alternatively, these equations can be expressed as a system of
three LMI’s. Efficient convex optimization techniques are available in MATLAB to solve
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics84
this system of LMI’s. Its solutions parameterize the H∞ controller. The robust control
structure has finally been developed from the Kalman-Yakubovich-Popov Lemma(KYP)
lemma, otherwise known as the Bounded Real Lemma[41]. The detailed derivations and
discussions are presented in Appendix B. For the sake of continuity in this section, we
will discuss and mention only the relevant points of the theory. The synthesis problem
given in this work is concerned with the familiar feedback arrangement shown in Figure
3.3. As depicted here, the so-called closed-loop system has one external input and one
output, given by d and z respectively. We assume that this connection is well-posed. It
is convenient to introduce the following notation for the transfer function d to z in the
diagram.
Figure 3.3: General feedback arrangement
The uncertainties and various disturbances can be seen in the vehicles with aerody-
namic controls due to the change of aerodynamic coefficients, drag, thrust mis-alignments,
CG (center of gravity) shift etc. The primary function of the robust controller designed
for the inner rate loop is to minimize the impact of these disturbances on the overall
system. Figure 3.4 displays the block diagram robust control structure in the standard
mixed sensitivity configuration where G(s) is the open-loop plant which is 1s
here, K(s)
is the controller to be designed, and W1(s) and W3(s) are weights for shaping the char-
acteristics of the open-loop plant. The design objective is to minimize a weighted mix of
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics85
Figure 3.4: Mixed sensitivity configuration
the transfer function
S(G,K) := G11 + G12K(I − G22K)−1G21
where G =
[G11 G12
G21 G22
]and thus handling the robustness issues as well as the stability
and performance of the system. This is sometimes called the star-product between transfer
functions G and K, or equivalently their linear fractional transformation. This mixed
sensitivity design objective is represented as
∥∥∥∥∥
[W1(s)S(s)
W3(s)T (s)
]∥∥∥∥∥∞
< 1
where, S(s) = (I−G(s)K(s))−1 and T (s) = G(s)K(s)(I−G(s)K(s))−1 are sensitivity and
complementary sensitivity functions. The state space description of the augmented-plant
is given by
xp
z
y1
=
A B1 B2
C1 D11 D12
C2 D21 0
xp
d
v1
(3.3)
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics86
where xp is the state variable vector of the plant G(s) and weights (W1(s),W2(s)) com-
bined, d is the disturbance input (contains the effect of plant uncertainties and various
disturbances), v1 is the plant input, y1 is the measured signals including disturbances and
z is the regulated output. The state space representation of the controller is given by
xk = Akxk + Bkeq
v1 = Ckxk + Dkeq
where xk represents the controller states; v1 is the controller output. The transfer matrix
between d and z is given by
Gzw(s) =
[W1(s)S(s)
W3(s)T (s)
]= CL(sI − AL)−1BL + DL
where,
AL =
[A + B2DkC2 B2Ck
BkC2 Ak
]
BL =
[B1 + B2DkD21
BkD21
]
CL =[
C1 + D12DkC2 D12Ck
]
DL = D11 + D12DkD21
(3.4)
We can parameterize the closed-loop relation in terms of the controller realization as
follows. First we make the following definitions.
A =
[A 0
0 0
], B =
[B1
0
], C =
[C1 0
], C =
[0 I
C2 0
]
B =
[0 B2
I 0
], D12 =
[0 D12
], D21 =
[0
D21
] (3.5)
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics87
which are entirely in terms of the state space matrices for G. Then we have
AL = A + BJpC, BL = B + BJpD21
CL = C + D12JpC, DL = D11 + D12JpD21
(3.6)
where
Jp =
[Ak Bk
Ck Dk
]
The crucial point here is that the parametrization of the closed loop state space matrices
is affine in the controller matrix Jp.
Now we formulate the H∞ control problem based on LMI approach in the following steps.
Any further details along with the proofs have been given in Appendix B.
• Suppose ML(s) = CL(Is − AL)−1BL + DL. Then the following are equivalent con-
ditions.
1. The matrix AL is Hurwitz and∥∥∥ML
∥∥∥∞
< γ
2. There exists a symmetric positive definite matrix XL such that
A∗LXL + XLAL XLBL C∗
L
B∗LXL −γI D∗
L
CL DL −γI
< 0
• The above inequality is equivalent to
HXL+ Q∗J∗pPXL
+ P ∗XL
JQ < 0 (3.7)
where,
HXL=
A∗XL + XLA XLB C∗
B∗XL −γI D∗11
C D11 −γI
,
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics88
PXL=
[B∗XL 0 D∗
12
]
and Q =[
C D21 0]
• Considering the above two cases there exists a controller synthesis K if and only if
there exists a symmetric matrix XL > 0 such that
W ∗PXL
HXLWPXL
< 0 and W ∗QHXL
WQ < 0 (3.8)
where ImW ∗PXL
= KerPXLand ImW ∗
Q = KerQ
However the above inequality is not an LMI condition as the variable XL appears
in both HXLand PXL
. So now we will convert it to an LMI condition.
• Given XL > 0;
W ∗PXL
HXLWPXL
< 0, if and only if, W ∗P TXL
WP < 0
where,
TXL=
AX−1L + X−1
L A∗ B X−1L C∗
B∗ −γI D∗11
CX−1L D11 −γI
(3.9)
and
P =[
B∗ 0 D∗12
]
Recall that XL is a real and symmetric (n + nK)× (n + nK) matrix; here n and nk
are state dimensions of G and K. Let us now define the matrices X and Y which
are submatrices of XL and X−1L , by
XL =:
[X X2
X∗2 X3
]and X−1
L =:
[Y Y2
Y ∗2 Y3
](3.10)
• Suppose X and Y are symmetric, positive definite matrices in Rn×n; and nk is a
positive integer. Then there exist matrices X2, Y2 ∈ Rn×nK and symmetric matrices
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics89
X3, Y3 ∈ RnK×nK , satisfying
[X X2
X∗2 X3
]> 0 and
[X X2
X∗2 X3
]−1
=
[Y Y2
Y ∗2 Y3
](3.11)
if and only if [X I
I Y
]≥ 0 and rank
[X I
I Y
]≤ n + nK (3.12)
• Suppose XL is a positive definite (n+nK)× (n+nK) matrix and X and Y are n×n
matrices satisfy 3.12. Then
W ∗P TXL
WP < 0 and W ∗QHXL
WQ < 0
If and only if, the following two matrix inequalities are satisfied
1.[
NX 0
0 I
]∗
A∗X + XA XB1 C∗1
B∗1X −I D∗
11
C1 D11 −I
[NX 0
0 I
]< 0 (3.13)
2.[
NY 0
0 I
]∗
AY + Y A∗ Y C1 B1
C1Y −I D11
B∗1 D∗
11 −I
[NY 0
0 I
]< 0 (3.14)
Where Nx and Ny are full-rank matrices whose images satisfy
ImNX = ker[
C2 D21
]
ImNY = ker[
B∗2 D∗
12
] (3.15)
• Comments: The above steps provide us with an explicit way to determine
whether a synthesis exists which solves the H∞ problem. Another point to be
mentioned is that the above synthesis exists if and only if nk ≥ n as well as rank(X−
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics90
Y −1) ≤ nk. But if nk ≤ n then the γ-suboptimal controller of order nk (reduced
order) exist if and only if (3.13)-(3.15) hold for some X and Y which further satisfy:
Rank(I −RS) ≤ nk
3.2.2 Design Steps
1. To compute X and Y from the system of LMI’s (3.13)-(3.15) and that
Rank(I −RS) ≤ nk
2. To construct an H∞ controller from this data, we can recall that X,Y are related by
(3.11) to solution XL of the Bounded Real Lemma inequality. We therefore begin by
computing a positive definite matrix XL ∈ R(n×nk)×(n×nk) compatible with (3.11).
To this end we have to compute two full-column-rank matrices X2, Y2 ∈ R(n×nk)
such that
X2YT2 = I −XY (3.16)
XL is then obtained as the unique solution of the linear equation:
[Y I
Y T2 0
]= XL
[I X
0 XT2
](3.17)
Note that (3.17) is always solvable when Y ≥ 0 and X2 has full column rank.
3. It can be proved that (X,Y) can solve (3.13) to (3.15) if and only if XL is given by
(3.11) is positive definite and satisfies (3.8).
4. This guarantees the existence of a solution Jp =
[Ak Bk
Ck Dk
]to the Bounded Real
Lemma inequality (3.7). And from Bounded Real Lemma K(s) = Dk + Ck(sI −Ak)
−1Bk is then a γ−suboptimal controller.
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics91
5. In summary, from the given X and Y first we will compute XL ≥ 0 by step 1 and
step 2. Then we will write the Bounded Real Lemma inequality for this XL as,
A∗LXL + XLAL XLBL C∗
L
B∗LXL −γI D∗
L
CL DL −γI
= HXL
+ Q∗J∗pPXL+ P ∗
XLJQ < 0 (3.18)
and solve this inequality (3.18) for the controller parameters Jp =
[Ak Bk
Ck Dk
]
Similarly, robust control laws can also be found out for yaw (Jy) and roll (Jr) channel.
3.2.3 Application to the Nonlinear Missile
After obtaining the desired robust control law v1, we can summarize the whole process in
four steps as given below:
• Feedback linearize the proposed nonlinear system
• Get the Linear Fractional Transformation (LFT) of the linear model obtained through
feedback linearization for the each channel.
• Get the disturbance model (d)
• Find the robust stabilizing controller v1 using LMI approach as discussed in the
Section B.4.3
So, from the list it is clear that we need to obtain disturbance model d which is given in
the next section.
3.2.3.1 Uncertainty and disturbance modeling between y and v
Let, P be the nominal feedback linearized plant transfer function, from (2.10) which is a
diagonal matrix with 1s
as the diagonal elements. Let, due to the variation in aerodynamic
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics92
coefficients, the nominal model P be perturbed to P which is also assumed diagonal. P is
estimated through system identification methods available in the MATLAB identification
toolbox using realistic simulation data obtained from FORTRAN simulation. Details of
the simulation data as well as the procedure of obtaining P have been given in the thesis.
Let ∆P denote the unmodelled dynamics for the nominal plant P .
∆P = P − P (3.19)
A brief discussion on constructing the ARMAX model has been given next.
Small theoretical background for constructing ARMAX model
It is well known that a linear, time invariant, causal system can be described by its impulse
response g(τ).
y(t) =
t∫
0
g(τ)u(t− τ)dτ
Though most processes are of continuous type, it is advantageous to choose a discrete time
model structure, compatible for parameter estimation simulation using digital computers.
Assuming a sampling interval of one time unit:
y(t) =∑
k=1
g(k)u(t− k)t = 0, 1, 2......
Now introducing the backward shift operator q−1 , we can write:
q−1u(t) = u(t− 1)
y(t) = G(q)u(t)
where, G(q) =∑
g(k)q−k known as the transfer operator of the linear system. From
Figure 3.5 the output error signal e(t) can be defined as e(t) = y(t) − ym(t) and it is
assumed that e(t) and em(t) are the gaussian and color noise respectively. H(q) is the
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics93
Figure 3.5: Complete model structure of the system and noisy signal
filter transfer function. To be able to estimate the functions G and H they have to be
parameterized as rational functions in the delay operator q−1. Assuming a basic input-
output configuration as depicted in Figure 3.5, we can write:
y(t) = G(q)u(t) + H(q)e(t)
A much used parametric model is the extended auto-regressive (ARX) that corresponds
to:G(q) = q−nk B(q)
A(q), H(q) = 1
A(q)
A(q) = 1 + a1q−1 + ................ + anaq
−na
B(q) = b1 + b2q−1 + ............... + anbq
−nb
where, the numbers na and nb are the orders of the respective polynomials, nk is the
number of delays from input to output. Another very common model structure is the
extended auto-regressive, moving average model (ARMAX):
G(q) = q−nk B(q)A(q)
, H(q) = C(q)A(q)
C(q) = 1 + c1q−1 + ................ + cncq
−nc
The prediction error can be computed having observed input-output data:
e(t) = H−1(q) [y(t)−G(q)u(t− nk)]
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics94
3.2.3.2 Uncertainty and disturbance model for nonlinear missile
Let assume that the disturbed plant or system is described as P and corresponding to
the theory discussed in Section 3.2.3.1, we get P as:
P = P + ∆P
where, P = 1s
and it is same for the all three channels. The values of P can be obtained
directly by using the MATLAB system identification toolbox. Now the ∆P ′s can be
written as:
∆pitch = −7.455e−006s6−s5+3.568s4−4.871s3+3.02s2−0.7227s+3.928e−005s6−3.568s5+4.871s4−3.02s3+0.7227s2−3.928e−005s
∆yaw = 2.455e−006s5−s4+2.985s3−3.182s2+1.34s−0.1435s5−2.985s4+3.182s3−1.34s2+0.1435s
∆roll = −0.007201s5+0.005238s4+0.001804s3−4.29e−005s2+0.0002332s+2.292e−018s5−3.051s4+3.913s3−2.642s2+0.923s−0.1419
Figures 3.6, 3.7, 3.8 shows that the maximum gain of the uncertainties and disturbances
Figure 3.6: Characteristics of ∆pitch
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics95
Figure 3.7: Characteristics of ∆yaw
Figure 3.8: Characteristics of ∆roll
can be reached up to a maximum of 20 dB. So the upper limit for γ can be obtained as
γ = 1‖∆‖∞ = 0.1
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics96
3.2.3.3 Robust controller formulation
With reference to the previous Section 2.2.1, the linearized input-output relation of the
missile model can be rewritten as
[y1 y2 y3
]T
=[
v1 v2 v3
]T
(3.20)
Now it is possible to find out three separate robust control law for designing v1, v2 and v3
for the linearized inner rate loop as discussed above. Now the remaining part of the design
is mainly concerned with the choice of weighting functions namely W1 and W3 here. In
a Riccati-based approach, the standard practice is to choose the weight W1(s) as a high
gain low-pass filter to adjust the tracking error of the system in low frequencies. The
weight W3(s) should be a high pass filter to shape the uncertainties in high frequencies.
After several trial and error effort it became possible to find the weights for which the
γ−optimal controller exists. The weights W1(s) and W3(s) are given by,
W1(s) = 100s+100
W2(s) = 0.01 s+0.04s+0.005
Note that same weights have been used for pitch, yaw roll channels. The multiobjective
(disturbance rejection and performance) feature of LMI was accessed through suitably
defined objective in the argument of the function hinflmi of the LMI Toolbox in MATLAB.
The order of the controller (nk) obtained from the LMI solution is 3. Some numerical
results have been given below for pitch plane only. Calculations for yaw and roll plane
will follow a similar pattern. According to the design steps described in the Section 3.2.2,
the values of X and Y are as follows:
X =
80877952.345 717.086 −805.697
717.086 0.0481 −0.0541
−805.697 −0.0541 2871.522
, Y =
89341574.796 −3.2181 −93147.491
−3.2181 705647.315 0.0160
−93147.491 0.016 465.7378
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics97
The value of γ is 0.08. Note that the value of γ is less than the upper limit of 0.1. So, in
this way the solvability condition has been satisfied with X,Y as positive definite matri-
ces. The augmented system matrices are given by:
A =
0 0 0
−1 −100 0
1 0 −0.005
, B1 =
0
1
0
, B2 =
1
0
0
C1 =
[0 100 0
0.01 0 0.00035
], C2 =
[−1 0 0
], D11 =
[0
0
], D12 =
[0
0
]
D21 = 1, D22 = 0
Now assuming
X1 =
1 0 0
0 1 0
0 0 1
, we can get the value of X2 as:
X2 =
−7.225e15 −245732514.558 7533578762612.71
−64065632132.52 −31695.417 66794823.456
72249781331.65 35595.128 −76386100.332
and from the equation 3.16, the value of XL obtained as
XL =
89341574.796 −3.2181 −93147.491 0.999 −2.123e− 10 −4.845e− 12
−3.2181 705647.315 0.016 2.392e− 21 1 2.31e− 18
−93147.4916 0.0160 465.737 1.483e− 17 −4.721e− 13 1
1 0 0 1.414e− 8 1.561e− 15 2.828e− 6
0 1 0 1.565e− 15 1.417e− 6 −4.745e− 11
0 0 1 2.828e− 6 −4.745e− 11 0.0027
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics98
and the value of HXLis obtained as:
HXL=
−186288.54 −705325.48 931.45 1.48e− 17 −1 1 −3.21 0 0.01
−705325.48 −141129463.1 −1.60 −2.39e− 19 −100 −2.31e− 16 705647.31 100 0
931.45 −1.60 −4.65 −7.41e− 20 2.36e− 15 −0.005 0.016 0 0.00035
0 0 0 0 0 0 0 0 0
−1 −100 0 0 0 0 1 0 0
1 0 −0.005 0 0 0 0 0 0
−3.2181 705647.31 0.016 2.39e− 21 1 2.31e− 18 −0.08 0 0
0 100 0 0 0 0 0 −0.08 0
0.01 0 0.00035 0 0 0 0 0 −0.08
Now from (3.18) it is possible to find out the robust controller as:
Jp =
[Ak Bk
Ck Dk
]=
−1658.461 −20.183 1089742.141 0.0003
−4.014 −0.0669 3519.74 −0.00042
1.9781 0.0277 −1480.52 3.8149
−1658.467 −20.183 1089747.94 0
3.2.3.4 Simulation results
The simulation results given in this section in order to show comparative performance
of FBLC with linear control (FBLCL) and FBLC with robust control (FBLCR) under
the following disturbance conditions shown in following Table 3.1. Note that, simulation
results have not yet been verified with seeker based closed loop guidance. Results shown
in this chapter are obtained from detailed 6-DOF model with open loop guidance. From
(2.10) it is evident that if the system is properly feedback linearized then (2.10) should be
satisfied in all the three channels pitch, yaw and roll. Here y1 = q, y2 = r, y3 = p. Figure
3.9 illustrates the comparison of FBLC with a linear rate loop controller (FBLCL) and
FBLC with a robust rate loop controller (FBLCR) in pitch plane. It can be seen that
the linearization for FBLCl is very poor. On the other hand the FBLCR shows better
performance in terms of linearization. From the figure it can be observed that maneuvers
for the missile take place around t = 0.5 unit and t = 0.8 unit. So, at these zones of flight,
the variations in aerodynamic coefficients and magnitude of disturbances become higher.
