644 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 2, MAY 2005 A Novel Time-Domain Current-Detection Algorithm for Shunt Active Power Filters Hongyu Li, Fang Zhuo, Zhaoan Wang, Senior Member, IEEE, Wanjun Lei, and Longhui Wu Abstract—A novel current-detection algorithm based on the time-domain approach for three-phase shunt active power filters (APFs) to eliminate harmonics, and/or correct power factor, and/or balance asymmetrical loads is analyzed in this paper. A basic overview and evaluation of the performance of existing current-detection algorithms for active power filters are presented. According to different complicated power quality issues and various compensation purposes, a novel current-detection algorithm is then proposed. Comparing with existing algorithms, this algorithm has shorter response time delay and clearer physical meaning. Different compensating current references can, thus, be accurately and easily obtained by adopting the proposed algorithm. It ensures that the shunt APF can very well achieve different compensation purposes. Moreover, it is very easy to implement this algorithm in a digital signal processor (DSP). Simulation results obtained with MATLAB and testing results on an experimental shunt APF are presented to validate the proposed algorithm. Index Terms—Active power filter (APF), current-detection algorithm, harmonic compensation, reactive power compensation, time domain. Fig. 1. SAPF control system. to some situation but not to all situation. A novel current-detection algorithm of SAPF for harmonic elimination, power factor correction, and balancing of nonlinear loads is proposed in this paper. II. TRADITIONAL AND PROPOSED CURRENT DETECTION ALGORITHMS There are many existing methods of current-detection algorithms. I. INTRODUCTION W ITH the proliferation of nonlinear loads such as diode/thyristor rectifiers, nonsinusoidal currents degrade power quality in power transmission/distribution systems [1]. Traditionally, passive filters have been used to attenuate the harmonic distortion and compensate the reactive power, but passive filters are bulky, detune with age, and can resonate with the supply impedance. As active power filters are powerful tools for the compensation not only of current harmonics produced by distorting loads but also of reactive power and unbalance of nonlinear and fluctuating loads [2], and they can be smaller, more versatile, better damped, more selective, and less prone to failure for component drift than its passive counterpart, they are studied widely, and great developments have taken place in theory and application of active power filters (APFs) [3]. The shunt APFs (SAPFs) are used most widely [4]. Fig. 1 shows the topology of SAPF, which is used to cancel the current distortion. The performance of SAPF strictly depends on the features of the current-detection algorithms and controllers. Many papers on current-detection algorithms have been published [5]–[17], and some algorithms have been applied in practice. However, usually one algorithm is only more appropriate Manuscript received March 1, 2004; revised July 2, 2004. This work was supported by a grant from the National Natural Science Foundation of China (Key Program no. 59737140). Paper no. TPWRS-00110-2004. The authors are with the Department of Electrical Engineering, Xi’an Jiaotong University, Xi’an, China (e-mail: lihy@mailst.xjtu.edu.cn; zffz@mail.xjtu.edu.cn). Digital Object Identifier 10.1109/TPWRS.2005.846215 A. Analogue Methods Analog filter has been used to detect the harmonics [5], [6], such as the band-pass filter and low-pass filter. The main disadvantage with this method is that the derived component has magnitude and phase errors; in other words, the precision is not satisfactory. B. Instantaneous Reactive Power Theory-Based Methods Instantaneous Reactive Power Theory (IRPT) (i.e., p-q theory), which was proposed by H. Akagi in 1983, promoted the use of APF [4]. This approach inspired the development of many other - theory-based methods for realizing the SAPF [7]–[9]. However, the IRPT-based method requires coordinate coordinates and the transformations between the coordinates, which increases the complexity of designing the SAPF controller. In 1998, Peng et al. proposed a theory that gave a general definition of the instantaneous reactive power in coordinate [10]. Although Peng’s approach does not the need the coordinate transformation, it requires an additional function for instantaneous reactive power vector calculation in the SAPF controller. Moreover, all of the above algorithms based on IRPT cannot be used to detect the selective order harmonics. C. Fourier Transform-Based Methods With the wide use of digital signal processor (DSP) and microcomputer, digital algorithms such as fast Fourier transform 0885-8950/$20.00 © 2005 IEEE LI et al.: A NOVEL TIME-DOMAIN CURRENT-DETECTION ALGORITHM FOR SHUNT