in-field simulation considering considering analog variability€¦ · in-field simulation...
TRANSCRIPT
ww
w.ic
t.tu
wie
n.a
c.at
Institute of
Computer Technology /99
In-field Simulation ConsideringConsidering Analog Variability
FAC’18 – May 16, 2018
Michael Rathmair1, Carna Radojicic2 and Christoph Grimm2
1Insttute of Computer Technology – TU Wien2Design of Cyber-Physical Systems – TU Kaiserslautern
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 2 of 15In-Field Simulation Considering Analog Variability
Motivation
System simulation processes during the design phase Model driven design Verification and implementation models Extend simulations to the operation phase
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 3 of 15In-Field Simulation Considering Analog Variability
Motivation
System simulation processes during the design phase Model driven design Verification and implementation models Extend simulations to the operation phase
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 4 of 15In-Field Simulation Considering Analog Variability
Concept and Approach
Compute simulation results during runtime Added value for process control Dynamic environment facing uncertainty Feedback loops
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 5 of 15In-Field Simulation Considering Analog Variability
Requirements
Challenges and requirements – a wish list Meet-in-the-middle design approach
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 6 of 15In-Field Simulation Considering Analog Variability
Requirements
Models• C/C++ based descriptions
• Use of verification and implementation models
• Models of the environment
Controlability / Observability• Process state evaluation
• Parameter tuning
• Operational bounds
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 7 of 15In-Field Simulation Considering Analog Variability
Requirements
Uncertainty modeling• Impact of parameter deviations
• Increasingly challenging
• Multi-run methods vs. Affine Arithmetic Forms
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 8 of 15In-Field Simulation Considering Analog Variability
Requirements
Analysis and decision-making• Objective -driven system analysis
• Application specific algorithms
• Expert-knowledge
• Machine-learning approaches
t t
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 9 of 15In-Field Simulation Considering Analog Variability
Requirements
Simulator Software• Discrete-event simulator
• Parallel and distributed
• C/C++ language
• Build-management
Deployment• Usage of permanently or partially free resources
• Coordinated integration of the simulation software
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 10 of 15In-Field Simulation Considering Analog Variability
Example Use-case
Analog behavior Optimize performance Industrial safety-critical application
• Continuously observing safety properties
• Re-Certification of configurable behavior
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 11 of 15In-Field Simulation Considering Analog Variability
Development Flow
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 12 of 15In-Field Simulation Considering Analog Variability
Work in Progress and Next Steps
What we have:• Simulation core for analog systems
• Design time verification models
• C++ library for uncertainty representation
• Objective driven system analysis
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 13 of 15In-Field Simulation Considering Analog Variability
Work in Progress and Next Steps
Next Steps:• Software build management
Platform / Library support Dependable analysis and debug functions
• Extend / modify the simulator core
• Guided integration into an application
ww
w.ic
t.tu
wie
n.a
c.at
Michael Rathmair
Institute of Computer Technology 14 of 15In-Field Simulation Considering Analog Variability
Conclusion
Extend simulation processes to operation phase Consideration of uncertainties and dynamic behavior Models used for design time verification Parallel and distributed simulation System Optimization and analysis
• Safety critical issues
• Predictive maintenance
• Etc.
ww
w.ic
t.tu
wie
n.a
c.at
Institute of
Computer Technology /99
Thank you for your attention
[email protected] of Computer Technology – TU Wien
Michael Rathmair, Carna Radojicic and Christoph Grimm