Figure 3.9 shows that the reference v1 and q have been overlapped each other when robust
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics99
Aerodynamic coefficients variation in %CL ±80CS ±10CN ±10Cm ±10Cn ±10Clζ ±10Clη ±10Cnδ
±10Cnη ±12Cmη ±12Thrust misalignment variation in %TmX
±5TmY
±2TmZ
±3
Table 3.1: Variation in aerodynamic coefficients and thrusts in x− y − z directions
control law has been applied. The same scenario can be seen for yaw and roll channel in
Figures 3.10 and 3.11 respectively. The last three Figures in 3.12 describe the control
input requirements in pitch,yaw and roll channels for robust control law. We can see that
the control input requirement is higher for FBLC and it varies rapidly. Figure 3.13 shows
the rate of control surface deflection and we can observe that except manuevering zone
the rate is well below the maximum limit (normalized to 1).
Note that we have not implemented this H∞ control law in closed loop guidance and
seeker. One main reason is that finding suitable weighting functions W1 and W3 requires
extensive efforts. We have left this part as future work.
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
Time(Normalised)
qd
ot
and
V1 (
No
rmal
ised
)
FBLC Controller
qdot
v1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−0.05
0
0.05
0.1
Time (Normalised)
qd
ot
and
V1 (
No
rmal
ised
)
Robust Controller
qdot
v1
Figure 3.9: Comparison of FBLC and robust controller in pitch plane
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics101
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
Time (Normalised)
rdo
t an
d V
2 (N
orm
alis
ed)
FBLC Controller
rdot
v2
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−0.05
0
0.05
Time (Normalised)
rdo
t an
d V
2 (N
orm
alis
ed)
Robust Controller
rdot
v2
Figure 3.10: Comparison of FBLC and robust controller in yaw plane
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics102
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−1.5
−1
−0.5
0
0.5
1
1.5
Time (Normalised)
pd
ot
and
V3 (
No
rmal
ised
)
FBLC Controller
pdot
v3
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
Time (Normalised)
pd
ot
and
V3 (
No
rmal
ised
)
Robust Controller
pdot
v3
Figure 3.11: Comparison of FBLC and robust controller in roll plane
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics103
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−5
0
5
10
Time (Normalised)
Pit
ch D
efle
ctio
n
RCFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−5
0
5
Time (Normalised)
Yaw
Def
lect
ion RC
FBLC
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−10
0
10
20
Time (Normalised)
Ro
ll D
efle
ctio
n RCFBLC
Figure 3.12: Control deflection comparison
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics104
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−1
−0.5
0
0.5
Time (Normalised)No
rmal
ized
Pit
ch D
efle
ctio
n R
ate
RCFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−1
0
1
Time (Normalised)No
rmal
ized
Yaw
Def
lect
ion
Rat
e
RCFBLC
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1−1
−0.5
0
0.5
Time (Normalised)No
rmal
ized
Ro
ll D
efle
ctio
n R
ate
RCFBLC
Figure 3.13: Control deflection rate
Chapter 3. H∞ Control of Feedback Linearized Inner Rate Loop Dynamics105
3.3 Comments
This chapter has demonstrated the use of robust feedback linearization for tackling the
aerodynamic uncertainties through a short range surface-to-surface homing missile. The
centralized design of the multivariable controller has been formulated as a multiobjective
optimization problem in the LMI framework. The solution is numerically sought through
the LMI solver. The performance robustness of the designed controller has been verified
through a realistic and practical 6-DOF simulation platform. Feedback linearization along
with a robust control law shows pretty good performance over a nominally designed feed-
back linearizing controller and thus it presents a robust feedback linearization approach.
Chapter 4
Sliding Mode control after Feedback
Linearization
4.1 Introduction
In this chapter, a robust control structure known as sliding mode control or variable
structure control, has been applied for designing rate loop controller. A variable structure
system is one whose structure can be changed or switched abruptly according to a certain
switching logic whose aim is to produce a desired overall behavior of the system. The
simplest example of variable structure systems are relay or on-off systems, in which the
control input can have only two values, on or off. Similar to the previous Chapter 3, sliding
mode controller has been formulated for the feedback linearized plant. We have already
discussed that the success of feedback linearization approach is hinged on the availability of
the accurate description of the model [14]. Indeed severe model uncertainty mainly due to
the aero-coefficients may degrade the performance of the feedback linearization approach.
In this regard, some robust scheme [28] such as sliding mode control is required. Figure
4.1 describes the proposed control structure and one can observe that this structure is
similar to the control structure proposed in Figures 2.1 and 3.1. The only difference is
that instead of linear controller block ’LC’ in Figure 2.1 and robust control block ’RC’ in
Chapter 4. Sliding Mode control after Feedback Linearization 107
Figure 4.1: Block diagram of the system representing robust feedback linearization foroutputs q, r and p
Figure 3.1, a new sliding mode control block ’SMC’ has been introduced. In Section 3.2 we
have discussed how uncertainties and various disturbances affect the nominally feedback
linearized plant. From (2.10) and (3.20), we can write the linearized input-output relation
of the plant model with q, r and p as outputs, as
y1
y2
y3
=
v1
v2
v3
(4.1)
In Chapter 3 we have designed H∞ robust control law for the new inputs v = [v1, v2, v3]T
to tolerate the model uncertainties and disturbances. Here in next sections we will discuss
about the formulation of sliding mode control law instead of H∞ control law for designing
v.
Chapter 4. Sliding Mode control after Feedback Linearization 108
4.2 Formulation of sliding mode controller
In this section a very short discussion on sliding modes which is a particular approach to
the design of variable structure systems, has been introduced. Developed in the Soviet
Union more than 40 years ago, sliding mode controllers differ from simpler relay controllers
in that they rely on extremely high speed switching among the control values. As discussed
already total design has been carried out in two steps.
• Performing input-output linearization of the nominal plant.
• Formulation of robust sliding mode control law for that feedback linearized plant.
We consider the system described by
·x (t) = f (x (t)) + g (x (t)) u (t)
y (t) = h (x (t)), (4.2)
where x(t) is the n-dimensional plant state, u is m-dimensional plant input, y is m-
dimensional plant output, f : Rn → Rn and g : Rn → Rn ×Rm and h : Rn → Rm are
smooth functions. Sliding mode controller for system 4.2 can be designed by the following
steps.
• STEP I : Performing input-output feedback linearization
• STEP II: Formulation of sliding mode control law for designing v
STEP I has been already done in Section 2.2.1 and the relevant equations that we will
demonstrate in this chapter will be exactly the same as Section 2.2.1.
4.2.1 Step II: Formulation of sliding mode control law for de-
signing v
The main idea behind sliding mode control is to choose a suitable surface in state space,
typically a linear hypersurface, called the switching surface, and switch the control input
Chapter 4. Sliding Mode control after Feedback Linearization 109
on this surface. The control input is then chosen to guarantee that the trajectories near
the sliding surface are directed toward the surface. Ideally then, any control input will
suffice so long as the resulting trajectories are pointing toward the surface. Once the
system is trapped on the surface, the closed loop dynamics are completely governed by
the equations that define the surface. In this way, since the parameters defining the surface
are chosen by the designer, the closed loop dynamics of the system will be independent
of perturbations in the parameters of the system and robustness is achieved. The design
of sliding mode control can be broken down into two steps:
• Specifying a suitable sliding surface
• Achieving the sliding condition and designing system dynamics on the surface
4.2.1.1 Specifying sliding surfaces
Let ei = yi − ri with ri the reference trajectories, be the tracking error for the output yi
and let
ei = yi − ri =[
ei ei ... ... eri−1i
]T
be the tracking error vector. Furthermore, let us define a time-varying surface Si(t) in
the state space Rri−1 by the scalar equation si(yi; t) = 0, where
s(yi; t) = (d
dt+ k)n−1ei (4.3)
and k is strictly positive constant. Or we can write (4.3) as
si (t) = e(ri−1)i + ki(ri−1)e
(ri−2)i + · · ·+ ki2e
(1)i + ki1e
(0)i + ki0
∫eidt (4.4)
In this way we can define m sliding surfaces si, i= 1, · · · ,m, based on the input-output
linearized system given in (4.1) ki(r−1), · · · , ki0 are such that
λri + ki(ri−1)λri−1 + · · ·+ ki1λ + ki0 (4.5)
Chapter 4. Sliding Mode control after Feedback Linearization 110
is Hurwitz polynomial. For the tracking task to be achievable using a finite control law
v, the initial state ri(0) must be such that
ri(0) = yi(0) (4.6)
Given initial condition (4.6), the problem of tracking ri = yi is equivalent to that of
remaining on the surface Si(t)∀t ≥ 0; indeed si ≡ 0 represents a linear differential equation
whose unique solution is ei = 0, given initial condition (4.6). Thus the problem of tracking
the n-dimensional vector ri can be reduced to that of keeping the scalar quantity s at
zero.
4.2.1.2 Achieving sliding condition
The closed-loop system is said to satisfy the sliding condition if the following applies [27].
1
2
ds2i
dt≤ −ηi |si| , (ηi > 0) (4.7)
where ηi, i = 1, · · · ,m are positive numbers. Note that sliding condition will make si (t) =
0 and si (t) = 0 in a finite time. Since is a stable differential equation, satisfaction of
si (t) = 0 by ei(t) in turn leads to asymptotic tracking.
Let
·s
def=
·s1
...·
sm
, y(ρ) =
y(r1)1...
y(rm)m
(4.8)
sgn (s) = [sgn (s1) , · · · , sgn (sm)]
where sgn(·) is the signum function. Now it has been reported in [36] that a control law
that achieves the sliding condition (4.7) is given by
u = E−1(( ·
s−y(ρ))−M − λsgn (s)
)(4.9)
Chapter 4. Sliding Mode control after Feedback Linearization 111
where λ = diag [λ1, · · · , λm] with λi a positive number greater than the given positive
number ηi. Note that
( ·si−yri
i
)= −rri
i + ki(ri−1)e(ri−1)i + · · ·+ ki2e
(2)i + ki1e
(1)i + ki0ei
so( ·s−y(ρ)
)does not depend on u. The integral term in (4.4) can be omitted by setting
ki0 = 0. Since the sliding condition also implies si(t) = 0, the asymptotical tracking can
still be achieved by the control law (4.9) as long as, for i = 1, · · · ,m, ki(r−1), · · · , ki1 are
such that λri−1 + ki(ri−1)λri−2 + · · · + ki1 are Hurwitz. In practice the implementation of
variable structure controllers results in control chattering. The ideal behavior of sliding
mode controller is achieved in the theoretical limit as the switching frequency becomes
infinite. In practice the small, but nonzero delay in control switching will cause the tra-
jectory to slightly overshoot or undershoot the switching surface each time the control is
switched. This is known as chattering. It has been observe that more the model uncer-
tainties, the chattering becomes severe. The approach taken in this thesis to overcome
the undesirable chattering is to introduce what is known as boundary layer around the
sliding surface and approximate the switching control law by a continuous control inside
this boundary layer. Thus the discontinuous control law sgn(si) is often replaced by the
saturation function sat( si
εi) where
sat (x) = x, if |x| ≤ 1
sat (x) = sgn (x) , if |x| ≥ 1(4.10)
4.2.2 Application to the STT missile model
As we have already tried input-output feedback linearization approach with nominal mis-
sile model in Section 2.2.1, let us start from the linearized input-output relation obtained
in 2.10,
y1
y2
y3
=
v1
v2
v3
(4.11)
Chapter 4. Sliding Mode control after Feedback Linearization 112
and we can easily observe that
r1 = 1, r2 = 1, r3 = 1. So, referring to equation (4.4) we can say that in this particular
case there will be three sliding surfaces namely
s1 = e1, s2 = e2, s3 = e3 where
e1 = q − qd
e2 = r − rd
e1 = p− pd
(4.12)
As discussed earlier qd, rd, pd can be obtained from outer loop or lateral acceleration error
dynamics. So from (4.8) we can write,
(s− y(ρ)) =
s1 − q
s2 − r
s3 − p
=
−k11qd
−k21rd
−k31pd
Now the control input can be easily found out from (4.9) with a boundary layer vector
ε =
ε1
ε2
ε3
=
0.1
0.2
0.2
The design parameters are given by k11, k21 and k31. These parameters have been chosen
by trial and error method. It has been observed for this particular case that using higher
values of kijs, the system becomes unstable and shows more chattering. For lower values
of kijs, the system performance shows a few second delay with respect to the nominal
FBLC. Thus we have chosen these kij values where the system performance is more or
less satisfactory. The 6-DOF simulation results have been shown in the next section to
check the system performance with sliding condition.
Chapter 4. Sliding Mode control after Feedback Linearization 113
4.2.2.1 Simulation Results
The simulation results shown for the sliding mode control is based on the detailed 6-DOF
given by (1.3) and (1.4). First we have tried with the same variations in aerodynamic
coefficients given in Table 3.1. But the control law (4.9) failed to handle that much
of variations and the system became unstable. A huge chattering was present in that
case. In the second stage we have decreased the variations in aerodynamic coefficients.
The simulations have been performed for open loop guidance as well as for closed loop
Aerodynamic coefficients variation in %CL ±30CS ±5CN ±5Cm ±5Cn ±5Clζ ±5Cnη ±5Thrust misalignment variation in %TmX
±1
Table 4.1: Variation in aerodynamic coefficients and thrusts in x− y − z directions
guidance with seeker. Open loop simulations have been performed to compare with the
H∞ robust control performance presented in Chapter 3. Figures 4.2,4.3 and 4.4 show
the input-output linearization with the sliding mode controller (FBLCSM) in comparison
with FBLC with linear controller (FBLCL) with guidance and seeker in open loop as done
before for the H∞ controller. But it can be seen that much improvement has not been
noticed for FBLCSM.
Figure 4.5 shows the input deflections for sliding mode and FBLC. Here also we can
see that in some cases sliding mode demands more input for almost the same performance.
So if we compare the open loop performances of SMC with H∞ control, we can see that
the latter one can handle more aerodynamic variations than the former one. As stated
before sliding mode control law has also been tested in FORTRAN 6-DOF platform with
Chapter 4. Sliding Mode control after Feedback Linearization 114
closed loop guidance and seeker. Some of these results have been explained below. We
will see that openloop and closed loop simulation results gives almost same performance
for sliding mode controller in comparison with FBLC. Figure 4.6 shows the pitch,yaw and
roll rates for FBLC and sliding mode controller. Here we can see that FBLC and the
sliding mode controller behave similarly.
Figures 4.7 and 4.8 show almost similar response for α, β and gimbal angle for both
FBLCL and FBLCSM. Figure 4.9 shows the pitch and yaw latax profiles from which
it can be concluded that performance is similar for small aerodynamic perturbations as
shown in Table 4.1. For heavy perturbation in aerodynamic coefficients, the FBLCL as
well as sliding mode controller become unstable. Figures 4.10 and 4.11 show pitch-yaw-
roll channel deflections and the fin distributions respectively. As expected, the sliding
mode controller takes more fin than the FBLCL. The rate of fin deflections are shown in
Figure 4.12 and the rate is below the maximum level.
Figure 4.13 represents the variation of aerodynamic coefficients and for small pertur-
bations in aerodynamic coefficients the response for FBLC and the sliding mode control
remain more or less the same. Thus we have seen that sliding mode control implemented
here can tolerate very small perturbations. Performance wise FBLCL and FBLCSM are
more or less similar and we have not found any significant improvements that is claimed
in [36].