APFs 645 (FFT) and discrete Fourier transform (DFT) began to be studied and used [11], which has good precision and can be used to detect the selective order harmonic, but one complete mains cycle delay is unavoidable for the algorithm and the physical meaning is not clear by the transformations between the time domain and the frequency domain. Although the modified Fourier transform-based methods adopting the sliding window can improve the dynamic response [12], [13], one main cycle is also required to track the load change completely. Therefore, it is suitable for slowly varying load conditions. Moreover, Fourier transform-based methods cannot be used to split the fundamental current up into the positive-sequence/negative-sequence component and active/reactive power component of the positive-sequence fundamental current. D. Other Methods Some other algorithms such as fuzzy control [14], adaptive [15], [16], and the wavelet and neural network [17], [18] algorithm have also been applied, which are quite accurate and, of course, have a much better dynamic response than FFT, but these algorithms require a large amount of calculation. Software implementations of signal filters as finite impulse response (FIR) or (IIR) filters can replace the analogue techniques outlined above, but the disadvantage of analog filters is also inherent. Fig. 2. Diagram of the symmetric weigh law. cared, and it is not necessary to decompose the harmonic. Then, the fundamental current component is expressed as follows: III. OPERATION PRINCIPLE OF THE PROPOSED ALGORITHM Based on IRPT and DFT, this paper presents a novel method based on time domain for determining the SAPF reference compensating currents. The operation principle of the proposed algorithm is introduced, and the physical meaning is analyzed in the following section. In the three-phase three-wire system, the instantaneous load , and ) can be disassemcurrents of phase “a,” “b,” “c” ( bled into positive-sequence and negative-sequence components according to the symmetrical weigh law, which was proposed by Fortescue separately. (1) where The subscripts 1 and 2 represent the positive-sequence and negative-sequence, respectively. represents harmonic order. represents the peak value of the current and is the initial phase ( is determined as the initial phase between the positive-sequence fundamental voltage of phase “a” in order to explain is the number of sampling point in the physical meaning). one fundamental cycle, and is the count value of sampling . Commonly, only the positive-sequence, negative-sequence, active power, and reactive power of the fundamental current are (2) where, as shown in Fig. 2, the first item of (2) corresponds to the positive-sequence component in phase with the phase voltage, which is called the active power component of the positive-sequence fundamental current; the second item of (2) corresponds to the positive-sequence component orthogonal with the line voltage, which is called the reactive power component of the positive-sequence fundamental current; the third item of (2) corresponds to the negative-sequence component in phase with the line voltage, which is called the active power component of the negative-sequence fundamental current; the forth item of (2) corresponds to the negative-sequence component orthogonal with the line voltage, which is called the reactive power component of the negative e-sequence fundamental current. Fig. 3 shows the block diagram of the proposed current-deis synchronous with the tection algorithm, where positive-sequence fundamental voltage of phase “a,” which determines the calculation precision of active and reactive power components. The low-pass filter used determines the performance of the system [20]. According to different compensation purposes, different components will be obtained by the segregator expediently, which is superior to the algorithm based on the instantaneous reactive power theory. 646 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 2, MAY 2005 to the total power delivered from source to load). The second item of the equation corresponds to the instantaneous reactive power component of the positive-sequence fundamental (the sum of three phases is zero, which circulates between the phases and can be compensated by a compensator without an energystorage element). The third item corresponds to the instantaneous active power component of the negative-sequence fundamental (the three-phase sum of the previous part of this item is zero, which circulates between the phases and can also be compensated by a compensator without an energy-storage element). The three-phase sum of the rest (including the posterior part of the third and fourth items, which is equal to and the same for each phase) is not zero, and its frequency is twice the fundamental, which can be compensated by a compensator with an energy-storage element). Therefore, the