single-phase hybrid active power filter with adaptive...

20 IETE JOURNAL OF RESEARCH | VOL 57 | ISSUE 1 | JAN-FEB 2011

Single-phase Hybrid Active Power Filter with Adaptive Notch Filter for Harmonic Current

EstimationSaad Mekhilef, Messikh Tarek and Nasrudin Abd. Rahim

Department of Electrical Engineering, University of Malaya, Kuala Lumpur, Malaysia

ABSTRACT

This paper presents a digital implementation of a new and simple control technique for single-phase Hybrid Active Power Filter (HAPF). The main control consists of an accurate estimation technique used to extract harmonic current generated by nonlinear loads. The estimation of harmonic current is done by using multiple units of an Adaptive Notch Filter (ANF) linked together in parallel structure; each unit is used to estimate a specified harmonic component. The frequency estimation is then carried out based on the output of these units. Therefore, the sum of these harmonic components will be the reference signal for the active power filter to generate the appropriate compensation signal. A laboratory prototype model of the HAPF was developed and its control strategy was implemented digitally in TMS320F2808. The theoretical expectations were verified and demonstrated experimentally.

Keywords:Hybrid active power filter, Adaptive notch filter, Harmonic estimation, Digital signal processor, Power quality.

1. INTRODUCTION

With the proliferation of nonlinear loads, a mass of harmonic current and reactive power flow into the network, and their contribution to the waveform distortion degrade the power quality in power transmission and distribution systems. Active power filter (APF) is an effective means that can be used to suppress harmonics and compensate reactive power dynamically. It basically works by injecting a current that has equal amplitude and opposite phase to loads’ harmonic current, into the utility AC power system. Thereby, the aim to suppress harmonics and purify network is achieved [1,2].

APF can be connected either in series or in parallel to power system; therefore, it can operate as either a voltage source or current source. The shunt APF is controlled to inject a compensate current into the utility system so that it cancels the harmonic current produced by nonlinear loads. The series APF is controlled to insert a distorted voltage such that the load voltage is sinusoidal and maintained at a rated magnitude. APFs, with a more autonomous operation, can proceed directly and dynamically to compensate harmonics or to damp harmonics resonance in power system. Hence, numerical methods for harmonic detection are needed to control APFs [3-5].

Harmonic components can be extracted or estimated

using the Fast Fourier Transform (FFT), the instantaneous p-q theory, the synchronous d-q reference frame theory or by using suitable analog or digital electronic filters separating successive harmonic components [6-8]. For harmonic current determination by digital filters, a cascade structure of a notch filter is widely used. However, notch filter gives better performance only when the frequency of the measured signal remains constant. So, when the frequency changes, the ideal solution is a notch filter capable of changing the notch frequency accordingly by tracking the frequency variation [9,10]. Based on adaptive filtering, this paper presents a new control strategy of single-phase APF using an adaptive notch filter as harmonics detector.

2. TOPOLOGY OF THE PROPOSED ACTIVE POWER FILTER

Among the various topologies of the active power filtering, the hybrid active power filter (HAPF) has been chosen to isolate the current source from harmonics generated by nonlinear load. An HAPF is a combination of a series APF and a shunt passive filter. The combined filter system aims to eliminate the problems encountered while using only conventional shunt passive or conventional shunt active filters. Its steady compensation characteristics have been demonstrated, and it has been verified theoretically and experimentally that the combined system has the following features [11-15, 16-17].

[Downloaded free from http://www.jr.ietejournals.org on Sunday, June 05, 2011, IP: 202.185.106.232] || Click here to download free Android application for this journal

https://market.android.com/details?id=comm.app.medknow

21IETE JOURNAL OF RESEARCH | VOL 57 | ISSUE 1 | JAN-FEB 2011

• The harmonic compensation effect is not influenced by the source impedance of the AC power system.

• There is no parallel resonance between the source impedance and the shunt passive LC filter because the series active filter is placed with the source.

• Ambient harmonics generated elsewhere on the AC system, including harmonics of the AC source, do not sink into the LC filter. At the same time, they are isolated from load harmonics by the series active filter.