Chapter 4. Sliding Mode control after Feedback Linearization 115
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−5
0
5
x 10−3
Time(Normalised)
qd
ot
and
V1 (
No
rmal
ised
)
FBLC Controller
qdot
v1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−0.05
0
0.05
0.1
Time (Normalised)
qd
ot
and
V1 (
No
rmal
ised
)
SMC Controller
qdot
v1
Figure 4.2: Linearization in pitch channel
Chapter 4. Sliding Mode control after Feedback Linearization 116
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−1
−0.5
0
0.5
1
x 10−3
Time (Normalised)
rdo
t an
d V
2 (N
orm
alis
ed)
FBLC Controller
rdot
v2
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−0.04
−0.03
−0.02
−0.01
0
0.01
0.02
0.03
Time (Normalised)
rdo
t an
d V
2 (N
orm
alis
ed)
SMC Controller
rdot
v2
Figure 4.3: Linearization in yaw channel
Chapter 4. Sliding Mode control after Feedback Linearization 117
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−2
−1
0
1
2x 10
−3
Time (Normalised)
pd
ot
and
V3 (
No
rmal
ised
)
FBLC Controller
pdot
v3
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−0.04
−0.02
0
0.02
0.04
Time (Normalised)
pd
ot
and
V3 (
No
rmal
ised
)
SMC Controller
pdot
v3
Figure 4.4: Linearization in roll channel
Chapter 4. Sliding Mode control after Feedback Linearization 118
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−4
−2
0
2
Time (Normalised)
Pit
ch D
efle
ctio
n
SMCFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−10
−5
0
Time (Normalised)
Yaw
Def
lect
ion
SMCFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
−10
−5
0
Time (Normalised)
Ro
ll D
efle
ctio
n
SMCFBLC
Figure 4.5: Control deflection comparison
Chapter 4. Sliding Mode control after Feedback Linearization 119
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−1
−0.5
0
0.5
1
Normalised Time
Nor
mal
ised
Pitc
h R
ate
Comparative Performance in 6−DOF Simulation
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−4
−2
0
2
Normalised Time
Nor
mal
ised
Yaw
Rat
e
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
0
1
Normalised Time
Nor
mal
ised
Rol
l Rat
e
SlidingFBLC
Figure 4.6: Pitch, yaw and roll rates
Chapter 4. Sliding Mode control after Feedback Linearization 120
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
−0.5
0
0.5
1
Normalised Time
Nor
mal
ised
Alp
ha (
Bod
y)
Comparative Performance in 6−DOF Simulation
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−2
−1
0
1
Normalised Time
Nor
mal
ised
Bet
a (B
ody)
Comparative Performance in 6−DOF Simulation
SlidingFBLC
Figure 4.7: α and β
Chapter 4. Sliding Mode control after Feedback Linearization 121
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
Normalised Time
Nor
mal
ised
Gim
bal A
ngle
(El)
Comparative Performance in 6−DOF Simulation
SlidingSliding
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−2
0
2
4
Normalised Time
Nor
mal
ised
Gim
bal A
ngle
(Az) Sliding
Sliding
Figure 4.8: Gimbal angle profile
Chapter 4. Sliding Mode control after Feedback Linearization 122
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−1
−0.5
0
0.5
Normalised Time
Nor
mal
ised
Pitc
h la
tax
Comparative Performance in 6−DOF Simulation
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−1
0
1
2
Normalised Time
Nor
mal
ised
Yaw
lata
x
Comparative Performance in 6−DOF Simulation
SlidingFBLC
Figure 4.9: Pitch and yaw latax
Chapter 4. Sliding Mode control after Feedback Linearization 123
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−2
−1
0
1
Normalised Time
Eff.
pitc
h de
fln (
δ PB
)
Comparative Performance in 6−DOF Simulation
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−2
0
2
Normalised Time
Eff.
yaw
def
ln (
δ YB
) SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−3
−2
−1
0
Normalised Time
Rol
l def
ln (
δ R )
SlidingFBLC
Figure 4.10: Pitch,yaw and roll deflections
Chapter 4. Sliding Mode control after Feedback Linearization 124
5 10 15 20
−2
−1
0
1
2
Normalised Time
Nor
mal
ised
Fin
−1 d
efl
n SlidingFBLC
0.2 0.4 0.6 0.8 1
−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
Nor
mal
ised
Fin
−2 d
efl
n
SlidingFBLC
0.2 0.4 0.6 0.8 1
−2
−1
0
1
Normalised Time
Nor
mal
ised
Fin
−3 d
efl
n
SlidingFBLC
0.2 0.4 0.6 0.8 1
−1.5
−1
−0.5
0
0.5
1
1.5
Normalised Time
Nor
mal
ised
Fin
−4 d
efl
n SlidingFBLC
Figure 4.11: Fin demands
Chapter 4. Sliding Mode control after Feedback Linearization 125
5 10 15 20
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Normalised Time
Nor
mal
ised
Fin
−1 d
efl
n r
ate Sliding
FBLC
0.2 0.4 0.6 0.8 1
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Normalised Time
Nor
mal
ised
Fin
−2 d
efl
n r
ate Sliding
FBLC
0.2 0.4 0.6 0.8 1
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Normalised Time
Nor
mal
ised
Fin
−3 d
efl
n r
ate Sliding
FBLC
0.2 0.4 0.6 0.8 1
−0.5
0
0.5
Normalised Time
Nor
mal
ised
Fin
−4 d
efl
n r
ate Sliding
FBLC
Figure 4.12: Fin deflection rate
Chapter 4. Sliding Mode control after Feedback Linearization 126
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1−0.3
−0.2
−0.1
0
Comparative Performance in 6−DOF Simulation
Rol
ling
mom
ent c
oeffi
cien
t CL
Normalised Time
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
0
0.5
1
Sid
e fo
rce
coef
ficie
nt C
S
Normalised Time
SlidingFBLC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
−8
−6
−4
−2
0
2
Yaw
ing
mom
ent c
oeffi
cien
t Cny
Normalised Time
SlidingFBLC
Figure 4.13: Force and moment Coefficients
Chapter 4. Sliding Mode control after Feedback Linearization 127
4.3 Comments
The above simulation results describe the performance of sliding mode controller for a
feedback linearized skid-to-turn homing missile. The main idea in designing sliding mode
controller is to tackle model uncertainties which are inherently present in the system. The
simulation results presented here constitute a case study of robust feedback linearization
where the parametric uncertainties are small enough. In comparison with the Chapter 3,
one can say that sliding mode control law, does not give a very satisfactory performance
even in small aerodynamic perturbations shown in Table 4.1. However it has been tested
successfully in closed loop seeker based detailed 6-DOF model. On the other hand, H∞robust control law, gives satisfactory performance with huge aerodynamic uncertainties
shown in Table 3.1, has been tested in open loop guidance only.
Chapter 5
Conclusions
In this thesis, an approach has been presented for controlling a highly maneuverable short
range surface-to-surface missile. Our choice has been oriented by the industrial as well
as academic context. The necessity of increasing missile performances with reduced flight
range has motivated new investigations in order to estimate not only the potential of
recent theoretical control methods, but also their application level, in order to meet the
industrial demands.
In this thesis, the application of input-output linearization has been presented for a short
range surface-to-surface skid-to-turn homing missile. Here, a state dependent nonlinear
feedback control law has been designed for improving the overall missile performance in
view of reduced range, less control effort etc. Aerodynamic missiles generally suffer from
coupling effect among pitch-yaw-roll axes. This coupling effect results in large variations
in aerodynamic coefficients, side-slip, control effort in yaw etc. These in turn may cause
fin saturation and gimbal angle limit violation which cause the missile to lose its actual
track. The conventional three loop autopilot generally fails to solve this coupling prob-
lem. To get rid of this problem a nonlinear multivariable approach has been proposed.
In this approach, the whole missile model has been feedback linearized. Then a linear
proportional controller has been designed for obtaining the tracking performance in inner
rate loop. Thus the pitch-yaw-roll dynamics, specifically rate dynamics, of the missile
becomes linear and decoupled which are the most essential requirement for most of the
Chapter 5. Conclusions 129
aerodynamic missiles. The simulation results show the evidence of linearization and de-
coupling and the improved performance of the Feedback Linearizing Control (FLC) law
over the traditional three loop autopilot.
Although input-output linearization is successful in decoupling and linearizing nonlinear
dynamics, it fails to handle the same problem when plant dynamics become uncertain.
Generally short range aerodynamic missiles, in which boost time is small enough, suffers
from the problem of uncertainties due to aerodynamic coefficients. This problem becomes
more acute when the missile tries to turn rapidly to track the target. In view of this prob-
lem, two different robust control laws have been formulated. In both cases the nominal
missile dynamics has been input-output linearized as before. In the first case, a linear
matrix inequality approach to H∞ control has been presented along with its application to
uncertain missile dynamics. It has been shown through open loop MATLAB simulation
results that upto 100% uncertainty in aerodynamic coefficients can be tolerated through
this method.
In the second case, another approach of sliding mode control has been considered. Here
also, sliding mode control law has been applied to an input-output linearized plant. Sim-
ulation results have been presented with closed loop guidance and seeker and it has been
seen that this method is not so useful when the system poses very high uncertainties in
terms of aerodynamic coefficients.
All of the above control laws presented in this thesis, require whole state measurements.
But only some of the states are measured (rates in three channels) and the rest are un-
measured (velocity components in three channels). That means the unmeasured states
are to be estimated to realize the control laws practically. For that purpose, a nonlinear
Luenberger observer has been used. The forward velocity of the missile is assumed to
be known from thrust profile and observer equations have been formulated in terms of
α and β as input to the observer i.e., the output of the system are direct functions of α
and β. Thus α and β have been first estimated and then velocity components are derived
easily from standard mathematical relations. Simulation results show good tracking per-
formances in presence of sensor noise.
Chapter 5. Conclusions 130
The main contributions of the thesis can be summarized as follows:
1. Minimized roll rate even during high angle of attack maneuver.
2. Good decoupling among the three axes during pitch maneuver.
3. Estimation of the unmeasured states of the system using a nonlinear observer re-
quired for the computation of nonlinear feedback.
4. Design of a robust H∞ control law that retains performance even with aerodynamic
uncertainties.
5. Design of a robust sliding mode controller to tolerate the uncertainties caused due
to aerodynamic coefficients.
Some recommendations can be drawn in view of the overall work done in this thesis.
• Outer loop feedback linearization along with the rate loop feedback linearization
will make the whole plant dynamics linear and decoupled.
• Some more advanced robust control laws such as µ−synthesis, H∞ loop shaping etc,
are to be tested in order to cater more aerodynamic uncertainties and disturbances.
• Observer gain has to be adaptive and robust irrespective of different flight condi-
tions.
The analysis and numerical results presented in this thesis amply demonstrate the fea-
sibility of designing nonlinear control systems for the next generation high performance
missile autopilots. Nonlinear design methods have the potential for enhancing missile
performance while simplifying the design process. This can result in a lighter and more
accurate missile system.
Appendix A
A Brief Theory of Feedback
Linearization
A.1 Introduction
A single input-single output control affine system obeys the following state and output
equation x = f(x) + g(x)u
y = h(x), (A.1)
x ∈ Rn is a vector, u is a scalar input, y is a scalar output,f : Rn → Rn and g : Rn → Rn
are vector fields and h : Rn → R is a scalar function.Suppose that f(·), g(·) and h(·) are
differentiable function.
Obviously y does not depend explicitly on u since u is not an argument of function
h. If u is changed instantaneously, there will be no immediate change in y. The change
comes gradually via x. To check the corresponding behavior of ywe can differentiate the
Chapter A. A Brief Theory of Feedback Linearization 132
output equation [54]:
y = ∂h∂x
x = ∂h∂x
(f(x) + g(x)u)
= ∂h∂x
f(x) + ∂h∂x
g(x)u
(A.2)
This shows that y depends directly on u if and only if [∂h/∂x] g(x) 6= 0 we will say that
the system has relative degree 1 if [∂h/∂x] g(x) 6= 0
Now, let us assume that [∂h/∂x] g(x) = 0 differentiating the output y once more, we
obtain:
y =∂
∂x
[∂h
∂x
]f(x) +
∂
∂x
[∂h
∂xf(x)
]g(x)u (A.3)
The system is said to have relative degree 2 if:
∂h
∂xg(x) = 0 and
∂
∂x
[∂h
∂xf(x)
]g(x) 6= 0 (A.4)
The idea of relative degree can be easily generalized with the help of the following notation.
Definition 1 (Lie Derivative) Let λ : Rn → R be a differentiable function and f a
vector field, both defined on an open subset U of Rn.The derivative of λ along f or Lie
derivative of λ along f is given by the inner product
⟨∂λ
∂x, f(x)
⟩=
∂λ
∂xf(x) (A.5)
The lie derivative of λ along f is usually denoted as Lfλ, so that:
Lfλ(x) =n∑
i=1
∂λ
∂xi
fi(x) (A.6)
Chapter A. A Brief Theory of Feedback Linearization 133
If Lfλ is again differentiated along another vector field g, the following is obtained:
LgLfλ(x) =∂ (Lfλ)
∂xg(x) (A.7)
This operation could be used recursively along the same vector field.Lkfλ indicates that λ
is being differentiated k times along f such that
Lk
fλ(x) =∂(Lk−1
f λ)∂x
f(x)
L0fλ(x) = λ(x)
(A.8)
Definition 2 (Relative Degree) The SISO system given by Equation A.1 is said to
have a relative degree r at a point xo if
1. LgLk−1f h(x) = 0 ∀ x in a neighborhood of xok = 1, ..., r − 1
2. LgLr−1f h(xo) 6= 0
The following formula describing the time derivatives of the output is an immediate
consequences of Definition 2 :
dky
dtk= y(k) =
Lk
fh(x), k = 1, ..., r − 1
Lkfh(x) + LgL
k−1f h(x)u, k = r
, (A.9)
Remark 1 For SISO linear systems:
x = Ax + Bu
y = Cx(A.10)
Chapter A. A Brief Theory of Feedback Linearization 134
The relative degree is the difference in degree between the nominator and denominator of
its equivalent transfer function:
G(s) = C(sI − A)−1B (A.11)
Equivalently, the relative degree of a SISO linear system is the positive integer r such
that:
LgLr−1f h = CAr−1B 6= 0 (A.12)
Definition 3 (Strong Relative Degree) A system is said to have a strong relative de-
gree if the relative degree is r for all xo ∈ Rn
A.1.1 Input-output linearisation
Consider a system with a strong relative degree r.Then we have that the rth derivative of
the output is given by:
y(r) = Lrfh + LgL
r−1f hu, LgL
r−1f h 6= 0 (A.13)
There is an interesting possibility. Introduce the feedback law:
u =1
LgLr−1f h
(v − Lr
fh), (A.14)
Where v is a reference signal The resulting relationship is between v and y(r) is:
y(r) = v (A.15)
Chapter A. A Brief Theory of Feedback Linearization 135
Which is a linear system. Taking the Laplace transform:
y(s) =1
snv(s) (A.16)
By using the feedback law A.14, we have obtained a system that is linear from the reference
signal v to the output y.
Proposition 1 Suppose a system of the form of equation A.1 has relative degree r at a
point xo. Define,
ϕ1(x) = h(x)
...
ϕr(x) = Lr−1f h(x)
, (A.17)
If r is strictly less than n, it is always possible to find n−r arbitrary functions ϕr+1(x), ..., ϕn(x)
such that the mapping
Φ(x) = [ϕ1(x), ..., ϕr(x), ..., ϕn(x)]T (A.18)
has a jacobian matrix which is nonsingular at xo and therefore represents a local coordi-
nates transformation in a neighborhood of xo. Moreover, it is always possible to choose
ϕr+1(x), ..., ϕn(x) in such a way that
Lgϕi(x) = 0, r + 1 ≤ i ≤ n, (A.19)
Chapter A. A Brief Theory of Feedback Linearization 136
Differentiating Equation A.17 with respect to time and using Equation A.9, we obtain:
dφ1(x)dt
= dh(x)dt
= Lfh(x) = φ2(x)...dφr−1(x)
dt= Lr−1
f h(x) = φr(x)dφr(x)
dt= Lr
fh(x) + LgLr−1f h(x)u
, (A.20)
By introducing the variables:
ξ =
ξ1
...
ξr
=
φ1(x)...
φr(x)
(A.21)
η =
η1
...
ηn−r
=
φr+1(x)...
φn(x)
(A.22)
Then Equation A.20 can be written as follows:
dξ1dt
= ξ2(t)...dξr−1
dt= ξr(t)
dξr
dt= b(ξ, η) + a(ξ, η)u
, (A.23)
where, y = ξ1 a(ξ, η) = LgL
r−1f h (Φ−1(ξ, η))
b(ξ, η) = Lrfh (Φ−1(ξ, η))
, (A.24)
Notice that
x = Φ−1(ξ, η), (A.25)
Chapter A. A Brief Theory of Feedback Linearization 137
As far as the other new co-ordinates are concerned, corresponding to η = [η1, ..., ηn−r]T ,
we cannot expect any special structure in the corresponding equations. However, if
φr+1, ..., φn have been chosen such that Lgφr+i = 0, i = 1, ..., (n− r), then
dηi
dt= Lfφr+i (x(t)) , i = 1, ..., (n− r) (A.26)
By defining qi (ξ, η) = Lfφr+i (Φ−1 (ξ, η)) , then we have that the derivatives of ηi can be
written as follows:dηi
dt= qi (ξ, η) , i = 1, ..., (n− r) (A.27)
or, in vector form:dη
dt= q (ξ, η) (A.28)
In summary, the normal form of a control affine SISO nonlinear system of Equation A.1,
with relative degree r around a point xo, is given by:
ξ1 = ξ2
ξ2 = ξ3
...
ξr−1 = ξr
ξr = b(ξ, η) + a(ξ, η)u
η = q(ξ, η)
y = ξ1
(A.29)
Notice that if we choose the feedback law:
u = 1a(ξ,η)
(v − b(ξ, η))
= 1LgLr−1
f h(Φ−1(ξ,η))
(v − Lr
fh(Φ−1(ξ, η)))
= 1LgLr−1
f h(x)
(v − Lr
fh(x)) (A.30)
Chapter A. A Brief Theory of Feedback Linearization 138
Then the resulting system from the reference input v to the output y = ξ1 is linear. The
normal form is given by:
ξ1 = ξ2
ξ2 = ξ3
...
