Amercan Journal of Appled Scences 5 (8): 1064-1070, 008 ISSN 1546-939 008 Scence Publcatons Modelng and Control of Three-Phase Shunt Acte Power Flter 1 Moleykutty George and Kartk Prasad Basu 1 Faculty of Engneerng and Technology, Multmeda Unersty, 75450 Melaka, Malaysa Faculty of Engneerng, Multmeda Unersty, 63100 Cyberjaya, Malaysa Abstract: In ths research the performance of a three-phase shunt acte power flter (APF) usng Model Reference Controller (MRC) has been compared wth that usng nstantaneous acte and reacte (p-q) theory. The noelty of ths research les n the applcaton of MRC to generate the ampltude of the reference supply current requred by the APF crcut and the successful mplementaton of the APF system for harmonc elmnaton. The entre system has been modeled usng MATLAB 6.1 toolbox. Smulaton results demonstrate the applcablty of MRC for the control of APF. Key words: APF, MRC, VSI, harmoncs, THD, p-q theory, power qualty INTRODUCTION The major causes of power qualty problems are due to the wde spread applcaton of statc power electronc conerters, saturable deces, fluorescent lamps and arch furnaces. Some of the aderse effects of poor power qualty are reduced motor lfe, ncreased losses, mal-operaton, electromagnetc nterference, ncreased heatng, and faulty tmng sgnals. Een though there are no standard waeforms for the purpose of specfyng power qualty problems, IEEE standard, Amercan natonal standard gudes (ANSI), Brtsh standards (BS), European norms (EN), etc. are wdely followed to mantan electrcal power qualty. The IEEE standard 519 s a recommended practce for power factor correcton and harmonc mpact lmtaton at statc power conerters. IEEE-519 standard lmts the total harmonc dstorton (THD) of oltage and current below 5 %. Acte power lne condtoners hae been proposed for harmonc elmnaton and power factor mproement [1-11], cancellaton of negate and zero sequence components [1-15], oltage sag and swell [16]. Many conentonal control strateges hae been proposed and mplemented for the successful control of APF system. Recent research shows the effecteness of artfcal ntellgent (AI) based controllers such as fuzzy logc controller and neural network controllers for the control of APF system [17-5]. Ths research proposes MRC for the control of APF system. The noelty of ths research les n the applcaton of MRC for the determnaton of ampltude of reference supply current requred n an APF system. Ths research also dscusses the control of APF system usng p-q theory. The control strateges of APF system are detaled n the second part of ths research. Smulaton results n the thrd part demonstrate the effecteness of MRC for the control of APF system. CONTROL TECHNIQUES OF APF SYSTEM In ths part of study control scheme of APF system usng p-q theory s dscussed and compared wth that usng MRC. Prncple of operaton: A three-phase system feedng an nerter load has been to study the performance of the APF system. It has been obsered that due to the non-lnear characterstcs of power electroncs loads the THD s of source current and termnal oltage fall well below the IEEE-519 standard and n prncple APF system s used to nject a current equal n magntude but n phase opposton to harmonc current to achee a purely snusodal current wae nphase wth the supply oltage. Fgure 1 [6] shows the sngle-lne dagram of a smple power system wth APF system ON. The heart of the APF system s the IGBT based oltage source nerter (VSI). A dc capactor s used to deler power for the VSI. For the successful operaton of APF, capactor oltage should be at least 150 % of maxmum lne-lne supply oltage. Snce the PWM VSI s assumed to be nstantaneous and nfntely fast to track the compensaton currents, t s modeled as a current amplfer wth unty gan and the Correspondng Author: Moleykutty George, Faulty of Engneerng and Technology, Melaka Campus, Multmeda Unersty, 75450, Melaka, Malaysa Tel: +606 5386 1064