negative-sequence fundamental currents do not contribute to the power delivered to the load. Similarly, the instantaneous power of harmonics can be obtained by multiplying current by Fig. 3. Block diagram of the proposed current-detection algorithm. Fig. 4. Control diagram of the phase locked loop. The synchronous signal can be obtained by soft phase loop lock (SPLL) [19]. The control diagram of the SPLL is shown is set to zero, the calcuin Fig. 4. When the reference lated is synchronous with the positive-sequence component of is not set to zero, a fixed phase fundamental voltage. When and the positive-sequence comdifference is between the ponent of fundamental voltage, which make the control of the displacement factor easy. Moreover, this phase difference will not affect the validity of the selected harmonics detection. The can be obtained as . A “table look-up” scheme is employed to get , and for calculation, which has been calculated and stored in a data table to reduce processing time. The phase voltage is expressed by per-unit; the base quantities for per-unit value is the peak value of positive-sequence fundamental phase voltage. Then, three phase voltages can be , respectively. The instantaexpressed as neous power of the fundamental current can be obtained by multiplying current by phase voltage (4) It can be seen that either the negative-sequence or positive-sequence component has one part of which the three-phase sum up to zero, which can be compensated by a compensator without energy storage, and the rest can be compensated by a compensator with energy storage. In (4) and (5), the lowest frequency component is twice the fundamental frequency; the dc component can be obtained by a low-pass filter with a cutoff frequency lower than twice the funsamples. damental frequency or by a sliding-window with Then, multiplying by 2, the following equation can be obtained: (5) Multiplying (1) by tion can be obtained: , the following equa- (3) where the first item of the equation corresponds to the instantaneous active power component of the positive-sequence fundamental (the sum of three phases is constant, which contributes Using the same method, the following equations can also be obtained: (6) LI et al.: A NOVEL TIME-DOMAIN CURRENT-DETECTION ALGORITHM FOR SHUNT APFs 647 Here, define (7) where and are the peak values of the reactive power component and active power component of positive-sequence and are the peak fundamental current, respectively. values of the reactive power component and active power component of negative-sequence fundamental current of phase “a,” respectively. According to phase “b” and “c,” the peak values and are , respectively. and Similarly, multiplying (1) by , respectively, and going through the low-pass filter, and can be obtained: Then, define (8) (9) (10) (11) Equations (7)–(10) compose the segregator shown in Fig. 1, by which various results can be achieved according to different compensation purposes. If the SAPF is used to compensate harmonics and the negative-sequence component of the fundamental current, the positive-sequence component of the can be obtained from (8), and the fundamental current current reference can be obtained as by subtracting from the load current. If the line current after compensation is expected to be a symmetrical three-phase fundamental current, and the power factor is 1, the active power component of the positive-sequence fundamental current can be obtained by assuming in (8), and the current by subreference can be obtained as from the load current. Similarly, by setting tracting to zero, the reactive power component of the positive-sequence fundamental current can be obtained. The negative-sequence Fig. 5. Simulated three-phase voltage and load current. Fig. 6. Estimated fundamental current by IRPT and the proposed algorithm. component of the fundamental current can be obtained by (9). If the APF is used to compensate the selected order harmonics, the compensating reference can be obtained by (11). In fact, the “active power” component and “reactive power” component of harmonics do not need to be divided, so the factors and can be and , separately. replaced with Then, the programming can be greatly simplified. Based on the detection methods, different compensation aims can be achieved by using specific combinations. It can be seen from the above analysis that the delay resulting from the proposed algorithm is less than half of the main cycle, which is half of that of DFT and the same as that of the algorithm based on IRPT. Besides, the algorithm proposed could detect the positive/negative-sequence fundamental current, active/reactive power component of positive-sequence fundamental current, and selective harmonics expediently, which is more flexible than the algorithm based on IRPT and DFT. Moreover, the proposed algorithm has clear physical meaning and does not need the