ξr−1 = ξr
ξr = v
η = q(ξ, η)
y = ξ1
(A.31)
Notice, however, that the internal dynamics η = q(ξ, η) are possibly nonlinear, so that
the system has not been fully linearised by the feedback law.
If the feedback law is changed to:
u =1
a (ξ, η)(v − b (ξ, η)− λ0ξ1 − ...− λr−1ξr) (A.32)
Then the normal form is given by:
ξ1 = ξ2
ξ2 = ξ3
...
ξr−1 = ξr
ξr = −λ0ξ1 − ...− λr−1ξr + v
η = q(ξ, η)
y = ξ1
(A.33)
and the transfer function from input to output is:
y(s)
v(s)=
1
λr−1sr + ... + λ0
(A.34)
Chapter A. A Brief Theory of Feedback Linearization 139
It is easy to see that the unforced internal dynamics (or zero dynamics) which are defined
for ξ = 0 (η = −η3) , are asymptotically stable around η = 0, so that the control signal
should be bounded for bounded v and ξ. Notice that ξ is bounded if v is bounded since
with a > 0 the transfer function y(s)/v(s) is stable.
A.2 Multi input multi output systems
The concepts used in the above section for SISO systems, such as input-state linearization,
input-output linearization, normal forms, zero-dynamics, and so on, can be extended to
MIMO systems. For the MIMO case, we consider square systems (i.e., systems with the
same numbers of inputs and outputs) of the following form
x = f(x) + g1(x)u1 + ..... + gm(x)um
y1 = h1(x)
· · ·ym = hm(x)
(A.35)
Where is the state vector,u′is(i = 1, ....., m) are control inputs, y′js(j = 1, ....., m) are
outputs, and , f and g′is are the smooth vector fields, and h′js are smooth scalar functions.
If we collect the control inputs ui into a vector u, the corresponding vectors into a matrix
G, and the outputs into a vector y, the system’s equations can then be compactly written
asx = f(x) + G(x)u
y = h(x)(A.36)
A.2.1 Feedback Linearization of MIMO Systems
The approach to obtain the input-output linearization of the MIMO systems is again to
differentiate the outputs yj of the system until the inputs appear, similarly to the SISO
Chapter A. A Brief Theory of Feedback Linearization 140
case. To start with,
yj = Lfhj +m∑
i=1
(Lgihj)ui, (A.37)
If Lgihj(x) = 0 ∀i ,then the inputs do not appear and we have to differentiate again.
Assume that rjis the smallest integer such that at least one of the inputs appear in y(rj)j ,
then
y(rj)j = L
rj
f hj +m∑
i=1
LgiL
rj−1f hjui, (A.38)
With LgiL
rj−1f hj(x) 6= 0 for at least one i, ∀x ∈ Ω.If we perform the above procedure for
each output yj, we can obtain a total of m equations in the above form, which can be
written compactly as
y(r1)1
· · ·· · ·y
(rm)m
=
Lr1f h1
· · ·· · ·Lrm
f hm
+ E(x)
u1
· · ·· · ·um
, (A.39)
Where the m×m matrix E is defined as
E(x) =
Lg1Lr1−1f h1........LgmLr1−1
f h1
................
................
Lg1Lrm−1f hm........LgmLrm−1
f h1m
(A.40)
Chapter A. A Brief Theory of Feedback Linearization 141
The matrix E(x) is called the decoupling matrix for the MIMO system. If the decoupling
matrix is non-singular in a region Ω around a point x0, then the input transformation
u = −E−1
Lr1f h1
· · ·· · ·Lrm
f hm
+ E−1
v1
· · ·· · ·vm
, (A.41)
yields a linear differential relation between the output y and the new input v
y(r1)1
· · ·· · ·y
(rm)m
=
v1
· · ·· · ·vm
(A.42)
Note that the above input-output relation is decoupled, in addition to being linear. Since
only affects the corresponding output yj , but not the others, a control law of the form
A.41 is called a decoupling control law, or non-interacting control law. As a result of
the decoupling, one can use SISO design on each y − v channel in the above decoupled
dynamics to construct tracking or stabilization controllers. It is useful to formalize the
concept of relative degree for MIMO systems at this point. Since there is a relative degree
associated with each output, the relative degree of the MIMO system is defined by m
integers.
Definition 4 The system A.35 is said to have relative degree (r1, · · · , rm) at x0 if there
exists a neighborhood Ω of x0 such that ∀x ∈ Ω,
• LgiLk
fhj (x) = 00 ≤ k ≤ ri − 11 ≤ i, j ≤ m
• E(x) is non-singular
Chapter A. A Brief Theory of Feedback Linearization 142
The total relative degree of the system is defined by
r = r1 + · · ·+ rm
Let us consider the case of r < n first. A normal forma can be obtained for the system in
a manner similar to the SISO case, as we now show. First, choose as coordinates
ζ11 = h1 (x) ζ1
2 = Lfh1 (x) · · · ζ11 = Lr1−1
f h1 (x)
· · ·ζm1 = hm (x) ζm
2 = Lfhm (x) · · · ζmm = Lrm−1
f hm (x)
(A.43)
These are simply the m outputs yj and their derivatives up to order rj.
Similarly to the SISO case, the r coordinates ζji (j = 1, . . . , m; i = 1, . . . , rj) are in-
dependent and can be used as a partial set of a new state vector. This is because the
gradient vectors
Lifhj (x) 0 ≤ i ≤ rj − 1 1 ≤ j ≤ m
are linearly independent, as can be shown in a manner analogous to SISO case, using the
non-singularity of the decoupling matrix E. Now, let us complete the choice of the new
state vector by choosing n− r more functions η1, . . . , ηn−r (x) which are independent with
respect to each other and to the r coordinates chosen earlier. This can always be done,
based on the Frobenius theorem. However, unlike the SISO case, it is no longer possible
to guarantee that
∀x ∈ Ω Lgiηk (x) = 0 1 ≤ i ≤ m 1 ≤ k ≤ n− r
unless the vector fields g1, . . . , gm are involutive on Ω . As a result, the state equations
for these n− r coordinates will have the input vector u appearing.
Chapter A. A Brief Theory of Feedback Linearization 143
With (ζ, η) as coordinates, the system equations can also be put into a ”normal form”.
Specifically, the external dynamics is
ζj1 = ζj
2
· · ·ζjrj
= aj (ζ, η) +m∑
i=1
bij (ζ, η) ui
where j = 1, 2, . . . ,m, and
aj (ζ, η) = Lrj
f hj (x)
bij (ζ, η) = Lgi
Lfr−1j hj (x)
The internal dynamics is·η = w (ζ, η) + P (ζ, η) u
with (k = 1, . . . , n− r) and (i = 1, . . . ,m)
wk (ζ, η) = Lfηk (x)
Pki (ζ, η) = Lgiηk (x)
Note that P ∈ R(n−r)×m and w ∈ Rn−r. As in the SISO case, feedback law of A.41
renders the (n− r) states η unobservable.
An interesting case of the above input-output linearization corresponds to the total
relative degree being n,i.e.,m∑
j=1
rj = n
In this case, there is no internal dynamics. With the control law in the form of A.41, we
obtain an input-state linearization of the original nonlinear system. With the equivalent
inputs vi designed as in the SISO case, both stabilization and tracking can then be achieved
for the system without any worry about the stability of the internal dynamics. We remark
Chapter A. A Brief Theory of Feedback Linearization 144
that the necessary and sufficient conditions for input-state linearization of multi-input
nonlinear systems are similar to and more complex than those for single input systems.
A.2.2 Zero-dynamics and control design
When designing controllers based on the linear input-output relation in (A.39)[63], one
has to be concerned with the stability of the internal dynamics. It is therefore of interest to
study the stability of the zero-dynamics, an extent case of internal dynamics with output
being exactly zero.Similarly to the SISO case, the zero-dynamics of a MIMO system is
obtained by constraining the output to zero [70],[19].
Definition 5 The zero-dynamics of the MIMO nonlinear system is the dynamics of the
system when the outputs are constrained to be identically zero.
since the constraint that the output identically equal to zero implies that all the
derivatives of the output are zero, we have
ζ(t) ≡ 0
In order to keep the outputs identically zero, the control inputs must be chosen as
u(t) = −E−1(0, η)a(0, η)
where η(t) is the soltion of the differential equations.
η(t) = w(0, η)− P (0, η)E−1(0, η)a(0, η)
with η(0) arbitary. In the original x coordinates, when the system operates in zero-
dynamics, the system states x evolve on the surface.
M∗ =
x ∈ Ω|hj(x) = Lfhj(x) = ..... = Lrj−1f hj(x) = 0, 1 ≤ j ≤ m
Chapter A. A Brief Theory of Feedback Linearization 145
Of course, the initial states x(0) must be chosen to b on this surface. In terms of x(0)
the constraining control inputs u are
u∗(x) = −E−1(x)
Lr1f h(x)
.....
.....
Lrmf hm(x)
(A.44)
The zero-dynamics is given by the equation
x = f(x) + g(x)u∗(x) (A.45)
with the states constrained on the surface M∗
Similarly to the SISO case, we can define the notation of minimum phase systems.
Definition 6 The MIMO nonlinear system A.35 is said to be asymtotically minimum
phase if the zero-dynamics is locally asymtotically stable.
The definition of exponentially minimum phase is similar.
For minimum phase systems, the control design results in section 2.2 can easily ex-
tended to MIMO case.
Appendix B
LMI Approach to H∞ Control
B.1 The Theory of H∞ Control based on LMI Ap-
proach
A brief introduction to the robust control has been given in the Chapter 3. Here we have
given more detailed analysis of robust control formulation.
B.1.1 Singular value decomposition
In classical control theory, Bode magnitude plot gives the gain of a SISO system at
different frequencies. But if the system is a MIMO one then the Bode plots of all the
elements of the transfer matrix give little idea about the gain of the system since it
undermines the interactions among the elements. Eigen values of the transfer matrix can
be an answer but then they are not a very good measure of the gain or size of a matrix.
Moreover, eigen values are limited to square transfer matrices (equal no. of inputs and
outputs) only. Therefore, a more general entity is required and singular values have been
found suitable.
Chapter B. LMI Approach to H∞ Control 147
The singular values of a complex valued matrix A are defined as the positive square
roots of the eigen values of A × A. Singular values are always real and non-negative.
They are generally denoted by σ The same way as a square matrix can be digitalized by
a similarity transformation with the modal matrix, similarly a non-square matrix can be
digitalized by a method called singular value decomposition. Let A be an m × n matrix
with rank r. Then it can be decomposed into singular values as,
A = U∑
V ∗
where, U is an m×m unitary matrix,V is an n× n unitary matrix and
∑=
σ1 0 .... 0
0 : ... 0
: : σr :
0 0 ... 0
is an m× n matrix.
B.1.2 Norms
Norms can be viewed as the measure of the size of a matrix or a vector. A generalized
p− norm of a vector x ∈ Cn is defined as,
‖x‖p
∆=
(n∑
i=1
|xi|p)1/p
, for 1 ≤ p ≤ ∞
Chapter B. LMI Approach to H∞ Control 148
In particular, 1-norm, 2-norm and ∞-norm are most commonly used. They are denoted
and defined in the following way:
‖x‖1
∆=
(n∑
i=1
|xi|)
‖x‖2
∆=
(n∑
i=1
|xi|2)1/2
‖x‖∞∆= max
i|xi|
The following are some properties of any of the vector norms:
• ‖x‖ ≥ 0
• ‖x‖ = 0 if and only if x = 0
• ‖αx‖ = |α| ‖x‖ for any scalar α
• ‖x + y‖ ≤ ‖x‖+ ‖y‖
Matrix norms are the extension of the concept of length in three-dimensional space to
higher dimensional hyperspace. On the basis of the vector norm, matrix norms or operator
norms of matrices are defined as,
‖A‖p
∆= sup
x 6=0
‖Ax‖p
‖x‖p
where A ∈ Cm×n
In particular, operator 2-norm can be computed as, ‖A‖2 = σ (A) Apart form the prop-
erties of the vector norms, any of the matrix norms obey the following relations:
• ρ (A) ≤ ‖A‖ where ρ denotes spectral radius
• ‖AB‖ ≤ ‖A‖ . ‖B‖
• ‖UAV ‖ = ‖A‖ for any appropriately dimensioned unitary matrices U and V
Chapter B. LMI Approach to H∞ Control 149
B.1.3 Vector Spaces
It is a set whose elements are real or complex valued vectors and where vector addition
and multiplication of vectors by a scalar are defined i.e. if v1, v2 ∈ Cn where Cn denotes
an n-dimensional vector space then,
(α1v1 + α2v2) ∈ Cn, for any α1 and α2 and
βv1, βv2 ∈ Cn for any β
B.1.4 Basis Vector
It is a set of linearly independent vectors that span the vector space (i.e. linear combina-
tions of the basis vectors can produce any vector that belongs to that vector space). A
set of basis vectors u1, u2....un is said to be orthogonal if the inner product of any two
of them is zero i.e. 〈ui, uj〉 = 0∀i, j and i 6= j An orthogonal set of unit vectors is called
an orthonormal set (i.e. in addition to 〈ui, uj〉 = 0, 〈ui, ui〉 = 1,∀i)
B.1.5 L2 space
Let f(t) and g(t) be two matrix valued time functions. An L2[or, L2(−∞,∞)] space is
defined as the space of real matrix valued time functions with finite inner products given
by,
〈f(t), g(t)〉 =
∞∫
−∞
tr[fT (t)× g(t)
]dt (B.1)
The norm induced by the inner product is given by,
‖f(t)‖22 =
∞∫
−∞
tr[fT (t)× f(t)
]dt (B.2)
Chapter B. LMI Approach to H∞ Control 150
This is how the 2-norm of signals is defined. If this norm exists then it is said that,f(t) ∈L2(−∞,∞). For causal signals (i.e.f(t) = 0 for t < 0) if the 2-norm exists then, f(t) ∈L2 [0,∞).
B.1.6 L∞ space
A matrix valued time function f(t) is said to belong to L∞ Space if it has a finite ∞-norm
defined as ‖f(t)‖∞ = supt
σ [f(t)]. If f(t) is a scalar function then the ∞-norm is defined
as,‖f(t)‖∞ = supt|f(t)|
B.1.7 H2 space
H2 space is nothing but the frequency-domain version of the space. In other words, it is
the space of complex matrix functions having bounded 2-norms or H2-norms i.e.
‖F (jω)‖22 =
1
2π
∞∫
−∞
tr [F ∗(jω).F (jω)]dω < ∞ (B.3)
It may be noted that the above 2-norm will exist for strictly proper functions with no
pole on the imaginary axis. The real rational subspace of H2 which consists of all strictly
proper real rational stable transfer matrices is denoted by RH2
B.1.8 H∞ space
Simply stating H∞ space is the space of frequency dependent matrix functions having
bounded H∞-norm,which is defined as
‖F‖∞ = supω
σ [F (jω)] (B.4)
Chapter B. LMI Approach to H∞ Control 151
For scalar functions, the H∞-norm is nothing but the peak value of the Bode magnitude
plot. H∞-norm exists for proper transfer functions with no pole on the imaginary axis.
RH∞ is a subspace of H∞ space consisting of real rational proper and stable transfer
matrices.
B.1.9 Packed Matrix Notation
A transfer matrix can be found from the A, B, C, D matrices as, G(s) = C(sI−A)−1B+D
In packed matrix notation, G(s) is represented in terms of A,B,C, D matrices as,
G =
[A B
C D
]
It is to be kept in mind that G written in the above form, is not a matrix in the
original sense but only a notation, which gives some computational advantages. The
following three formulae regarding the packed matrix notation of different interconnection
of systems are very useful and will be used frequently in later sections.
Figure B.1: System interconnections: (a) Series connection, (b) Inversion, (c) Parallelconnection
Chapter B. LMI Approach to H∞ Control 152
Series Connection Let there be two systems connected in series as shown in Fig. If,
G1 =
[A1 B1
C1 D1
]and G2 =
[A2 B2
C2 D2
]then G1G2 =
A1 0 B1
B2C1 A2 B2D1
D2C2 C2 D2D1
Inversion If ,G =
[A B
C D
]then G−1 =
[A−BD−1C BD−1
−D−1C D−1
],provided D is non-
singular.
Parallel Connection If G1 =
[A1 B1
C1 D1
]and G2 =
[A2 B2
C2 D2
]then it can be shown
that G1 + G2 =
A1 0 B1
0 A2 B2
C1 C2 D1 + D2
B.1.10 Robust Stability
Robustness of the stability in the face of model errors will be treated briefly[41],[20].The
whole concept is based on the so called small gain theorem which trivially applies to the
situation sketched in fig B.2.