• The required rating of the series active filter is considerably smaller than that of the conventional active filter; the initial and running costs of the combined system are as cheap as a usual shunt passive filter.

The circuit configuration, shown in Figure 1, is composed of a diode bridge rectifier with R-L load for harmonics, producing a passive filter in parallel with the load and an APF connected in series with line using a series transformer (CT). The principal function of the proposed topology can be examined using the equivalent circuit shown in Figure 2.

The purpose of the HAPF is to produce the compensation voltage that is injected into the main circuit by coupling transformer. The compensation voltage is proportional to the harmonic components of the supply current.

U R Isha a= ∑ (1)

Now on the supply side of the transformer, there is voltage that is proportional to the current of the transformer, which means that the transformer can be seen as an active resistance Ra by the main circuit. Steady-state vector equations can be written for every frequency

component, as follows:

Z I U Z I R I Z Is h s h a h z h s h a s h f h f h, , , , , , , ,+ = + = (2)

I I Is h f h l h, , ,= + (3)

In these equations, the subscripts s refer to source branch, h to harmonic and f to parallel passive filter branch. Combining Equations (2) and (3), we get the following relation:

IZ

Z Z RIs h

f h

f h s h al h,

,

, ,,=

+ + (4)

From Equation (4) it can be seen that when the value of the active resistance Ra increases, the amplitude of the harmonic components of the supply current decreases. This means in practice that the impedance ratio between source and filter branches increases, which forces the harmonic components to flow more effectively into the L-C shunt circuit.

3. CONTROL STRATEGY OF THE HYBRID ACTIVE POWER FILTER

Basically, for harmonic current suppression, the power converter of the series APF is presented as a harmonic resistor to reduce the main harmonic current. The output voltage of the series active filter is related to harmonic current. Thus, the inverter injects harmonic voltage which is proportional to the harmonic current into the line. Based on the above fundamental concept, the control strategy of the HAPF consists of sensing the current source (Ish), extracting the harmonic current, multiplying the results by a gain and that is the reference

Figure 1: The proposed topology of hybrid active power filter topology. Figure 2: Equivalent circuit of hybrid active power filter.

Mekhilef S, et al.: Single-phase Hybrid Active Power




signal for the pulse width modulation (PWM) voltage source inverter. Therefore, the main control consists of an accurate estimation technique for harmonic current.

The estimation of harmonic current is done by using multiple units of an ANF linked together in a parallel structure; each unit is used to estimate a specified harmonic component. Moreover, each unit is governed by two ordinary differential equations and a second-order notch filter that is further furnished with a nonlinear differential equation to update frequency [12,18,20]. The frequency estimation is then carried out based on the output of these units. Therefore, the sum of these harmonic components will be the reference signal for the APF to generate the appropriate compensation signal.

3.1 Harmonic Current Estimation

To determine Ish, an adaptive notch filter is used. It consists of n ANF units linked together in parallel structure to decompose the current into its constituting components. Each unit is governed by two sets of differential equations. The frequency estimation is carried out based on the outputs of these units. The particular ANF structure, which is our concern, originated from [11,22-23]. It is characterized by the dynamic behavior of the following differential equations:

x i x e t i n

x e t

e t y t x t

i i

t

..

.

.

( ) , , ...,

( )

( ) ( ) ( )

+ = =

= −

= −

2 2

1

2 1 2θ ζθ

θ γ θ

ii

n

=∑

1

(5)

where θ is the estimated frequency, ξ is the depth of ANF, γ determines the adaptation speed, e(t) is the adjusting error and x is the estimated signal for the input signal y(t).