coordinate transformation. IV. SIMULATION RESULTS In order to test the operation of the novel current-detection algorithm, the simulation model has been built. The simulation was performed by MATLAB dynamic simulation tool—Simulink. A. Steady-State Conditions Simulation The test system for steady-state conditions consisted of a three-phase diode rectifier with a 4 ohm dc-side resistor and a 1.8 mH ac-side inductor in phase “b” branch. Fig. 5 shows waveforms of three-phase voltage, which is symmetrical, and load current, which is nonsinusoidal and unbalanced. Figs. 6–9 show waveforms of the derived fundamental current, which include both the positive-sequence and negative-sequence components; the derived positive-sequence fundamental current, where both active-power and reactive-power components exist; the derived active-power component of the 648 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 2, MAY 2005 Fig. 7. Estimated positive-sequence fundamental current by IRPT and the proposed algorithm. Fig. 12. Estimated fundamental current by DFT and the error. Fig. 8. Estimated active power component of positive-sequence fundamental current by IRPT and the proposed algorithm. Fig. 13. Estimated fundamental current by instantaneous reactive power theory and the error. Fig. 9. Estimated negative-sequence fundamental current by IRPT and the proposed algorithm. Fig. 14. error. Estimated fundamental current by the proposed algorithm and the Fig. 10. Estimated fifth-order harmonics by DFT and the proposed algorithm. Fig. 15. Difference of the estimated fundamental current between the instantaneous reactive power theory and proposed algorithm. Fig. 11. Simulated load current and the fundamental component. positive-sequence fundamental current; the derived negative-sequence fundamental current, which when obtained by the proposed algorithm is the same as that obtained by the IRPT-based algorithm [4]. As shown in Fig. 10, the firth-order harmonics obtained by the proposed algorithm and DFT [11] are also the same. Therefore, the proposed algorithm has the functions of both IRPT and DFT, which is more flexible. B. Dynamic Response Simulation In order to compare the dynamic response of the DFT, IRPT, and the proposed algorithm, a different test system for transient experiments is built, which consists of a three-phase diode bridge rectifier with a 20–10 ohm varying dc-side resistor; negative-sequence fundamental voltage that is 10% of the positive-sequence fundamental voltage is added. Fig. 11 shows the load current and the fundamental component. Figs. 12–15 show estimated the fundamental current and the error by adopting different algorithms. The sliding-window average is adopted to obtain the dc component in IRPT and the proposed algorithm. As shown in Fig. 9, there is one main cycle delay in the estimated fundamental current by adopting DFT. This leads to a residual fundamental component in the reference current, which, in turn, causes a large real power flow in the APF. However, the delay by adopting the algorithm based on IRPT and proposed algorithm is half of one main cycle, and the estimated fundamental current error is smaller than adopting DFT. As shown in Fig. 15, the response time and precision of the proposed algorithm and the algorithm based on IRPT are almost the same. Fig. 16 shows the firth-order harmonics of the load current. Figs. 17 and 18 show estimated fifth-order harmonics and the detection error by adopting DFT and the proposed algorithm, respectively. As shown in these two figures, the precision in steady state by adopting both algorithms are the same; however, the proposed algorithm has better character of dynamic response than DFT, whose response time is half of that of DFT, as pointed out in the analysis in Section III. LI et al.: A NOVEL TIME-DOMAIN CURRENT-DETECTION ALGORITHM FOR SHUNT APFs Fig. 16. Fig. 17. 649 Fifth-order harmonics. Fig. 20. Phase voltage and line current before compensation. Fig. 21. Phase voltage and line current after harmonics is compensated. Estimated fifth-order harmonics by adopting the DFT and the error. Fig. 18. Estimated fifth-order harmonics by adopting the proposed algorithm and the error. Fig. 22. Phase voltage and line current after compensation, where only the active-power component of positive-sequence fundamental current remains. Fig. 19. Configuration of experimental system. V. EXPERIMENTAL VERIFICATIONS An experimental system was constructed in order to demonstrate the operation of the proposed algorithm. Fig. 19 shows the configuration of the experimental system. The objectives are to show that the proposed algorithm has better performance than conventional algorithms, such as the instantaneous reactive power theory-based algorithm and the DFT-based algorithm, and can be implemented well in real time. mented by analog circuit instead of digital circuit to avoid using too many A/D