Figure B.2: Closed loop with loop transfer H
The stable stable transfer H represents the total looptransfer in a closed loop. If we
require that the modulus (amplitude) of H is less than 1 for all frequencies it is clear
from fig B.6 that the polar curve cannot encompass the point −1 and thus we know
Chapter B. LMI Approach to H∞ Control 153
Figure B.3: Small gain stability in Nyquist space
from nyquist criterion that the loop will always constitute a stable system.So stability is
guaranteed as long as
‖H‖∞def= sup
ω|H(jω)| < 1 (B.5)
sup stands for supremum which effectively indicates the maximum.(only in the case that
the supremum is approached at within any small distance but never really reached it is
not allowed to speak of a maximum.) Notice that no information concerning phase angle
has been used which is typically H∞. In the above formula we get the first taste of H∞by simultaneous definition of the infinity norm which has been discussed already.For the
MIMO system the small gain condition is given by
‖H‖∞def= sup
ωσ (H(jω)) < 1 (B.6)
Where σ denotes the maximum singular value (always real) of the transfer H (for the
ω under consideration).All together, these conditions may seem somewhat exaggerated,
because transfers, less than one, are not so common. The actual application is therefore
Chapter B. LMI Approach to H∞ Control 154
somewhat ”nested” and very depictively indicated in the literature as ”the baby small gain
Figure B.4: Baby small gain theorem for additive model error
theorem” illustrated in fig . In the upper block scheme all relevant elements of fig B.2 have
been displayed in case we have to deal with an additive model error ∆P . We now consider
the ”baby” loop as indicated containing ∆P explicitly. The lower transfer between the
Figure B.5: Control sensitivity guards stability robustness for additive model error
output and the input of ∆p, as once again illustrated in fig , can be evaluated and happens
to be equal to the control sensitivity R as shown in the lower block scheme. (Actually we
get a minus sign that can be joined to ∆P .Because we only consider absolute values in the
small gain theorem, this minus sign is irrelevant: it just causes a phase shift of 180 which
leaves the conditions unaltered.) Now it is easy to apply small gain theorem to the total
loop transfer H = R∆P . The infinity norm will appear to be an induced operator norm
in the mapping between identical signal spaces L and such it follows Schwartz inequality
Chapter B. LMI Approach to H∞ Control 155
so that we may write:
‖R∆P‖∞ ≤ ‖R‖∞ ‖∆P‖∞ (B.7)
Ergo,if we can guarantee that:
‖R‖∞ ≤ 1
α(B.8)
a sufficient condition for stability is :
‖R‖∞ ≤ α (B.9)
If all we require from ∆P is stated in equation B.7 then it is easy to prove that the
condition on R is also necessary condition. till this is rather crude condition but it can
be defined by weighting over the frequency axis. Once again from fig we recognize that
the robustness stability constraint effectively limits the feedback from the point, where
both the disturbance and the output of the model error block ∆P enter, and the point
of the plant such that the loop transfer is less than one. The smaller error bound 1α
the
greater the feedback α can be and vice versa! We so analyzed the effect of additive model
error ∆P . Similarly we can study the effect of multiplicative error which is very easy if
we take:
Ptrue = P + ∆P = (I + ∆) P (B.10)
where obviously ∆ is the bounded multiplicative model error. ( Together with P it
evidently constitutes the additive model error ∆P .) In similar blockschemes we now get
Figure B.6: Baby small gain theorem for multiplicative model error
Chapter B. LMI Approach to H∞ Control 156
figures . The ”baby” -loop now contains ∆ explicitly and we notice that transfer P is
somewhat ”displaced” out of the additive perturbation block. The result is that ∆ sees
Figure B.7: Complementary sensitivity guards stability robustness for multiplicativemodel error
itself fed back by (minus) the complementary sensitivity T . (The P has, so to speak ,
been taken out of ∆P and adjoined to R yielding T .) If we require that:
‖∆‖∞ ≤ 1
β(B.11)
the robust stability follows from:
‖T∆‖∞ ≤ ‖T‖∞ ‖∆‖∞ ≤ 1 (B.12)
yielding as final condition:
‖T‖∞ ≤ β (B.13)
Again proper weighting may refine the condition.
Chapter B. LMI Approach to H∞ Control 157
B.2 A Linear Matrix Inequality Approach to H∞ Con-
trol
The continuous and discrete-time H∞ control problems are solved via elementary ma-
nipulation on linear matrix inequalities (LMI)[26]. Two interesting new features emerge
through this approach: solvability conditions valid for both regular and singular problems,
and an LMI-based parametrization of all H∞-suboptimal controllers, including reduced-
order controllers.The solvability conditions involve Riccati inequalities rather than the
usual indefinite Riccati equations. Alternatively, these conditions can be expressed as a
system of three LMI’s. Efficient convex optimization techniques are available to solve this
system. Moreover, its solutions parameterize the set of H∞ controllers and bear impor-
tant connections with the controller order and the closed loop Lyapunov functions. In
this thesis after some brief introduction on LMI, the H∞ synthesis has been described in
the view of LMI.
B.2.1 Brief theory
The history of LMIs in the analysis of dynamical systems goes back more than 100 years.
The story begins in about 1890, when Lyapunov published his seminal work introducing
what we now call Lyapunov theory. It has been reported in his seminal work that the
differential equation,d
dxx(t) = Ax(t) (B.14)
is stable (i.e., all trajectories converge to zero) if and only if there exists a positive definite
matrix P such that
AT P + PA < 0 (B.15)
The requirement ,P > 0, AT P + PA < 0 is what we now call a Lyapunov inequality on
P , which is a special form of an LMI. Lyapunov also showed that this first LMI could
be explicitly solved. Indeed, we can pick any Q = QT > 0 and then solve the linear
Chapter B. LMI Approach to H∞ Control 158
equation AT P + PA = −Q for the matrix P , which is guaranteed to be positive-definite
if the system B.14 is stable. In summary, the first LMI used to analyze stability of a
dynamical system was the Lyapunov inequality B.15, which can be solved analytically
(by solving a set of linear equations).The important role of LMIs in control theory was
already recognized in the early 1960’s, especially by Yakubovich . The Positive-Real
lemma and extensions were intensively studied in the latter half of the 1960s. By 1970, it
was known that the LMI appearing in the Positive-Real lemma could be solved not only
by graphical means, but also by solving a certain Algebraic Riccati Equation (ARE). In
1971, a paper on quadratic optimal control by J. C. Willems led to the LMI
[AT P + PA + Q PB + CT
BT P + C R
]≥ 0 (B.16)
and pointed out that it can be solved by studying the symmetric solutions of the ARE
AT P + PA− (PB + CT )R−1(BT P + C) + Q = 0 (B.17)
B.2.2 Advantages of LMI
Linear matrix inequalities (LMIs) and LMI techniques have emerged as powerful tools
in areas ranging from control engineering to system identification and structural design.
Three factors make LMI techniques appealing:
• A variety of design specifications and constraints can be expressed as LMIs.
• Once formulated in terms of LMIs, a problem can be solved exactly by efficient
convex optimization algorithms.
• While most problems with multiple constraints or objectives lack analytical solutions
in terms of matrix equations, they often remain tractable in the LMI framework.
This makes LMI based design a valuable alternative to classical analytical method.
Chapter B. LMI Approach to H∞ Control 159
• iv. If the system is strictly proper, the solution of H∞ control problem is not possible
using DGKF method but with LMI technique, it is possible to solve sub-optimal
H∞ control using Bounded Real Lemma
B.2.3 Basic Idea about LMI
A linear matrix inequality (LMI) has the form
F (η) = F0 +m∑
i=1
ηiFi > 0 (B.18)
Where η ∈ <m the variable and the symmetric matrices are Fi = F Ti ∈ <n×n, i =
0, 1..........,m given. The inequality symbol in B.18 means that F (η) is positive definite,
i.e.,uT F (η)u > 0 for all nonzero u ∈ <n. Of course, the LMI B.18 is equivalent to a set of
n polynomial inequalities in η , i.e., the leading principal minors of F (η) must be positive.
We will also encounter non-strict LMIs, which have the form
F (η) ≥ 0 (B.19)
The LMI B.18 is a convex constraint on η, i.e., the set η|F (η) > 0 is convex. Al-
though the LMI B.18 may seem to have a specialized form, it can represent a wide
variety of convex constraints on η. In particular, linear inequalities, quadratic inequal-
ities, matrix norm inequalities, and constraints that arise in control theory, such as
Lyapunov and convex quadratic matrix inequalities, can all be cast in the form of an
LMI. Multiple LMIs F (1)(η) > 0, ......., F (η)(p) > 0 can be expressed as the single LMI
diag(F (1)(η), ....., F (p)(η)) > 0. Therefore we will make no distinction between a set of
LMIs and a single LMI, i.e., ”the LMI F (1)(η) > 0, ......., F (η)(p) > 0” will mean ”the LMI
diag(F (1)(η), ....., F (p)(η)) > 0”. When the matrices Fiare diagonal, the LMI F (η) > 0is
just a set of linear inequalities. Nonlinear (convex) inequalities are converted to LMI form
Chapter B. LMI Approach to H∞ Control 160
using Schur complements. The basic idea is as follows: The LMI
[Q(η) S(η)
S(η)T R(η)
]> 0 (B.20)
Where Q(η) = Q(η)T , R(η) = R(η)T and depend affinely on ’η’, is equivalent to, R(η) > 0,
Q(η)− S(η)R(η)−1S(η)T > 0
B.2.4 Matrices as variable
We will often encounter problems in which the variables are matrices, e.g., the Lyapunov
inequality AT P + PA < 0. Where A ∈ <n×n is given and P = P T is the variable. In this
case we will not write out the LMI explicitly in the form F (η) > 0, but instead make clear
which matrices are the variables. The phrase the LMI AT P + PA < 0 in P means that
the matrix P is a variable. As another related example, consider the quadratic matrix
inequality
AT P + PA + PBR−1BT P + Q < 0 (B.21)
Where A,B, Q = QT , R = RT > 0 are given matrices of appropriate sizes, and P = P T
is the variable. Note that this is a quadratic matrix inequality of the variable P. It can
be expressed as the linear matrix inequality using the Schur complement as,
[−AT P − PA−Q PB
BT P R
]> 0
Chapter B. LMI Approach to H∞ Control 161
B.2.5 Lyapunov’s Inequality
We have already mentioned the Linear Matrix Inequality Problem (LMIP) associated
with Lyapunov’s inequality, i.e.
P > 0 , AT P + PA < 0
Where P is a variable and A ∈ <n×n is given. It can be shown that this LMI is feasible
if and only if the matrix A is stable, i.e., all trajectories of x = Ax converge to zero as
t →∞, or equivalently, all eigen values of P must have negative real part. To solve this
LMIP, we pick any Q > 0 and solve the Lyapunov equation AT P + PA = −Q, which is
nothing but a set of n(n+1)2
linear equations for the n(n+1)2
scalar variables in P . This set
of linear equations will be solvable and result in P > 0 if and only if the LMI is feasible.
In fact this procedure not only finds a solution when the LMI is feasible, it parameterizes
all solutions as Q varies over the positive-definite cone.
B.3 Stabilizing Controllers
A necessary feature of any feedback system is that it be stable in some appropriate sense.
In this chapter we introduce the feedback arrangement we will be studying for the rest of
the course. Once introduced, our main objective is to precisely define feedback stability
and then to parametrize all controllers that stabilize the feedback system. The general
feedback setup we are concerned with is shown above. As depicted the so-called closed-
loop system has one external input and one output, given by w and z respectively. The
signal or function w captures the effects of the environment on the feedback system; for
instance noise, disturbances and commands. The signal z contains all characteristics
of the feedback system that are to be controlled. The maps G and K represent linear
subsystems where G is a given ”plant” which is fixed, and K is the controller or control
law whose aim is to ensure that the mapping from w to z has the desired characteristics.
To accomplish this task the control law utilizes signal y, and chooses an action u which
Chapter B. LMI Approach to H∞ Control 162
Figure B.8: General feedback arrangement
directly affects the behavior of G.
Here G and K are state space systems, with G evolving according to
x(t) = Ax(t) +[
B1 B2
] [w(t)
u(t)
],
[z(t)
y(t)
]=
[C1
C2
]x(t) +
[D11 D12
D21 D22
][w(t)
u(t)
],
(B.22)
and K being described by
xK(t) = AKxK(t) + BKy(t)
u(t) = CKxK(t) + DKy(t)(B.23)
Throughout this section we have the standing assumption that the matrix triples (Ak, Bk, Ck)
and are both stabilizable and detectable
As shown in the figure, G is naturally partitioned with respect to its two inputs and two
Chapter B. LMI Approach to H∞ Control 163
outputs. We therefore partition the transfer function of G as
G(s) =
A B1 B2
C1 D11 D12
C2 D21 D22
=
[G11(s) G12(s)
G21(s) G22(s)
](B.24)
so that we can later refer to these constituent transfer functions.
At first we must determine under what conditions this interconnection of components
makes sense. That is, we need to know when these equations have a solution for an
arbitrary input w.
The system of Figure B.8 is well-posed if unique solutions exist x(t), xK(t), y(t) and u(t),
for all input functions w(t), and all initial conditions x(0), xK(0).
Proposition 2 The connection of G and K in Figure B.8 is well-posed, if and only if,
I −D22DK is nonsingular.
Proof : The proof of this result amounts to simply writing out the system state
equations. So we have
x(t) = Ax(t) + B1w(t) + B2u(t)
xK(t) = AKxK(t) + BKy(t),(B.25)
and
[I −DK
−D22 I
][u(t)
y(t)
]=
[0 CK
C2 0
][x(t)
xK(t)
]+
[0
D21
]w(t) (B.26)
Now it is easily seen that the left hand side matrix is invertible if and only if I −D22DK
is nonsingular. If this holds, clearly one can substitute u, y into B.25 and find a unique
solution to the state equations. Conversely if this does not hold, from B.26 we can find
a linear combination of x(t), xK(t), and w(t) which must be zero, which means that,
x(0), xK(0), w(0) cannot be chosen arbitrarily.
Chapter B. LMI Approach to H∞ Control 164
We have the following result which is frequently used.
Corollary 1 If either D22 = 0 or Dk = 0, then the interconnection in Figure B.8 is
well-posed.
We are now ready to talk about stability. From now on we tacitly assume that our
feedback system is well-posed.
B.3.1 System Stability
In this section we introduce the notion of internal stability, and discuss its relation to the
boundedness of input-output maps [91].
Definition 7 The system in Figure B.8 is internally stable if for every initial condition
x(0) of G, and xk(0) of K, the limits
x(t), xK(t)t→∞→ 0
hold
When w = 0
The following is an immediate test for internal stability.
Proposition 3 Suppose that the system of Figure B.8 is well-posed. Then the system is
internally stable if and only if
A =
[A 0
0 AK
]+
[B2 0
0 BK
][I −DK
−D22 I
]−1 [0 CK
C2 0
](B.27)
Chapter B. LMI Approach to H∞ Control 165
is Hurwitz
Proof : This is easily seen by noting that is the A-matrix of the closed-loop; this follows
from B.25 and B.26.
As defined, internal stability refers to the autonomous system dynamics in the absence of
an input w; in this regard it coincides with the standard notion of asymptotic stability of
dynamical systems. However it has immediate implications on the input-output properties
of the system.
In particular, the transfer function from w to z, denoted T (s), will have as poles a subset
of the eigenvalues of A; for example, when Dk = 0 we have
T (s) =[
C1 D12CK
](Is− A)−1
[B1
BKD21
]+ D11
If A is Hurwitz, this function has all its poles in the left half plane of C. An important
consequence is that w 7→ z defines a bounded operator on L2 [0,∞); this is termed input-
output stability.
The question immediately arises as to whether the two notions are interchangeable, i.e.
whether the boundedness of w 7→ z implies internal stability; clearly, the answer is nega-
tive: an extreme example would be to have C1, D11, D12 be all zero which gives T (s) = 0
but clearly says nothing about A. In other words, the internal dynamics need not be
reflected in the external map.
There is, however, a way to characterize internal stability in terms of the boundedness of
an input-output operator, by considering the map between injected interconnection noise
in the feedback loop, to the interconnection variables. The relevant diagram is given in
Figure B.9, where the controller K has the same description as in Figure B.8. The system
G22 is the lower block of G, described by the state space equations
x22(t) = Ax22(t) + B2v1(t)
v2(t) = C2x22(t) + D22v1(t)
Chapter B. LMI Approach to H∞ Control 166
Figure B.9: Input-output stability
where (C2, A, B2, D22) are the same matrices as in the state space description of G. We
have also included the external inputs d1 and d2 at the interconnection between G22 and
K.
As with our more general system, we say that this new system is well-posed if there
exist unique solutions for x22(t), xK(t), v1 and v2 for all inputs d1(t) and d2(t) and initial
conditions x22(0), xK(0). We say it is internally stable if it is well posed and for di = 0
x(t), xK(t)t→∞→ 0 holds for every initial condition.
It is an easy exercise to see that the system is well-posed, if and only if, I − D22DK
is nonsingular; this is the same well-posedness condition we have for Figure B.8. Also
noticing that all the states in the description of G are included in the equations for G22,
it follows immediately that internal stability of one is equivalent to internal stability of
the other.
Lemma 1 Given a controller K, Figure B.8 is internally stable, if and only if, Figure
B.9 is internally stable.
The next result shows that with this new set of inputs, internal stability can be charac-
terized by the boundedness of an input-output map.
Lemma 2 Suppose that (C2, A, B2) is stabilizable and detectable. Then Figure B.9 is
Chapter B. LMI Approach to H∞ Control 167
internally stable if and only if the transfer function of
[d1
d2
]7→
[v1
v2
]
has no poles in the closed right half plane of C.