3.2 Performance of the Proposed ANF

To test the effectiveness of the estimation technique, a current supply of one phase of controlled three-phase rectifier defined as

y(t) = 100 sin(ωt - 0.3) + 20 sin(5ωt + 0.49) + 14 sin (7ωt - 0.57) + 8 sin(11ωt + 0.45) + 6 sin(13ωt - 0.61) + 3 sin(17ωt + 0.41) + sin(19ωt - 0.65) (6)

is taken for estimation. The ANF algorithm is implemented using a sampling time Ts = 0.0001 second, ξ = 0.08, γ = 0.04 and f = 50 Hz. The input signal y(t) passes trough n parallel units to estimate both even and odd order harmonic components that exist in y(t). Figures 3–11 show, respectively, the comparison between the real

component and the estimated one generated by different ANF units. When the simulation results are compared with the actual one, a good match is observed. The proposed ANF structure has estimated both magnitude and phase of the fundamental frequency and each odd harmonic component. The results obtained with the even units are zero because the even harmonic components are not included in the input signal.

Another key feature of this adaptive filter is the adaptation speed defined by the coefficient γ. A simulation result is carried out with different values of γ to show the speed of the filter in estimating the harmonic components. From Figures 12–17, it can be seen that when γ increases, the capability of the ANF for tracking frequency increases and the time of converging is fast. However, with a very high value of γ, the filter will not converge and the estimated signal will have a distorted shape. Thus, it is

Figure 3: Fundamental real and estimated.

Figure 4: Total harmonics real and estimated.





Figure 5: 5th Harmonic real and estimated.


Figure 9: 17th Harmonic real and estimated. Figure 10: 19th Harmonic real and estimated.







very important to choose an accurate value of γ. On the other hand, increasing the value of γ can achieve faster estimation, but at the same time, ζ should be increased in order to avoid oscillatory behavior.

4. SYSTEM CONTROL IMPLEMENTATION

The harmonic current determination by the ANF and the HAPF control implementation are carried out with Matlab simulation and hardware implementation in a digital signal processor (DSP). All calculations are accomplished digitally with the parameters given in Table 1. The current source (Is) is sensed and fed to the

Figure 11: 11th even harmonics.

Figure 13: Comparison between real and estimated total harmonics.


Figure 14: Frequency evaluation with γ = 0.6.







Table 1: Hybrid active power filter parametersComponents Values Source frequency 50 HzAC supply voltage 110 VPWM frequency 20 kHzDC bus capacitor 470 uFLoad L=200 mH, R=100 ΩPassive filter L=3 mH,C=60 uF, R=100 Ω


Figure 18: Flowchart of the proposed hybrid active power filter control algorithm.

analog digital converter (ADC) of the TMS320F2808, and then the data are filtered by the ANF function, multiplied by a gain and sent to the PWM block in DSP to generate the accurate pulses for the inverter.

Figure 18 illustrates the operations sequence of the C code of the PWM control that is executed on the CC Studio 3.1 platform. As shown in this flowchart, the controller sample times are governed by the DSP timer called a CpuTimer0, which generates periodic interrupt at each sampling time Ts. At each interrupt service routine (ISR), the following tasks must be executed: • Read the sampling value of the current source Is from

the ADC.• Process the data obtained from the ADC to filter it

using the adaptive notch filter equations.• Multiply the output from the ANF processing by a

gain K.• Update the compared value of the pulse width

modulation module (ePWM) built in TMS320F2808 DSP.

The ePWM module generates the appropriate pulses for the inverter switches. The output of the inverter depends on the modulation index represented by K and the DC voltage input (VDC).

Figures 19–22 show, respectively, the simulation and hardware results for unfiltered current source, filtered

current source with only passive filter, total current source harmonic estimated by the adaptive notch filter and the current source filtered with the HAPF. From the results obtained, it is clear that the single-phase HAPF proposed for harmonic current mitigation features great simplicity in the control side with a good performance in harmonic current mitigation shown by keeping the Total Harmonic Distortion (THD) of the current source within acceptable limits. It reduces the supply current THD from 24.33% down to 4.57%. The ability of harmonic current filtering is due to the ability of the proposed ANF to track the frequency variation and extract the harmonic components of the measured signal.

Table 2 provides a summary of the proposed estimation technique and active power filtering. As it can be seen, the results obtained from the simulation and experimental tests are almost the same. However, when using the compensation with the HAPF, the simulation results





Figure 19: Simulation and experimental results of unfiltered current source (0.5 A/div., 5 ms/div.).