Am. J. Appled Sc., 5 (8): 1064-1070, 008 s(t) l(t) Accordng to p-q theory, determnaton of nstantaneous real power p l (t) and magnary power q l (t) s gen by the expresson pl sα ql s β sβ sα lα lβ APF ON Nonlnear Load where p l (t) and q l (t) contan dc and ac terms and can be wrtten as p l (t) p ~ + p q l (t) q + q ~ Fg. 1: Sngle-lne dagram of a smple power system wth APF ON To achee unty power factor and harmonc elmnaton, the ac term p and the magnary power q l (t) hae to be elmnated. The compensaton power p could be obtaned by flterng out the ac components from p l(t). Thus p c q c p~ and q l (t) The reference compensaton current n the (α-β) plane s gen by the expresson Fg. : Inerter sub-system Smulnk model of the nerter sub system used s shown n Fg. [7]. Control of APF system usng p-q theory: The p-q theory proposed by Akag [1] to determne the compensaton current to be njected by the APF system for harmonc elmnaton and reacte power uses Park s transformaton from three-phases (a,b,c) to two phases (αandβ). Thus the three phase supply oltages and load currents could be transformed nto the (α-β) orthogonal coordnates as follows: sα sβ 1 3 0 1 3 1 3 la(t) lα (t) 1 1/ 1/ lb(t) lβ(t) 3 0 3 / 3 / lc (t) sa sb sc 1065 1 c (t) α sα(t) sβ(t) p c(t) s (t) s (t) cβ(t) β α q c(t) and the reference compensaton currents for phase a, phase b and phase c could be ealuated usng Park s backward transformaton and gen n matrx form as follows: ca cb cc 3 1 1 1 0 3 3 cα cβ Fgure 3 shows the smulnk model of APF control system usng p-q theory. Control of APF system usng MRC: MRC s successfully used for the control of UPQC [8]. Fg. 4 shows the smulnk model of MRC controlled APF system, whereas n Fg. 5 the detaled control structure of APF system usng MRC s llustrated. Model reference controller uses two neural networks: a plant model network and a controller network. To tran the controller, frst of all neural network plant model shown n Fg. 6 has been dentfed and traned. Followng that
Am. J. Appled Sc., 5 (8): 1064-1070, 008 Fg. 3: Smulnk model of APF system usng p-q theory Fg. 4: APF system usng MRC tranng data has been generated usng a smulnk reference model. The Leenberg-Marquadrat algorthm s used for tranng the neural network plant model. The controller tranng s computatonally expense and tme consumng as t requres dynamc back propagaton. The BFGS (Broyden, Fletcher, Goldfarb, and Shanno) tranng algorthm was used to tran the controller. As shown n Fg. 4, the control nput of the plant s the ampltude of the desred mans current and the capactor oltage s the plant output. MRC checks the desred capactor oltage and the actual capactor oltage and the control nput s adjusted to achee the reference alue. Table 1 and show the specfcatons of the plant model network and controller network. It has been obsered that the complcated equatons n p-q 1066
Am. J. Appled Sc., 5 (8): 1064-1070, 008 c L - + sr Multpler MRC - V dc I sm + V dcr PLL ( Unt ampltude generaton, V a n, V b n, & V cn ) Fg. 5: Control crcut usng MRC Fg. 6: Plant model Table 1: Plant model specfcatons Sze of hdden layer 1 Samplng nteral (s) 6.54e -5 No. of delayed plant nputs No. of delayed plant outputs 1 Tranng samples 50000 Maxmum plant nput 1.8 Mnmum plant nput 1.5 Maxmum nteral alue (s) 0.05 Mnmum nteral alue (s) 0.01 Maxmum plant output 800 Mnmum plant output 700 Tranng Epochs 100 Tranng Functon tranlm Use current weghts Use aldaton data Use testng data Table : Specfcatons of Model reference control Sze of hdden layer 1 Samplng nteral (s) 6.54e -5 No. delayed reference nputs 1 No. delayed controller outputs 1 No. delayed plant outputs 1 Tranng samples 50000 Maxmum plant nput 1.8 Mnmum plant nput 1.5 Maxmum nteral alue (s) 0.05 Mnmum nteral alue (s) 0.01 Tranng Epochs 10 Controller tranng segments Use current weghts Use cumulate tranng un 1067 Fg. 7: Performance graph Fg. 8: Tranng data theory could be elmnated by the use of MRC. Sample performance graph, tranng data, reference model and neural network outputs obtaned are llustrated n Fg. 7 and Fg. 8 and 9 respectely. RESULTS AND DISCUSSION An APF system based on MRC has been successfully modeled and tested usng MATLAB 6.1 toolbox. The effecteness of the system has been tested for arous frng angles (α) n the range of 0 and 180.