Proof : We begin by finding an expression for the transfer function. For convenience
denote
D =
[I −DK
−D22 I
],
then routine calculations lead to the following relationship:
[v1(s)
v2(s)
]=
D−1
[0 CK
C2 0
](Is− A
)−1
[B2 0
0 BK
]D−1 + D−1 +
[0 0
0 −I
][d(s)
d2(s)
]
where is the closed loop matrix from B.27. Therefore the ”only if” direction follows
immediately, since the poles of this transfer function are a subset of the eigenvalues of A
, which is by assumption Hurwitz ; see Proposition 3 and Lemma 1.
To prove ”if”: assume that the transfer function has no poles in C+, therefore the same
is true of [0 CK
C2 0
]
︸ ︷︷ ︸C
(Is− A
)−1
[B2 0
0 BK
]
︸ ︷︷ ︸B
We need to show that A is Hurwitz ; it is therefore sufficient to show that(C, A, B
)is a
stabilizable and detectable realization. Let
F =
[F 0
0 FK
]− D−1
[0 CK
C2 0
]
Chapter B. LMI Approach to H∞ Control 168
Where F and Fk are chosen so that A + B2F and AK + BKFK are both Hurwitz. It is
routine to show that
A + BF =
[A + B2F 0
0 AK + BKFK
]
and thus(A, B
)is stabilizable.
A formally similar argument shows that(C, A
)is detectable.
B.3.2 Stabilization
In the previous section we have discussed the analysis of stability of a given feedback
configuration; we now turn to the question of design of a stabilizing controller. The
following result explains when this can be achieved.
Proposition 4 A necessary and sufficient condition for the existence of an internally
stabilizing K for Figure B.8, is that (C2, A, B2) is stabilizable and detectable. In that
case, one such controller is given by
K(s) =
A + B2F + LC2 + LD22F −L
F 0
Where F and L are matrices such that A + B2F and A + LC2 are Hurwitz.
Proof : If the stabilizability or detectability of (C2, A, B2) is violated, we can choose an
initial condition which excites the unstable hidden mode. It is not difficult to show that
the state will diverge to infinity regardless of the controller. Details are left as an exercise.
Consequently no internally stabilizing K exists, which proves necessity. For the sufficiency
side, it is enough to verify that the given controller is indeed internally stabilizing. Start
Chapter B. LMI Approach to H∞ Control 169
by noting that Dk = 0 and so the configuration is well-posed. Now substitute the state
space for the controller into the expression for A given in Proposition 3.
A =
[A B2F
−LC2 A + B2F + LC2
]
Let
T =
[I 0
I I
]
and notice
T−1AT =
[A + B2F B2F
0 A + LC2
]
Since the eigenvalues of A are therefore given by those of A + B2F and A + LC2 we see
is A Hurwitz.
B.4 H∞ Synthesis
Now we consider optimal synthesis with respect to the H∞ norm introduced already[21].
Again we are concerned with the feedback arrangement of figure where we have two state
space systems G and K, each having their familiar role. We will pursue the answer to the
following question: does there exist a state space controller K such that
• The closed loop system is internally stable;
• The closed loop performance satisfies
∥∥∥∥S
(∧G,
∧K
)∥∥∥∥∞
< 1
Chapter B. LMI Approach to H∞ Control 170
Thus we only plan to consider the problem of making the closed loop contractive in the
sense of H∞ . It is clear, however, that determining whether there exists a stabilizing
controller so that
∥∥∥∥S
(∧G,
∧K
)∥∥∥∥∞
< γ, for some constant γ , can be achieved by rescaling
the γ dependent problem to arrive at the contractive version given above. Furthermore,
by searching over γ , our approach will allow us to get as close to the minimal H∞ norm as
we desire, but in contrast to our work on H2 optimal control, we will not seek a controller
that exactly optimizes the H∞ norm. There are many approaches for solving the H∞
control problem. Probably the most celebrated solution is in terms of Riccati equations.
Here we will present a solution based entirely on linear matrix inequalities, which has the
main advantage that it can be obtained with relatively straightforward matrix tools, and
without any restrictions on the problem data. In fact Riccati equations and LMIs are
intimately related, an issue we will explain when proving the Kalman-Yakubovich-Popov
lemma concerning the analysis of the H∞ norm of a system, which will be key to the
subsequent synthesis solution.
Before getting into the details of the problem, we make a few comments about the
motivation for this optimization.
As we know already that the H∞ norm is the L2-induced norm of a causal, stable,
linear-time invariant system. More precisely, given a causal linear time-invariant operator
G : L2(−∞,∞) → L2(−∞,∞), the corresponding operator in the isomorphic space∧L2
(jR) is a multiplication operator M∧G
for a certain∧G (s) ∈ H∞, and
‖G‖L2→L2= ‖MG‖L2→L2
=∥∥∥G
∥∥∥∞
The motivation for minimizing such an induced norm lies in the philosophy making
error signal small (z), we are minimizing the maximum ”gain” of the system in the energy
or L2 sense. Equivalently, the excitation w is considered to be an arbitrary L2 signal and
we wish to minimize its worst-case effect on the energy of z.
Chapter B. LMI Approach to H∞ Control 171
B.4.1 Two important matrix inequalities
The entire synthesis approach of the chapter revolves around the two technical results
presented here. The first of these is a result purely about matrices; the second is an
important systems theory result and is frequently called the Kalman-Yacubovich-Popov
lemma, or KYP lemma for short. We begin by stating the following which the reader can
prove as an exercise.
Lemma 3 Suppose P and Q are matrices satisfying ker P = 0 and ker Q = 0. Then for
every matrix Y there exists a solution J to
P ∗JQ = Y
The above lemma is used to prove the next one which is one of the two major technical
results of this section
Lemma 4 Suppose,
1. P , Q and H are matrices and that H is symmetric;
2. The matrices WP and WQ are full rank matrices satisfying ImWP = kerP and
ImWQ = kerQ then there exists a matrix J such that
H + P ∗J∗Q + Q∗JP < 0 (B.28)
if and only if, the inequalities
W ∗P HWP < 0 and W ∗
QHWQ < 0
Chapter B. LMI Approach to H∞ Control 172
both hold.
Observe that when the kernels of P and Q are not both nonzero the result does not apply
as stated. However it is readily seen from Lemma 1, that if both of the kernels are zero
then there is always a solution J . if for example only kerP = 0 then W ∗QHWQ < 0
is a necessary and sufficient condition for a solution to lemma 1 to exist, as follows by
simplified version of the following proof.
Proof : We will show the equivalence of the conditions directly by construction. To
begin define V1 to be a matrix such that
ImV1 = ker P ∩ ker Q,
and V2 and V3 such that
Im [V1V2] = ker P and Im [V1V3] = ker Q
without loss of generality we assume that V1, V2 and V3 have full column rank and define
V4 so that
V =[
V1 V2 V3 V4
]
is square and nonsingular. Therefore the LMI B.21 above holds, if and only if
V ∗HV + V ∗P ∗J∗QV + V ∗Q∗JPV < 0 (B.29)
does. Now PV and QV are simply the matrices P and Q on the domain basis defined by
V ; therefore they have the form
PV =[
0 0 P1 P2
]and QV =
[0 Q1 0 Q2
];
Chapter B. LMI Approach to H∞ Control 173
we also define the block components
V ∗HV =:
H11 H12 H13 H14
H∗12 H22 H23 H24
H∗13 H∗
23 H33 H34
H∗14 H∗
24 H∗34 H44
Further define the variable Y by
Y =
[Y11 Y12
Y21 Y22
]=
[P ∗
1
P ∗2
]= J∗
[Q1 Q2
]
from their definitions ker[
P1 P2
]= 0 and ker
[Q1 Q2
]= 0, and so by Lemma
1 we see that Y is freely assignable by choosing an appropriate matrix J . Writing out
inequality B.29 using the above definitions we get
H11 H12 H13 H14
H∗12 H22 H23 + Y ∗
11 H24 + Y ∗21
H∗13 H∗
23 + Y11 H33 H34 + Y12
H∗14 H∗
24 + Y21 H∗34 + Y ∗
12 H44 + Y22 + Y ∗22
< 0
Apply the Schur complement formula to the upper 3 × 3 block, and we see the above
holds, if and only if, the two following inequalities are met.
−H :=
H11 H12 H13
H∗12 H22 H23 + Y ∗
11
H∗13 H∗
23 + Y11 H33
< 0
Chapter B. LMI Approach to H∞ Control 174
and H44 + Y22 + Y ∗22 −
H14
H24 + Y ∗21
H34 + Y12
∗
H−1
H14
H24 + Y ∗21
H34 + Y12
< 0
as already noted above Y is freely assignable and so we see that provided the first in-
equality can be achieved by choosing Y11, the second can always be met by appropriate
choice of Y12, Y21 and Y22. That is the above two inequalities can be achieved, if and only
if,−H < 0 holds for some Y11. Now applying a Schur complement on
−H with respect to
H11, we obtain
H11 0 0
0 H22 − Y ∗12Y
−111 H12 Y ∗
11 + X∗
0 Y11 + X H33 −H∗13H
−111 H13
< 0,
where X = H∗23 −H∗
13H−111 H12. Now since Y11 is freely assignable we see readily that the
last condition can be satisfied, if and only if, the diagonal entries of the lift hand matrix
are all negative definite. Using the Schur complement result twice these three conditions
can be converted to the equivalent conditions
[H11 H12
H∗12 H22
]< 0 and
[H11 H13
H∗13 H33
]< 0
by the choice of our basis we see that these hold, if and only if, W ∗P HWP < 0 and W ∗
QHWQ <
0 are both met. Having proved this matrix result we move on to our second result, the
KYP lemma.
B.4.2 The KYP Lemma
There are many versions of this result, which establishes the equivalence between a fre-
quency domain inequality and a state-space condition in terms of either a Riccati equation
Chapter B. LMI Approach to H∞ Control 175
or an LMI. The version given below turns an H∞ norm condition into an LMI. Being able
to do this is very helpful for attaining our goal of controller synthesis, however it is equally
important simply as a finite dimensional analysis test for transfer functions.
Lemma 5 Suppose∧
M (s) = C (Is− A)−1 B + D . Then the following are equivalent
conditions.
1. The matrix A is Hurwitz and ∥∥∥∥∧
M
∥∥∥∥∞
< 1;
2. ii. There exists a symmetric positive definite matrix X such that
C∗
D∗
[C D
]+
A∗X + XA XB
B∗X −I
< 0 (B.30)
the condition in (ii) is clearly an LMI and gives us a very convenient way to evaluate the H
norm of a transfer function. In the proof below we see proving that condition (ii) implies
that (i) holds is reasonably straightforward, and involves showing the direct connection
between the above LMI and the state space equations that describe M . proving the
converse is considerably harder; fortunately we will be able to exploit the Riccati equation
techniques. An alternative proof, which employs only matrix arguments is beyond the
scope. Proof :We begin by showing (ii) implies (i). The top left block in B.30 states that
A∗X + XA + C∗C < 0. Since X > 0 we see that A must be Hurwitz. It remains to
show contractiveness which we do by employing a system-theoretic argument based on
the state equations for M . using the strict inequality B.30 choose 1 > ε > 0 such that
[C∗
D∗
] [C D
]+
[A∗ + XA XB
B∗X − (1− ∈) I
]< 0 (B.31)
Chapter B. LMI Approach to H∞ Control 176
holds. Let ω ∈ L2 [0,∞] and realize that in order to show that M is contractive, it is
sufficient to show that ‖z‖2 ≤ (1− ∈) ‖ω‖2 , where z := Mω. The state space equations
relating ω and z are·x (t) = Ax (t) + Bω (t) , x (0) = 0,
z (t) = Cx (t) + Dω (t) .
Now multiplying inequality B.31 on the left by [x∗(t)w∗(t)] and on the right by the adjoint
we have
|z(t)|22 + x∗(t)X(Ax(t) + Bw(t)) + (Ax(t) + Bw(t))∗Xx(t)− (1− ∈) |w(t)|22 ≤ 0
By introducing the storage function V : Rn → R , defined by V (x(t) = x∗(t)Xx(t), we
arrive at the so-called dissipation inequality
V + |z(t)|22 ≤ (1− ∈) |w(t)|22
Integrating on an interval [0, T ], recalling that x(0) = 0, gives
x(T )∗Xx(T ) +
∫ T
0
|z(t)|22 dt ≤ (1− ∈)
∫ T
0
|w(t)|22 dt
Since X > 0, we can suppress the first term above; now taking the limit as T → ∞, we
find that
||z||22 ≤ (1− ∈) ||w||22which completes this direction of the proof. We now tackle the direction (i) implies (ii).
To simplify the expressions we will write the derivation in the special case D = 0, but an
analogous argument applies to the general case. Starting from
M(s) =
[A B
C 0
]
Chapter B. LMI Approach to H∞ Control 177
and from the definition of para-Hermitian conjugate, if M∼(s) :=
M∼(−s∗)∗
we derive
the state-space representation
I − M∼(s)M(s) =
A 0 −B
−C∗C −A∗ 0
0 B∗ I
It is easy to verify that
[I − M∼(s)M(s)
]−1
=
A BB∗ B
−C∗C −A∗ 0
0 B∗ I
(B.32)
Since∥∥∥M
∥∥∥∞
< 1 by hypothesis, we conclude that[I − M∼(s)M(s)
]−1
has no poles on
the imaginary axis. Furthermore we now show, using the PBH test, that the realization
B.32 has no unobservable eigen values that are purely imaginary. Suppose that
jω0I − A −BB∗
C∗C jω0I + A∗
0 B∗
[x1
x2
]= 0
for some vectors x1 and x2. Then we have the following chain of implications
B ∗ x2 = 0 ⇒ (jω0I − A) x1 = 0,
therefore x1 = 0 since A is Hurwitz;
this means (jω0I − A∗) x2 = 0;
which implies x2 = 0 again because A is Hurwitz.
We conclude that B.32 has no un-observable eigenvalues on the imaginary axis; an anal-
ogous argument shows the absence of uncontrollable eigenvalues. This means that the
Chapter B. LMI Approach to H∞ Control 178
matrix
H =
[A BB∗
−C∗C −A∗
]
has no purely imaginary eigenvalues. According to the lemma
Lemma 6 Suppose that H is a Hamiltonian matrix and
• H has no purely imaginary axis eigen value;
• R is either positive or negative semidefinite;
• (A,R) is a stabilizable matrix pair
Then H is in the domain of the Riccati operator
Here notice thatBB∗ ≥ 0 and (A,BB∗) is stabilizable since A is Hurwitz. Hence H is in
the domain of the Riccati operator, and we can define X0 = Ric(H) satisfying
A∗X0 + X0A + C∗C + X0BB∗X0 = 0 (B.33)
and A + BB∗X0 Hurwitz. Also note that B.33 implies A∗X0 + X0A ≤ 0, therefore from
our work on Lyapunov equations we see that
X0 ≥ 0
since A is Hurwitz. To obtain the LMI characterization of (ii) we must slightly strengthen
the previous relationships. For this purpose define X to be the solution of the Lyapunov
equation
(A + BB∗X0)∗X + X(A + BB∗X0) = −I (B.34)
Chapter B. LMI Approach to H∞ Control 179
Since (A + BB∗X0) is Hurwitz we have X > 0. Now let X = X0+ ∈ X > 0, which is
positive definite for all ε > 0. Using B.33 and B.34 we have
A∗X + XA + C∗C + XBB∗X = − ∈ I+ ∈2 XBB∗X
Choose ε > 0 sufficiently small so that this equation is negative definite. Hence we have
found X > 0 satisfying the strict Riccati Inequality
A∗X + XA + C∗C + XBB∗X = 0
Now applying a Schur complement operation, this inequality is equivalent to
[A∗X + XA + C∗C XB
B∗X −I
]< 0
which B.30 for the special case D = 0.
The preceding proof illustrates some of the deepest relationships of linear systems
theory. We have seen that frequency domain inequalities are associated with dissipativity
of storage functions in the time domain, and also the connection between LMIs (linked
to dissipativity) and Riccati equations (which arise in quadratic optimization).
In fact this latter connection extends as well to problems of H∞ synthesis, where both
Riccati equations and LMIs can be used to solve the suboptimal control problem.
Chapter B. LMI Approach to H∞ Control 180
B.4.3 Synthesis
Lets start with the state space realizations that describe the systems G and K [? ]:
G(s) =
A B1 B2
C1 D11 D12
C2 D21 0
, K(s) =
[AK BK
CK DK
]
Notice that we have assumed D22 = 0. Removing this assumption leads to more compli-
cated formulae, but the technique is identical. We make no other assumptions about the
state space systems. The state dimensions of the nominal system and controller will be
important: A ∈ Rn×n, Ak ∈ Rnk×nk
Our first step is to combine these two state space realizations into one which describes
the map from w to z. We obtain
S(G,K) =
[AL BL
CL DL
]=
A + B2DKC2 B2CK B1 + B2DKC21
BKC2 AK BKD21
C1 + D12DKC2 D12CK D11 + D12DKD21
Now define the matrix
J =
[AK BK
CK DK
]
which collects the representation for K into one matrix. We can parameterize the closed-
loop relation in terms of the controller realization as follows. First make the following
Chapter B. LMI Approach to H∞ Control 181
definitions.