Figure 20: Simulation and experimental results of current source filtered with passive filter (0.5 A/div., 5 ms/div.).

Figure 21: Simulation and experimental results of harmonic current source (50v/div., 5 ms/div.).

Figure 22: Simulation and experimental results of current source filtered with HAPF (0.5 A/div., 5 ms/div.).





show a decrease in the amount of THD for the 7th–19th harmonic components, whereas for the experimental results there is an increase in the amount of THD for some harmonic components. This is due to the injection of a waveform which is not very smooth and also to the characteristics of the series transformer which behaves as a second filter for the injected waveform. On the other hand, the components used in simulation are ideal compared to the experimental results.

5. CONCLUSION

The single-phase HAPF with an adaptive notch filter for harmonics estimation, presented in this paper, was effective to mitigate harmonics generated by nonlinear load and keep the utility AC power supply line current sinusoidal. The advantage of the proposed method is the capability of the ANF for extracting the harmonics as well as the fundamental component. Furthermore, The ANF has a simple structure which could be easily implemented in both software and hardware.

REFERENCES

1. H Akagi, Modern active filters and traditional passive filters, Bulletin of the polish academy of sciences technical science, Vol. 54, No.3: 255-6, 2006.

2. A M Massoud, S J Finney, and B W Williams, Review of harmonic current extraction techniques for an active power filter, IEEE International Conference on Harmonics and Quality of Power, 154-9, 2004.

3. F Z Peng, Harmonic sources and filtering approaches, IEEE Industry

Table 2: Simulation and experimental comparison of THD No

compensation (%)

Compensation with

PF (%)

Compensation with HAPF

(%)3rd Harmonic

Simulation 14.66 6.52 3.23Experiment 17.33 1.23 1.17

5th Harmonic


7th Harmonic


9th Harmonic


11th Harmonic


13th Harmonic


15th Harmonic


17th Harmonic


19th Harmonic


Applications Magazine, Vol. 7, 18-25, 2001.

4. M McGranaghan, Active filter design and specification for control of harmonics in industrial and commercial facilities, Electrotek Concepts, Inc, Knoxville TN, USA, 2005.

5. R ElShatshat, M Kazerani, and M M A Salama, On-line tracking and mitigation of power system harmonics using ADALINE-based active power filter system, Electrical and Computer Engineering, Canadian Conference, Vol. 4, 2119-24, 2004.

6. M Longhua, Application of adaptive filtering in harmonic analysis and detection, IEEE/PES, transmission and distribution Conference & Exhibition: Asia and Pacific Dalian, China, pp. 1-4, 2005.

7. A Ozdemir, A digital adaptive filter for detecting harmonic, active and reactive current, Institute of Physics Publishing, Meas Sci, pp. 1316-22, 2004.

8. K Nishida, M Rukonuzzaman, and M Nakaoka, A novel single-phase shunt active power filter with adaptive neural network based harmonic detection, IEEJ Trans. LA, Vol. 125, N0. 1: 9-15, 2005.

9. M Mojiri and A R Bakhshai, An adaptive notch filter for frequency estimation of a periodic signal, IEEE Transaction on Automatic Control, Vol.49, N0.2: pp. 315-8, Feb 2004.

10. A H Sayed and T Kailath, A state-space approach to RL adaptive filtering, IEEE Signal Processing Magazine, pp. 18-60, July 1994.

11. M Karimi-Ghartemani, M Mojiri, and A R Bakhshai, A technique for extracting time-varying harmonics based on adaptive notch filter, IEEE Conference on control applications, Toronto, Canada, pp. 624-8, 2005.

12. L Asiminoaei, F Blaabierg, and S Hansen, Evaluation of harmonic detection methods for active power filters application, APEC, Vol. 1, pp. 635-41, 2005.

13. S Jain, P Agarwal, H O Gupta, and G Agnihotri, Modeling of frequency domain control of shunt active power filter using Matlab simulink and power system blockset, Proc. IEEE in Electrical Machines and Systems, ICEMS, Vol. 2, pp. 1124-9, 2005.