Am. J. Appled Sc., 5 (8): 1064-1070, 008 Fg. 9: Reference model and neural network outputs Fg. 13: Frequency spectrum of source current after compensaton usng MRC based APF system Fg. 10: Plots of sa(t), sa(t), la(t), ca(t) usng MRC Fg. 14: Frequency spectrum of source current after compensaton usng p-q theory based APF system Fg. 11: Plots of sa(t), sa(t), la(t), ca(t) usng p-q theory Fg. 15: THD of source and load currents after compensaton usng MRC Fg. 1: Frequency spectrum of load current The performance of the deeloped system s llustrated wth the one usng p-q theory for α 165 as shown n Fg.10-15. It has been obsered that usng p-q theory, for α 165, as the load s n the nerter mode, the source currents are 180 out of phase wth the respecte supply oltages. Howeer usng MRC source 1068
Am. J. Appled Sc., 5 (8): 1064-1070, 008 currents are n-phase wth the supply oltages. One may obsere that n MRC based APF system, reference source currents are obtaned by multplyng the requred ampltude of the source currents wth the unt ampltude waeform n-phase wth the supply oltages. CONCLUSION An MRC based APF system has been modeled and successfully tested for the control of APF. The noelty of ths research les n the applcaton of MRC to determne the ampltude of the reference source current requred n an APF system. Ths research also dscusses modelng and control of APF system usng p-q theory. The performance of the dfferent system has been compared. It has been obsered that the complcated calculatons used n p-q theory could be elmnated by the use of MRC. ACKNOWLEDGMENT The authors are grateful to Alan Tan Wee Chat, Dr. Abdulazeez S. Boujarwah, Chrsto George, K.S. Suresh Kumar, Prof. K. Ramar and Dr. Saurabh Kumar Mukerj for aluable dscussons. REFERENCES 1. Akag, H., 1994. Trends n Acte Power Lne Condtoners: IEEE Tran. Power Electroncs, 9(3): 63-68.. Nastran, J., R. Cajhen, M. Selger and P. Jereb, 1994. Acte Power Flter for Nonlnear AC Loads: IEEE Trans. Power Electroncs, 9(1): 9-96. 3. E. Destobbeleer and L. Protn, 1996. On the Detecton of Load Acte Currents for Acte Flter Control: IEEE Trans. Power Electroncs, 11(6): 768-775. 4. Aredes, M. J. Hafner and K. Heumann, 1997. Three-Phase Four-Wre Shunt Acte Flter Control Strateges: IEEE Trans. Power Electroncs, 1(): 311-318. 5. Fujta, H. and H. Akag, 1998. The Unfed Power Qualty Condtoner: The Integraton of Seres- and Shunt-Acte Flters: IEEE Tran. Power Electroncs, 13(): 315-3. 6. Peng, F.Z., W. George, Jr. Ott and J. Donald Adams, 1998. Harmonc and Reacte Power Compensaton Based on the Generalzed Instantaneous Reacte Power Theory for Three- Phase Four-Wre Systems: IEEE Trans. Power Electroncs, 13(6): 1174-1181. 7. Chatterjee, K., B.G. Fernandes and K. Gopal Dubey, 1999. An Instantaneous Reacte Volt- Ampere Compensator and Harmonc Suppressor System: IEEE Trans. Power Electroncs, 14(): 381-39. 8. Cheng, P.T., S. Bhattacharya and D.D. Dan, 1999. Lne Harmoncs Reducton n Hgh-Power Systems Usng Square-Wae Inerters-Based Domnant Harmonc Acte Flter: IEEE Trans. Power Electroncs, 14(): 65-7. 9. Huang S.J. and J.C. Wu, 1999. A Control Algorthm for Three-Phase Three-Wred Acte Power Flters Under Nondeal Mans Voltages: IEEE Trans. Power Electroncs, 14(4): 753-760. 10. Chandra, A., B. Sngh, B.N. Sngh and K.A. Haddad, 000. An Improed Control Algorthm of Shunt Acte Flter for Voltage Regulaton, Harmonc Elmnaton, Power-factor Correcton, and Balancng of Nonlnear loads: IEEE Trans. Power Electroncs, 15(3): 495-507. 