A =
[A 0
0 0
], B =
[B1
0
]
C =[
C1 0], C =
[0 I
C2 0
]
B =
[0 B2
I 0
], D12 =
[0 D12
]
D21 =
[0
D21
]
(B.35)
which are entirely in terms of the state space matrices for G. Then we have
AL = A + BJC, BL = B + BJD21
CL = C + D12JC, DL = D11 + D12JD21
(B.36)
The crucial point here is that the parametrization of the closed loop state space matrices
is affine in the controller matrix J .
Now we are looking for a controller K such that the closed loop is contractive and inter-
nally stable. The following form of the KYP lemma will help us.
Corollary 2 Suppose ML(s) = CL(Is−AL)−1BL+DL. Then the following are equivalent
conditions.
1. The matrix AL is Hurwitz and∥∥∥ML
∥∥∥∞
< 1
Chapter B. LMI Approach to H∞ Control 182
2. There exists a symmetric positive definite matrix XL such that
A∗LXL + XLAL XLBL C∗
L
B∗LXL −I D∗
L
CL DL −I
< 0
This result is readily proved from Lemma 5 by applying the Schur complement formula.
Notice that the matrix inequality in (b) is affine in XL and J individually, but it is not
jointly affine in both variables. The main task now is to obtain a characterization where
we do have a convex problem.
Now define the matricesPXL
=[
B∗X 0 D∗12
]
Q =[
C D21 0]
And further
HXL=
A∗XL + XLA XLB C∗
B∗XL −I D∗11
C D11 −I
It follows that the inequality in (b) above is exactly
HXL+ Q∗J∗PXL
+ P ∗XL
JQ < 0
Lemma 7 Given the above definitions there exists a controller synthesis K if and only if
there exists a symmetric matrix XL > 0 such that
W ∗PXL
HXLWPXL
< 0 and W ∗QHXL
WQ < 0
Chapter B. LMI Approach to H∞ Control 183
Where are as defined in Lemma 4
Proof : From the discussion above we see that a controller K exists if and only if there
exists XL > 0 satisfying
HXL+ Q∗J∗PXL
+ P ∗XL
JQ < 0
Now invoke Lemma 4.
This lemma says that a controller exists if and only if the two matrix inequalities can be
satisfied. Each of the inequalities is given in terms of the state space matrices of G and
the variable XL. However we must realize that since XL appears in both HXLand WPXL
,
that these are not LMI conditions. Converting to an LMI formulation is our next goal,
and will require a number of steps. Given a matrix XL > 0 define the related matrix
TXL=
AX−1L + X−1
L A∗ B X−1L C∗
B∗ −I D∗11
CX−1L D11 −I
(B.37)
And the matrix
P =[
B∗ 0 D∗12
](B.38)
Which only depends on the state space realization of G.the next lemma converts one of
the two matrix inequalities of the lemma, involving HXL, to one in terms of TXL
.
Lemma 8 Suppose XL > 0. Then
W ∗PXL
HXLWPXL
< 0 , if and only if, W ∗P TXL
WP < 0
Proof :Start by observing that
PXL= PS where,
S =
XL 0 0
0 I 0
0 0 I
Chapter B. LMI Approach to H∞ Control 184
Therefore we have ker PXL= S−1 ker P
Then using the definitions of WPXLand WP we can set
WPXL= S−1WP
Finally we have that W ∗PXL
HXLWPXL
< 0 if and only if
W ∗P (S−1)∗HXL
S−1WP < 0
And it is routine to verify (S−1)∗HXLS−1 = TXL
Combining the last two lemmas we see
that there exists a controller of state dimension nk if and only if there exists a symmetric
matrix XL > 0 such that
W ∗P TXL
WP < 0 and W ∗QHXL
WQ < 0 (B.39)
The first of these inequalities is an LMI in the matrix variable XL, where as the second
is an LMI in terms of XL. However the system of both inequalities is not an LMI. Our
intent is to convert these seemingly non-convex conditions into an LMI condition. Recall
that XL is a real and symmetric (n + nK) × (n + nK) matrix; here n and nk are state
dimensions of G and K. Let us now define the matrices X and Y which are submatrices
of XL and X−1L , by
XL =:
[X X2
X∗2 X3
]and X−1
L =:
[Y Y2
Y ∗2 Y3
](B.40)
We now show that the two inequality conditions listed in B.39, only constrain the sub-
matrices X and Y
Lemma 9 Suppose XL is a positive definite (n + nK) × (n + nK) matrix and X and Y
Chapter B. LMI Approach to H∞ Control 185
are n× n matrices defined as in B.40. Then
W ∗P TXL
WP < 0andW ∗QHXL
WQ < 0
If and only if, the following two matrix inequalities are satisfied
1.
NX 0
0 I
∗
A∗X + XA XB1 C∗1
B∗1X −I D∗
11
C1 D11 −I
NX 0
0 I
< 0
2.
NY 0
0 I
∗
AY + Y A∗ Y C1 B1
C1Y −I D11
B∗1 D∗
11 −I
NY 0
0 I
< 0
Where Nx and Ny are full-rank matrices whose images satisfy
ImNX = ker[
C2 D21
]
ImNY = ker[
B∗2 D∗
12
]
Proof : The proof amounts to writing out the definitions and removing redundant con-
straints. Let us show that W ∗P TXL
WP < 0 is equivalent to the LMI in (ii). From the
Chapter B. LMI Approach to H∞ Control 186
definitions of TXL in B.37, and A, B and C in B.35 we get
TXL=
AY + Y A∗ AY2 B1 Y C∗1
Y ∗2 A∗ 0 0 Y ∗
2 C∗1
B∗1 0 −I D∗
11
C1Y C1Y2 D11 −I
Also recalling the definition of P in B.38, and substituting for B and D12 from B.35 yields
P =
[0 I 0 0
B∗2 0 0 D∗
12
]
Thus the kernel of P is the image of
WP =
V1 0
0 0
0 I
V2 0
Where
[V1
V2
]= NY
spans the kernel of [B∗2D
∗12] as defined above. Notice that the second block row of WP is
exactly zero, and therefore the second block-row and block-column of TXL, as explained
above, do not enter into the constraint W ∗P TXL
WP < 0. Namely this inequality is
V1 0
0 I
V2 0
∗
AY + Y A∗ B1 Y C∗1
B∗1 −I D∗
11
C1Y D11 −I
V1 0
0 I
V2 0
< 0
Chapter B. LMI Approach to H∞ Control 187
By applying the permutation
V1 0
0 I
V2 0
=
I 0 0
0 0 I
0 I 0
[NY 0
0 I
]
We arrive at (ii). Using a nearly identical argument, we can readily show that W ∗QHXL
WQ <
0 is equivalent to LMI (i)in the theorem statement.
What we have shown is that a controller synthesis exists if and only if there exists an
(n + nK) × (n + nK) matrix XL that satisfies conditions (i) and (ii) of the last lemma.
These latter two conditions only involve X and Y , which are submatrices of respectively
XL and X−1L . Our next result tell us under what conditions, given arbitrary matrices X
and Y , it is possible to find a positive definite matrix that satisfies B.40.
Lemma 10 Suppose X and Y are symmetric, positive definite matrices in Rn×n; and nk
is a positive integer. Then there exist matrices X2, Y2 ∈ Rn×nK and symmetric matrices
X3, Y3 ∈ RnK×nK , satisfying
[X X2
X∗2 X3
]> 0and
[X X2
X∗2 X3
]−1
=
[Y Y2
Y ∗2 Y3
]
if and only if [X I
I Y
]≥ 0 and rank
[X I
I Y
]≤ n + nK (B.41)
Proof : First we prove that the first two conditions imply the second two. From
[X X2
X∗2 X3
][Y Y2
Y ∗2 Y3
]= I (B.42)
Chapter B. LMI Approach to H∞ Control 188
It is routine to verify that
0 ≤[
I 0
Y Y2
][X X2
X∗2 X3
][I Y
0 Y ∗2
]=
[X I
I Y
]
Also the schur component relationship
[X I
I Y
]=
[I Y −1
0 I
][X − Y −1 0
0 Y
][I 0
Y −1 I
](B.43)
implies that
rank
[X I
I Y
]= n + rank(X − Y −1) = n + rank(XY − I) ≤ n + nK
where the last inequality follows from B.42 : I −XY = X2Y∗2 andX2 ∈ Rn×nK . To prove
”if” we start with the assumption that B.41 holds; therefore B.43 gives
X − Y −1 ≥ 0andrankX − Y −1 ≤ nK
These conditions ensure that there exists a matrix X2 ∈ Rn×nK so that
X − Y −1 = X2X∗2 ≥ 0
From this and the Schur component argument we see that
[X X2
X∗2 I
]> 0
Chapter B. LMI Approach to H∞ Control 189
Also [X X2
X∗2 I
]−1
=
[Y −Y X2
−X∗2Y X∗
2Y X2 + I
]
and so we set X3 = I
The lemma states that a matrix XL in R(n+nK)×(n+nK), satisfying B.40, can be constructed
from X and Y exactly when the LMI and rank conditions in B.41 are satisfied. The rank
condition is not in general an LMI, but notice that
rank
[X I
I Y
]≤ 2n
Therefore we see that if nK ≥ n in the lemma, the rank condition becomes vacuous and
we are left with only the LMI condition. We can now prove the synthesis theorem.
Theorem 1 A synthesis exists to the H∞ problem, if and only if there exist symmetric
matrices X > 0 and Y > 0 such that
1.
NX 0
0 I
∗
A∗X + XA XB1 C∗1
B∗1X −I D∗
11
C1 D11 −I
NX 0
0 I
< 0
2.
NY 0
0 I
∗
AY + Y A∗ Y C1 B1
C1Y −I D11
B∗1 D∗
11 −I
NY 0
0 I
< 0
3.
X I
I Y
≥ 0
Chapter B. LMI Approach to H∞ Control 190
Where NX and NY are full-rank matrices whose images satisfy
ImNX = ker[
C2 D21
]
ImNY = ker[
B∗2 D∗
12
]
Proof : Suppose a controller exists, then by Lemma 9 a controller exists if and only if the
inequalities
W ∗P TXL
WP < 0 and W ∗QHXL
WQ < 0
hold for some symmetric, positive definite matrix XL in R(n+nK)×(n+nK). By Lemma 9
these LMIs being satisfied imply that (1) and (2) are met. Also invoking Lemma 10 we see
that (3) is satisfied. Showing that (1-3) imply the existence of a synthesis is essentially
the reverse process. Since nk ≥ n we have by Lemma 10 that there exists an X2 in
R(n+nK)×(n+nK) which satisfies B.40.
The proof is now completed by XL using and (1-2) together with Lemma 9. This theorem
gives us exact conditions under which a solution exists to our H∞ synthesis problem.
Notice that the conditions are totally independent of the controller state dimensions nk,
provided that nk is not smaller than the system state dimension n. This clearly means
that a synthesis exists if and only if one exists with state dimension nk = n.
Theorem 1 gives us necessary and sufficient conditions for the existence of a synthesis,
however we still need to trace these conditions backwards to explicitly construct such a
controller.
B.4.4 Controller reconstruction and connections
The results of the last section provide us with an explicit way to determine whether a
synthesis exists which solves the H∞ problem. Implicit in our development is a method
to construct controllers when the conditions of Theorem 1 are met. We now outline this
procedure, which simply retraces our steps so far [92].
Chapter B. LMI Approach to H∞ Control 191
Suppose X and Y have been found satisfying Theorem 1 then by Lemma 10 there exists
a matrix satisfying
XL =
[X ?
? ?
]and X−1
L =
[Y ?
? ?
]
From the proof of the lemma we can construct XL by finding a matrix X2 ∈ Rn×nK such
that X − Y −1 = X2X∗2 .Then
XL =
[X X∗
2
X2 I
]
has the properties desired above. Notice that there is some freedom in choosing the
controller state dimension nk. We typically would like to select nk as small as possible.
Immediately we know that need be no larger than n, and can be chosen to be the rank
of X − Y −1 Next by Lemma 4 we know that there exists a solution to
HXL+ Q∗J∗PXL
+ P ∗XL
JQ < 0
and that any such solution J provides the state space realization for a feasible controller
K. The solution of this LMI can be accomplished using standard techniques, and there
is clearly an open set of solutions J .
Finally let us observe that we can extend Theorem 1 to provide conditions for controllers
of state dimension nk less than n. To do this we add the constraint that
rank(X − Y −1) ≤ nK
to the list (1) to (3).Unfortunately this constraint is not convex in general when nk < n,
however it does provide an explicit condition which can be exploited in certain situations
Bibliography
[1] O. Akhrif, F. Okou, L. A. Dessaint, and R. Champagne. Multi-input multi-output
feedback linearization of a synchronous generator. Canadian Conference on Electrical
and Computer Engineering, 2:586–590, 26-29 May 1996. 2.2.1.1
[2] JosE Alvarez-Ramirez. Observers for a class of continuous tank reactors via temper-
ature measurement. Journal of Chemical Engineering Science, 50(9):387–404, May
1995. 1.5.6
[3] E. R. Bachmann, R. B. McGhee, X. Yun, and M. J. Zyda. Rigid body dynamics,
inertial reference frames, and graphics coordinate systems: A resolution of conflicting
conventions and terminology. IEEE Symposium on Computational Intelligence in
Robotics and Automation, 2000. 1.1.3
[4] Mark D. J. Brown and Yuri B. Shtessel. Disturbance cancellation techniques for mimo
smooth sliding mode control. AIAA Guidance, Navigation, and Control Conference
and Exhibit, August 2001. 1.5.3
[5] K. Busawon, M. Farza, and H. Hammouri. A simple observer for a class of nonlinear
systems. Applied Mathematics Letters, 11(3):27–31(5), May 1998. 1.5.6
[6] M. Casdagli, S. Eubank, J.D. Farmer, and J. Gibson. A luenberger like observer for
nonlinear systems. Int. J. Control, 57:536–556, 1993. 2.3.2
[7] B.M. Chen, T.H. Lee, P. Kemao, and V. Venkataramanan. Composite nonlinear
BIBLIOGRAPHY 193
feedback control for linear systems with input saturation: theory and an application.
IEEE Transactions on Automatic Control, 48(3):427– 439, March 2003. 1.5.7
[8] Dong Kyoung Chwa and Jin Young Choi. New parametric affine modeling and control
for skid-to-turn missiles. IEEE Trans. On Control System Technology, 9(2):335–347,
March 2001. 2.2
[9] G. Ciccarella, M. Dalla Mora, and A. Germani. A luenberger-like observer for non-
linear systems. Int. J. Control, 57(3):537–556, 1993. 2.3.2
[10] M. Dalla Mora, A. Germani, and C. Manes. A state observer for nonlinear dynamical
systems. Proc. of the 2nd World Congress of Nonlinear Analysis, Theory, Methods
and Applications, 30(7):4485–4496, December 1997. 1.5.6, 2.3.2
[11] A. Das, R. Das, S. Mukhopadhyay, and A. Patra. Nonlinear autopilot and observer
design for a surface-to-surface, skid-to-turn missile. Second India annual conference,
Proceedings of the IEEE INDICON 2005, pages 304 – 308, 11-13 December 2005.
1.1.3
[12] A. Das, R. Das, S. Mukhopadhyay, and A. Patra. Sliding mode controller along with
feedback linearization for a nonlinear missile model. 1st International Symposium on
Systems and Control in Aerospace and Astronautics, 19-21 January 2006. 1.1.3
[13] A. Das, T. Garai, S. Mukhopadhyay, and A. Patra. Feedback linearization for a
nonlinear skid-to-turn missile model. First India annual conference, Proceedings of
the IEEE INDICON 2004, pages 586–589, 20-22 December 2004. 1.1.3
[14] R.A. Decarlo, S.H. Zak, and G.P. Matthews. Variable structure control of nonlinear
multivatiable systems:a tutorial. Proceedings of the IEEE, 76(3):212–232, March
1988. 4.1
[15] Joachim Deutscher. Asymptotically exact input-output linearization using carleman
linearization. Proc. European Control Conference ECC 2003, 2003. 1.5.6
BIBLIOGRAPHY 194
[16] Emmanuel Devaud, Jean-Philippe Harcaut, and Houria Siguerdidjane. Three-axes
missile autopilot design: From linear to nonlinear control strategies. Journal of
Guidance, Control, and Dynamics, 24(1):64–71, January-February 2001. 1.5.2
[17] F. Deza, D. Bossanne, E. Busvelle, J. P. Gauthier, and D. Rakotopara. Exponential
observers for nonlinear systems. IEEE Trans. Automat. Contr., 38:482–484, March
1993. 1.5.6
[18] Denis Dochain. State and parameter estimation in chemical and biochemical
processes: a tutorial. Journal of Process Control, 13(8):801–818, December 2003.
1.5.6
[19] F.J. Doyle, F. Allgower, and M. Morari. A normal form approach to approximate
input-output linearization for maximum phase nonlinear siso systems. IEEE Trans.
Automatic Control, 41:305–309, February 1996. A.2.2
[20] J.C. Doyle, K. Glover, P.P. Khargonekar, and B.A. Francis. State-space solutions to
standard h2 and h∞ control problems. IEEE Trans. Automatic Control, 34:831–847,
August 1989. B.1.10
[21] G. Dullerud and F. Paganini. h∞ synthesis. Lecture Note Series, 1997. B.4
[22] J. Eker and K. J. Astrom. A nonlinear observer for the inverted pendulum. The 8th
IEEE Conference on Control Application, pages 332 – 337, September 1996. 1.5.6
[23] M. Farza, M. M. Saad, and L. Rossignol. Observer design for a class of mimo nonlinear
systems. Automatica, 40(1):135–143, 2004. 1.5.6
[24] G. Ferreres, V. Fromion, G. Duc, and M.M. Saad. Non conservative robustness
evaluation of a multivariable h∞ missile autopilot. Proceedings of the American
Control Conference, 3:3283–3287, June-July 1994. 1.5.4
[25] G.C. Freeland and T.S. Durrani. Nonlinear state observers for chaotic systems and
their application to communications. IEE Colloquium on Exploiting Chaos in Signal
Processing, 10, 6th June 1994. 1.5.6
BIBLIOGRAPHY 195
[26] P. Gahinet and P. Apkarian. A linear matrix inequality approach to h∞ control.