14. H J Zhao, Y F Pang, Z M Qiu, and M Chen, Study on UPF harmonic current detection method based on DSP. Institute of Physics Publishing, Journal of Physics, International Symposium on Instrumentation Science and Technology, Vol. 48, pp. 1327-31, 2006

15. Z Pan, and F Z Peng, Power factor correction using a series active filter, IEEE Transactions on Power Electronics, Vol. 20, No. 1: pp. 148-53, 2005.

16. F B Libano, S L Muller, R A Marques Braga, J V Rossoni Nunes, O S Mano, and I A Paranhos, Simplified control of the series active power filter for voltage conditioning, International Symposium on Industrial Electronics (ISIE), Vol. 21: pp. 1706-11, 2006.

17. K Nishida, M Rukonuzzaman, and M Nakaoka, A novel single-phase shunt active power filter with adaptive neural network based harmonic detection, IEEJ Trans. LA, Vol. 125, N0. 1: pp. 9-15, 2005.

18. J Turunen, M Salo, and H Tuusa, Comparison of series hybrid active power filter based on experimental tests, IEEE European Conference on Power Electronics and Applications, pp. 1-10, 2005.

19. C C Hua and C W Chuang, Design and implementation of a hybrid series active power filter, IEEE international conference on PEDS, Vol. 2, pp. 1322-6, 2005.

20. H K Chiang, B R Lin, K T Yang, and K W Wu, Hybrid active power filter for power quality compensation, IEEE International conference on PEDS, Vol. 2: pp. 949-54, 2005.

21. W Xiaoyu, L Jmjun, Y Chang, and W Zhaoan, A comparative study on voltage source control and current source control of series active power filter, IEEE International Conference on APEC, pp. 1-6, 2006.

22. M S Feitas, J L Afonso, and J S Martins, A single-phase power series compensator for voltage distortion, International conference on electrical engineering, CEE’05, Coimbra, pp. 1-5, 2005.

23. S Zabihi and F Zarc, Active power filters with unipolar pulse width modulation to reduce switching losses, International Conference on Power System Technology, pp.1-5, 2006.





Department, Faculty of Engineering, University Malaya, Malaysia. His current research interests include power electronics and power system studies.

E-mail: [email protected]

Nasrudin Abd. Rahim was born in Johor, Malaysia, in 1960. He received the B.Sc. (Hons.) and M.Sc. degrees from the University of Strathclyde, Glasgow, U.K., and the Ph.D. degree in 1995 from Heriot–Watt University, Edinburgh, U.K. He is currently a Professor with the Department of Electrical Engineering, University of Malaya, Kuala Lumpur, Malaysia, and the Director of the

Center of Research for Power Electronics, Drives, Automation and Control. Dr. Rahim is a Fellow of the Institution of Engineering and Technology, U.K., and a Chartered Engineer. He is also Chairman of the Working Group WG-8, covering reluctance motors, of the IEEE Motor Subcommittee under IEEE Power Engineering Society/Electric Machinery Committee. His research interests include Power Electronics, Real-time Control Systems, and Electrical Drives.


AUTHORSSaad Mekhilef received the B.E. degree in Electrical Engineering from University of Setif, Algeria in 1995, and Master of Engineering Science and Ph.D. from University of Malaya in 1998 and 2003 respectively. He is currently professor at Department of Electrical Engineering University of Malaya. Prof. Saad is the author and co-author of more than 100 publications in

international journals and proceedings. He is actively involved in industrial consultancy, for major corporations in the power electronics projects. His research interest includes Power Conversion Techniques, Control of Power Converters, Renewable Energy and Energy Efficiency.


Messikh Tarek was born in Annaba, Algeria. He received his B.E. degree in Electrical Engineering from University of Annaba, Algeria and Master of Engineering Science in Electrical Engineering from University of Malaya, Malaysia in 1996 and 2007 respectively. He is currently a Ph.D. student and a research assistant at the Electrical

DOI: 10.4103/0377-2063.78316; Paper No. JR 433_09; Copyright © 2011 by the IETE




single-phase hybrid active power filter with adaptive...

Documents