converters. In the presented control system, the number of D/A converters is always three, no matter how many models exist in the APF system, because the current instructions of the models are the same. So, it is easier to expand the power rating of APF through the digital-analog hybrid control method. The load current is sampled 500 times in a fundamental cycle by three rapid 12-bit parallel A/D converters (AD7892). The current references calculated by DSP are given to the analog tracking circuit via a rapid parallel DAC- MAX547, which is 13 bits and has eight channels. The tri-carrying current tracking is adopted in current control in order to achieve the high performance of multiplex well. B. Steady-State Conditions Experiments A. Configuration of APF A 100-kVA SAPF is developed [21], which is multiplex and composed of two models. The detection algorithm proposed in this paper is adopted in the APF system and complemented by a 32-bit floating point DSP—TMS320C32, by which the precision of calculation can be ensured. The current control is imple- The test system for steady-state experiments consisted of a three-phase diode bridge rectifier with a 2 ohm dc-side resistor and a 1.8 mH ac-side inductor in phase “b” branch. Fig. 20 shows the phase voltage and the line current before compensation, which is nonsinusoidal and unbalanced and has a reactive power component. Figs. 21–24 show experimental re- 650 Fig. 23. Phase voltage and line current after compensation, where only the positive-sequence fundamental current remains. IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 20, NO. 2, MAY 2005 Fig. 26. Dynamic response of APF adopting the proposed algorithm when all harmonics below 25th are compensated (time base 40 ms/div). dynamic response, and the THD of the line current after compensation is lower than 4% in the transient process. VI. CONCLUSION Fig. 24. Phase voltage and line current after fifth-order harmonics is compensated. A novel current-detection algorithm based on time domain is proposed in this paper. The algorithm is realized by DSP. From the analysis, simulation, and experiment, we can see that the algorithm presented in this paper has some advantages: 1) clear physical meaning, 2) flexible, 3) rapid, 4) accurate, and 5) easy to implement. On this basis, a 100-kVA APF is developed, and the research of simulation and experiment is also done. The results of the simulation indicate that this algorithm can detect the active power, reactive power, positive-sequence, negative-sequence, active component of positive-sequence current, and each selected harmonics below 25th expediently and quickly. The experimental results indicate that the APF adopting the new algorithm performs well and can be used in practice. REFERENCES Fig. 25. Dynamic response of SAPF adopting DFT when all harmonics below 25th are compensated (time base 40 ms/div). sults according to different compensation purposes, where is phase voltage, is line current, and time base is 10 ms/div. As shown in these waveforms, the steady-state performance of APF adopting the proposed algorithm is perfect. C. Dynamic Response Experiments In Figs. 25 and 26, the dynamic response of the APF adopting the DFT or the proposed algorithm are shown, respectively, is the load where is the line current after compensation, current, and is the reference for compensating (amount of fundamental component is composed in the reference in order to control the dc voltage of the APF, and this part does not change with the load current varying). As shown in Fig. 25, there is one main cycle delay in the reference identifying. Therefore, the THD of the line current after compensation is higher than 10% in the transient process. However, as shown in Fig. 26, the APF adopting the proposed algorithm has good [1] P. Jintakosonwit, H. Fujita, and H. Akagi, “Control and performance of a fully-digital-controlled shunt active filter for installation on a power distribution system,” IEEE Trans. Power Electron., vol. 17, no. 1, pp. 132–140, Jan. 2002. [2] P. Mattavelli, “A closed-loop selective harmonic compensation for active filters,” IEEE Trans. Ind. Appl., vol. 37, no. 1, pp. 81–89, Jan./Feb. 2001. [3] W. Zhaoan, Y. Jun, and L. Jinjun, Harmonics Elimination and Reactive Power Compensation. Beijing, China: China Machine Press, 1998. [4] H. Akagi, “New trends in active filters for power conditioning,” IEEE Trans. Ind. Appl., vol. 32, no. 3, pp. 1312–1322, May/Jun. 1996. [5] W. M. Grady, M. J. Samotj, and A. H. Noyola, “Survey of active power line conditioning methodologies,” IEEE Trans. Power Del., vol. 5, no. 3, pp. 1536–1542, Jul. 1990. [6] L. A. Moran, J. W. Dixon, and R. R. Wallace, “A three-phase active power filter operating with fixed switching frequency for reactive power and current harmonic compensation,” IEEE Trans. Ind. Electron., vol. 42, no. 4, pp. 402–408, Aug. 1995. [7] H. Akagi, Y. Kanazawa, and