International Journal of Robust and Nonlinear Control, pages 1–30, 1994. B.2
[27] W. Gao and J.C. Hung. Variable structure control of nonlinear systems:a new ap-
proach. IEEE Trans. Industrial Electronics, 40:45–55, February 1993. 4.2.1.2
[28] G. Garcia, J. Bernussou, J. Daafouz, and D. Arzelier. Robust quadratic stabilization.
Robust-Flight Control. A Design Challenge. GARTEUR, Lecture Notes in Control
and Information Sciences, pages 42–51, 1997. 4.1
[29] P. Garnell and D. J. East. Guided Weapon Control System. Pergamon Press,2nd
Ed.,Oxford, 1980. 1.1.3
[30] A. L. Greensite. Analysis and Design of Space Vehicle Flight Control Systems. Perg-
amon Press,2nd Ed.,Oxford, 1970. 1.1.3
[31] G.U. Guoxiang, James R. Cloutiert, and Gisoon Kim. Gain scheduled missile au-
topilot design using observer-based h∞ control. Proceedings of the American Control
Conference, 3:1951–1955, June 1995. 1.5.4
[32] E.C. Gwo and J. Hauser. A numerical approach for approximate feedback lineariza-
tion. Proceedings in American Control Conference,San Fransisco,Callifornia, pages
1495–1499, 1993. 2.2
[33] Charles E. Hall and Yuri B. Shtessel. Rlv sliding mode control system using sliding
mode observers and gain adaptation. AIAA Guidance, Navigation, and Control
Conference and Exhibit, (AIAA 2003-5437). 1.5.3
[34] M. J. Hemsch. Tactical Missile Aerodynamics: General Topics, volume 146. AIAA,
1991. 1
[35] J. Huang, C. F. Lin, J. R. Cloutier, J. H. Evers, and C. D. Souza. Robust feedback
linearization approach to autopilot design. Proc. IEEE Conf. Contr. Applicat., 1:220–
225, 1992. 1.5.1, 1.5.3
BIBLIOGRAPHY 196
[36] J. Huang and C.F. Lin. Application of sliding mode control to bank-to-turn missile
systems. The First IEEE Regional Conference on Aerospace Control Systems, pages
569 – 573, May 1993. 1.5.3, 4.2.1.2, 4.2.2.1
[37] Richard A. Hull and Zhihua Qu. Dynamic robust recursive control design and its
application to a nonlinear missile autopilot. Proceedings of the American Control
Conference, 1:833–837, June 1997. 1.5.4
[38] Alberto Isidori. Nonlinear Control Systems: an introduction. Berlin: Springer-Verlag,
1989. 2.3.2
[39] Solsona J., Valla M.I., and Muravchik C. A nonlinear reduced order observer for
permanent magnet synchronous motors. 20th International Conference on Industrial
Electronics, Control and Instrumentation, 1:38–43, 1994. 1.5.6
[40] Solsona J., Valla M.I., and Muravchik C. A nonlinear reduced order observer for
permanent magnet synchronous motors. IEEE Transaction on Industrial Electronics,
43(4):492–497, August 1996. 1.5.6
[41] Kemin. Jhou and John C. Doyle. Essential of Robust Control. Prentice Hall, first
edition, 1997. 3.2.1, B.1.10
[42] Yuqiang Jin, Wen-Jin Gu, , Jinhua Wu, and Xuebao Wang. Sliding mode control of
btt missile based on fuzzy-neural approach. Proceedings of the 5th World Congress
on Intelligent Control and Automation, pages 5483–5486, June 2004. 1.5.3
[43] W. Jin-Yong, You-An Zhang, and Wen-Jin Gu. An approach to integrated guidance
and autopilot design for missiles based on terminal sliding mode control. Proceedings
of the Third Intemational Conference on Machine Learning and Cybernatics, pages
610–615, August 2004. 1.5.3
[44] Nam H. Jo and Jin H. Seo. Input output linearization approach to state observer
design for nonlinear system. IEEE Transactions On Automatic Control, 45(12):2388–
2393, December 2000. 1.5.6
BIBLIOGRAPHY 197
[45] Andreas. Johansson. Nonlinear Observers with applications in the steel industry.
Doctoral Thesis, 2001. 1.5.6
[46] S. Jorge, I.V. Maria, and M. Carlos. A nonlinear reduced order observer for per-
manent magnet synchronous motors. IEEE Transactions on Industrial Electronics,
43:492–497, August 1996. 2.3
[47] Nikolaos Kazantzis and Costas Kravaris. Nonlinear observer design using lyapunovs
auxiliary theorem. Proceedings of the 36th Conference on Decision and Control, pages
4802–4807, December 1997. 1.5.6
[48] H. K. Kkalil. Nonlinear System. Prentice Hall, 1996. 2.2
[49] Karlene A. Kosanovich, , Michael J. Piovoso, Vadim Rokhlenko, and AIIon Guezl.
Nonlinear adaptive control with parameter estimation of a cstr. J. Proc. Cont.,
5(3):137–148, 1995. 1.5.6
[50] A.J. Koshkouei, K.J. Burnham, and A.S.I. Zinober. Dynamic sliding mode control
design. Control Theory and Applications, IEE Proceedings, 152(4):392 – 396, July
2005. 1.5.3
[51] S. H. Lane and R. F. Stengel. Flight control design using nonlinear inverse dynamics.
Automatica, 24(4):471–483, 1988. 2.2.2.1
[52] Ju-Il. Lee and In-Joong. Ha. Autopilot design for highly maneuvering stt missiles via
singular perturbation-like technique. IEEE Trans. on Control Systems Technology,
7(5):527–541, September 1990. 1.3.2, 1.5.1
[53] Sang-Yong. Lee, Ju-Il. Lee, and In-Joong. Ha. A new approach to nonlinear autopilot
design for bank-to-turn missiles. Procedings of the 36th Conference on Decision and
Control, pages 4193–4197, December 1997. 1.5.1
[54] K. Y. Lian, L. C. Fu, D. M. Chuang, and T. S. Kuo. Nonlinear autopilot and guidance
for a highly maneuverable missile. Proc. Amer. Contr. Conf., Baltimore, MD, pages
2293–2297, June 1994. A.1
BIBLIOGRAPHY 198
[55] Kuang-Yow. Lian, Li-Chen. Fu, Dung-Ming. Chuang, and Teh-Son Kuo. Nonlin-
ear autopilot and guidance for a highly maneuverable missile. American Control
Conference, 2:2293– 2297, August 2002. 1.5.2
[56] C. F. Lin, J.R. Cloutier, and J.H. Evers. Missile autopilot design using a generalized
hamiltonian formulation. The First IEEE Regional Conference on Aerospace Control
Systems, pages 715–723, 1993. 1.5.4
[57] Ching-Fang Lin, James R. Cloutier, and Johnny H. Evers. Robust bank-to-turn
missile autopilot design. Procedings of the American Control Conference, pages 1941–
1945, June 1995. 1.5.4
[58] Farza M., Busawon K., and Hammouri H. Simple nonlinear observers for on-line
estimation of kinetic rates in bioreactors. Automatica, 34(3):301–318, March 1998.
1.5.6
[59] Timothy Massey and Yuri Shtessel. Satellite formation control using traditional and
high order sliding modes. AIAA Guidance, Navigation, and Control Conference and
Exhibit, (AIAA 2004-5021). 1.5.3
[60] P. K. Menon, V. R. Iragavarapu, and E. J. Ohlmeyer. Nonlinear missile autopi-
lot design using time-scale separation. AIAA Guidance, Navigation, and Control
Conference, 41:1791–1803, 11-13 Aug 1997. 1.5.2
[61] P.K. Menon and E.J. Ohlmeyer. Computer-aided synthesis of nonlinear autopilots
for missiles. Nonlinear Studies, 11(2):173–198, 2004. 2.2
[62] P.K. Menon and M. Yousefpor. Design of nonlinear autopilots for high angle of attack
missiles. Optimal Synthesis, 1996. 1.5.1, 1.5.2
[63] G. Meyer, R. Su, and L. R. Hunt. Application of nonlinear transformations to
automatic flight control. Automatica, 20:103–107, 1984. A.2.2
BIBLIOGRAPHY 199
[64] M.C. Mickle and J.J. Zhu. A nonlinear roll-yaw missile autopilot based on plant
inversion and pd-spectral assignment. Proceedings of the 37th IEEE Conference on
Decision and Control, 4:4679–4684, December 1998. 1.5.4
[65] Heon Seong Nam, Seung-Hwan Kim, Chanho Song, and Joon Lyou. A robust non-
linear control approach to missile autopilot design. Proceedings of the 40th SICE
Annual Conference.International Session Papers, pages 310–314, July 2001. 1.5.4
[66] F.W. Nesline, B.H. Wells, , and P Zarchan. A combined optimal/classical approach to
robust missile autopilot design. AIAA Guidance Navigation and Control Conference,
Collection of Technical Papers. (A79-45351 19-12):265–280, 6−8thAugust 1979. 1.5.5
[67] C.M. Ng, C.A. Rabbath, N. Hori, and M. Lauzon. Modern multivariable and multi-
loop digital redesign of missile autopilots. Proceedings of the American Control Con-
ference, 1:414– 419, June 2003. 1.5.7
[68] I.R Petersen. Guaranteed cost control of stochastic uncertain systems applied to a
problem of missile autopilot design. Proceedings of the American Control Conference,
3:1737–1741, June 1998. 1.5.4
[69] A.S. Poznyak, R. Martinez-Guerra, and A. Osorio-Cordero. Robust high-gain ob-
server for nonlinear closed-loop stochastic systems. Journal of Mathematical Prob-
lems in Engineering, 6:31–60, 2000. 1.5.6
[70] A. Rantzer and P.A. Parrilo. On convexity in stabilization of nonlinear systems.
Proceedings of the 39th IEEE Conference on Decision and Control, 2000., 3:2942–
2945, December 2000. A.2.2
[71] L. Richard, M. Lauzon, and A. Jeffrey. Integrated autopilot tuning methodology for
airframe parametric simulation. AIAA Guidance, Navigation and Control Conference
and Exhibit, AIAA-2002-4661, August 2002. 1.5.2
[72] R. Rysdyk, B. Leonhardt, and A. J. Calise. Development of an intelligent flight
BIBLIOGRAPHY 200
propulsion control system: Nonlinear adaptive control. Guidance Navigation and
Control Conference, AIAA-2000-3943, August 2000. 1.5.5
[73] C. Schumacher and P Khargonekar. A comparison of missile autopilot designs using
h∞ control with gain scheduling and nonlinear dynamic inversion. Proceedings of the
American Control Conference, 5:2759 – 2763, June 1997. 1.5.2
[74] Corey J. Schumacher, Gerald C. Cottrill, and Hsi-Han. Yeh. Optimal sliding mode
flight control. AIAA Guidance, Navigation, and Control Conference and Exhibit,
(AIAA-99-4002). 1.5.3
[75] Jeff S. Shamma and James R. Cloutier. Gain-scheduled missile autopilot design
using linear parameter varying transformations. Journal of Guidance, Control, and
Dynamics, 16(2):256–263, March-April 1993. 1.5.7
[76] A. Shang, Wen-jin Gu, Dehai Yu, and Yong Liang. Design of missile autopilot ap-
plying the immersion and invariance algorithm. Fifth World Congress on Intelligent
Control and Automation, 6:5479 – 5482, June 2004. 1.5.7
[77] Hyungbo Shim and Jin H. Seo. Semi..global observer for multioutput nonlinear
systems. Proc. of the Applied Mathematics Letters, 48(2):294–298, February 2003.
1.5.6
[78] Tal Shima, Moshe Idan, and Oded M. Golan. Sliding mode control for integrated
missile autopilot-guidance. AIAA Guidance, Navigation, and Control Conference
and Exhibit, (AIAA 2004-4884). 1.5.3
[79] I. Shkolnikov, Yuri B. Shtessel, and D. Lianos. Integrated guidance-control system
of a homing interceptor: Sliding mode approach. AIAA Guidance, Navigation, and
Control Conference and Exhibit, (AIAA-2001-4218). 1.5.3
[80] I. Shkolnikov, Yuri B. Shtessel, P. Zarchan, and D. Lianos. Simulation study of
the homing interceptor guidance loop with sliding mode observers versus kalman
BIBLIOGRAPHY 201
filter. AIAA Guidance, Navigation, and Control Conference and Exhibit, (AIAA-
2001-4218). 1.5.3
[81] Yuri B. Shtessel, James Strott, and J. Jim Zhu. Time-varying sliding mode control
with sliding mode observer for reusable launch vehicle. AIAA Guidance, Navigation,
and Control Conference and Exhibit, (AIAA 2003-5362). 1.5.3
[82] Yuri B. Shtessel, J. Jim. Zhu, and Dan. Daniels. Reusable launch vehicle attitude
control using time-varying sliding modes. AIAA Guidance, Navigation, and Control
Conference and Exhibit, (AIAA 2002-4779). 1.5.3
[83] Jean-Jacques E. Slotine and Weiping. Li. Applied Nonlinear Control. Prentice Hall,
1991. 2.2
[84] Seong-Ho Song, Sang-Yong Lee, Jeom-Keun Kim, Gyu. Moon, Seop Hyeong Park,
and Sun Yong Kim. Design of a stt missile autopilot using functional inversion and
lmi approach. Proceedings of the 39th IEEE Conference on Decision and Control,
4:3596 – 3597, December 2000. 1.5.1
[85] Masoud Soroush. Nonlinear state-observer design with application to reactors. Jour-
nal of Chemical Engineering Science, 52(3):387–404, February 1996. 1.5.6
[86] V. Sundarapandian. General observers for nonlinear systems. Mathematical and
Computer Modelling, 39(2):97–105, February 2004. 1.5.6
[87] Minjea Tahk, M. Michael Brigges, and P.K. Menon. Application of plant inversion
via state feedback to missile autopilot design. Procedings of the 27th IEEE Conference
on Decesion and Control, pages 730–735, December 1988. 1.5.4
[88] B. Targui, M. Farza, and H. Hammouri. Constant-gain observer for a class of multi-
output nonlinear systems. Proc. of the Applied Mathematics Letters, 15(6):709–720,
August 2002. 1.5.6
BIBLIOGRAPHY 202
[89] Antonios Tsourdos, Anna Blumel, and Brian White. Autopilot design of a non-
linear missile. UKACC IEE International Conference on CONTROL, (455):889–894,
September 1998. 1.5.1
[90] David P. White, Jason G. Wozniak, and Douglas A. Lawrence. Missile autopilot
design using a gain scheduling technique. Proceedings of the 26th Southeastern Sym-
posium on System Theory, pages 606–610, 20-22 March 1994. 1.5.2
[91] L. Xie, M. Fu, and C.E.de. Souza. h∞ control and quadratic stabilization of systems
with parameter uncertainity via output feedback. IEEE Trans. Automatic Control,
37:1253–1256, August 1992. B.3.1
[92] L. Xie, M. Fu, and C.E.de. Souza. h∞ analysis and synthesis of discrete-time systems
with time-varying uncertainity. IEEE Trans. Automatic Control, 38:459–462, March
1993. B.4.4
[93] Yi Xiong and Mehrdad Saif. Sliding mode observer for nonlinear uncertain systems.
IEEE Transactions On Automatic Control, 46(12):2012–2016, December 2001. 1.5.6
[94] Yuru Xu, Jinyong Yu, Yuman Yuan, and Wen-Jin Gu. Adaptive fuzzy sliding-mode
controller for btt missile. 8th Control, Automation, Robotics and Vision Conference,
2:1222 – 1226, December 2004. 1.5.3
[95] Jin-Yong Yu, You-An Zhang, and Wen-jin Gu. An approach to integrated guid-
ance/autopilot design for missiles based on terminal sliding mode control. Proceed-
ings of International Conference on Machine Learning and Cybernetics, 1:610– 615,
August 2004. 1.5.7
[96] Hongchao Zhao, Wen-Jin Gu, Yunan Hu, and Changpeng Pan. Second-order sliding
mode control for aerodynamic missiles using backstepping design. Proceedings of the
5th World Congress on Intelligent Control and Automation, pages 1375–1378, June
2004. 1.5.3
BIBLIOGRAPHY 203
[97] J.J. Zhu and M.C. Mickle. Missile autopilot design using the extended-mean as-
signment control. i. stabilization. Proceedings of the Twenty-Seventh Southeastern
Symposium on System Theory, pages 247 – 251, 12-14 March 1995. 1.5.7
[98] J.J. Zhu and M.C. Mickle. Missile autopilot design using the extended-mean as-
signment control. i. stabilization. Proceedings of the Twenty-Seventh Southeastern
Symposium on System Theory, pages 120–124, March 1996. 1.5.7