11. George, M., 004. Modelng and smulaton of a current controlled three-phase shunt acte power flter usng MATLAB/PSB: AIUB Journal of Scence and Engneerng, 3(1): 11-18. 1. M. George and C.L. Seen, 004. Modelng and control of zero-sequence current of parallel threephase conerters usng Matlab/power system blockset: IEEE Power Systems Conf. and Exp. 004, 3: 1440-1443. 13. Bhaaraju, V.B. and P.N. Enjet, 1993. Analyss and Desgn of an Acte Power flter for Balancng Unbalanced Loads: IEEE Trans. Power Electroncs, 8(4): 640-647. 14. Bhaaraju, V.B. and P.N. Enjet, 1996. An Acte Lne Condtoner to Balance Voltages n a Three- Phase System: IEEE Trans. Industry Applcatons, 3(): 87-9. 15. Enjet, P.N., W. Shreen, P. Packebush and I.J. Ptel, 1994. Analyss and Desgn of a New Acte Power Flter to Cancel Neutral Current Harmoncs n Three-Phase Four-Wre Electrc Dstrbuton Systems, IEEE Trans. Industry Applcatons, 30(6): 1565-157. 16. Kumar, S.V.R., S.S. Nagaraju, 007. Smulaton of D-STATCOM and DVR n power systems: ARPN J. Eng. Appled Sc., (3): 7-13. 17. Dxon, J.W., J.M. Contardo and L.A. Moran, 1999. A Fuzzy-controlled acte front-end rectfer wth current harmonc flterng characterstcs and mnmum sensng arables: IEEE Tran. Power Electroncs, 14(4): 74-79. 1069
Am. J. Appled Sc., 5 (8): 1064-1070, 008 18. Elmtwally, A., M.S. Kandl and M. Elkateb, 000. A fuzzy-controlled ersatle system for harmoncs unbalance and oltage sag compensaton: IEEE Power engneerng socety summer meetng, 3:1439-1444. 19. Dell Aqula, A., G. Delno, M. Lserre and P. Zanchetta, 000. A new fuzzy logc strategy for acte power flter: IEE Power electroncs and arable speed dres conf. pub., 475: 39-397. 0. A. Dell Aqula, A. Lecc, M. Lserre, and P. Zanchetta, 000. Desgn of the optmum duty cycle for a fuzzy controlled acte flter: Proceedngs of the 000 IEEE Intl. Symposum on Industral Electroncs, 1: 78-83. 1. Jan, S.K., P. Agarwal and H.O. Gupta, 00. Fuzzy logc controlled shunt acte power flter for power qualty mproement: IEE proceedngs on Electrcal power applcatons, 149(5): 317-38.. Round S.D. and N. Mohan, 1993. Comparson of frequency and tme doman neural network controllers for an acte power flter: Intl. Conference Industral Electroncs, Control, and Instrumentaton, : 1099-1104. 3. Karabag Y., T. Erfdan, S. Urgun and N. Abut, 004. Artfcal neural network based hysteress current controller for sngle-phase nerter: 1th IEEE Medterranean Electrotechncal Conference, 1: 339-341. 4. Kawagosh T., A. Kumamoto, T. Hkhara, Y. Hrane, K. Oku, O. Nakamura, S. Tada, K. Mzuk and Y. Inoue, 1993. Harmonc oltage suppresson by acte flter wth neural network controller: Second Intl. Forum on Applcatons Neural Networks to Power Systems: 93-98. 5. Elmtwally, A., S. Abdelkader and M. EI-Kateb, 000. Neural network controlled three-phase fourwre shunt acte power flter: IEE proceedngs on Generaton, Transmsson and Dstrbuton, 147(): 87-9. 6. http://www.elec.canterbury.ac.nz/research/ powerelectroncs/acte_flter.htm 7. Suresh Kumar K.S, 004, Three-phase shunt acte power flters, part III-Smulaton of control systems: Selected topcs n power qualty and custom power, pp: 3. 8. George, M., 007. Artfcal ntellgence based three-phase unfed power qualty condtoner: J. computer sc., 3(7): 465-477. 1070