A. Nabae, “Generalized theory of the instantaneous reactive power in three-phase circuits,,” in Proc. IEEJ Int. Power Electron. Conf., Tokyo, Japan, 1983, pp. 1375–1386. [8] , “Instantaneous reactive power compensators comprising switching devices without energy storage components,” IEEE Trans. Ind. Appl., vol. 20, no. 2, pp. 625–630, Mar./Apr. 1984. [9] J. L. Willems, “A new interpretation of the Akagi-Nabae power components for nonsinusoidal three-phase situations,” IEEE Trans. Instrum. Meas., vol. 41, no. 4, pp. 523–5297, Aug. 1992. [10] F. Z. Peng and J. S. Lai, “Generalized instantaneous reactive power theory for three-phase power systems,” IEEE Trans. Instrum. Meas., vol. 45, no. 1, pp. 293–297, Feb. 1996. [11] A. A. Girgis, W. B. Chang, and E. B. Makram, “A digital recursive measurement scheme for on-Line tracking of power system harmonics,” IEEE Trans. Power Del., vol. 3, pp. 1153–1160, Jul. 1991. LI et al.: A NOVEL TIME-DOMAIN CURRENT-DETECTION ALGORITHM FOR SHUNT APFs [12] S. Luo, Z. Lu, and Y. Zhou, “A novel real-time harmonics and reactive power current detection method,” Chongqing Univ. Trans., vol. 14, no. 6, pp. 28–31, Jun. 1991. [13] M. EI-Habrouk and M. K. Darwish, “Design and implementation of a modified Fourier analysis harmonic current computation technique for power active filters using DSP’s,,” Proc. Inst. Elect. Eng.—Elect. Power Appl., vol. 148, no. 1, pp. 21–28, Jan. 2001. [14] S. K. Jain, P. Agrawal, and H. O. Gupta, “Fuzzy logic controlled shunt active power filter for power quality improvement,” Proc. Inst. Elect. Eng.—Elect. Power Appl., vol. 149, no. 5, pp. 317–328, Sep. 2002. [15] S. Liu, “An adaptive Kalman filter for dynamic estimation of harmonic signals,” in Proc. 8th Int. Conf. Harmonics Quality Power, Athens, Greece, 1998, pp. 286–292. [16] S. Luo and Z. Hou, “An adaptive detecting method for harmonic and reactive currents,” IEEE Trans. Ind. Electron., vol. 42, no. 1, pp. 85–89, Feb. 1995. [17] B. Lin and R. G. Hoft, “Power electronics inverter control with neural networks,” IEEE Technol. Update Series, Neural Netw. Appl., pp. 211–217, 1996. [18] J. R. Varquez and P. Saimeron, “Active power filter control using neural network technologieds,” Proc. Inst. Elect. Eng.—Elect. Power Appl., vol. 150, no. 2, pp. 139–145, Mar. 2003. [19] V. Kaura and V. Blasko, “Operation of a phase locked loop system under distorted utility conditions,” IEEE Trans. Ind. Appl., vol. 33, no. 3, pp. 58–63, May/Jun. 1997. [20] Q. Wang, W. Yao, and Z. Wang, “A study about influence of high and low pass filter on detecting effect of harmonics detection circuits,” Trans. China Elect. Tech. Soc., vol. 14, no. 3, pp. 22–26, May 1997. [21] H. Li, F. Zhuo, and W. Lei, “Control system of the multiple large power rate active power filter,” Adv. Tech. Elect. Eng. Energy, vol. 23, no. 1, pp. 25–28, Jan. 2004. Fang Zhuo was born in Shanghai, China, on May 20, 1962. He received the B.S., M.S., and Ph.D. degrees from Xi’an Jiaotong University, Xi’an, China, in 1984, 1989, and 2001, respectively. Beginning in 1984, he was a Lecturer with Xi’an Jiaotong University, where he now is an Associate Professor. He is engaged in research on power electronics, motor driver control, active power filters, and power quality. Zhaoan Wang (SM’98) was born in Xi’an, China, on June 9, 1945. He received the B.S. and M.S. degrees from Xi’an Jiaotong University, Xi’an, China, in 1970 and 1982, respectively, and the Ph.D. degree from Osaka University, Osaka, Japan, in 1989. From 1970 to 1979, he was an Engineer at Xi’an Rectifier Factory. He was an Associate Professor at Xi’an Jiaotong University, where he is currently a Professor. He is engaged in research on power conversion system, harmonics suppression and reactive power compensation, and active power filters. Wanjun Lei was born in Shannxi, China, in 1978. He received the B.S. and M.S. degrees from Xi’an Jiaotong University, Xi’an, China, in 2000 and 2004, respectively. He is currently working toward the Ph.D. degree from the Department of Electrical Engineering, Xi’an Jiaotong University. His research interests are active power filter and power quality improvement. Hongyu Li was born in Shandong province, China, in 1978. He received the B.Sc. and M.Sc. degrees from Xi’an Jiaotong University, Xi’an, China, in 1999 and 2002, respectively. Since 2002, he has been pursuing the Ph.D. degree in the Department of Electrical Engineering, Xi’an Jiaotong University. His research interests are active power filters and power quality improvement. 返 651 Longhui Wu was born in Shandong, China, in 1981. He received the B.S. and M.S. degrees from Xi’an Jiaotong University, Xi’an, China, in 2002 and 2005, respectively. He is currently working toward the Ph.D. degree in the Department of Electrical Engineering, Xi’an Jiaotong University. His research interests are active power filter and power quality improvement. 回