Design of UPQC by Optimizing PI Controller using GA and PSO for Improvement of Power Quality

Transcription

1 Desgn of UPQC by Optmzng PI Controller usng GA and PSO for Improvement of Power Qualty T.Gunasekar, Dr.R.Anta Abstract Ths paper presents a new control desgn of an Unfed Power Qualty Condtoner (UPQC). The DC lnk capactor of UPQC s controlled by usng PI controller. The conventonal method of obtanng the coeffcents of proportonal plus ntegral (PI) controllers for the actve power flter utlzes a lnear model of the PWM nverter. Ths paper presents a new algorthm based genetc algorthm and Partcle Swam Optmzaton to optmze the coeffcents of the PI controller. Through the smulaton results, t s observed that the dynamc response of the Partcle Swarm Optmzaton algorthm PI (PSO PI) controller s qute satsfactory. A smulaton study of the proposed topology has been carred out usng MATLAB/Smulnk and the results are dscussed. Index Terms Couplng Transformer, Fuzzy Logc, Genetc algorthm, Hysteress Controller, Phase Locked Loop, Partcle Swam Optmzaton, Seres Actve Flter, Shunt actve flter, Synchronous Reference Frame, Dc lnk capactor I. INTRODUCTION In modern electrcal dstrbuton system there has been a sudden ncrease of nonlnear loads, such as power supples, rectfer equpment used n telecommuncaton networks, domestc applances, adjustable speed drves, etc. These power electronc converters are from low power domestc applcatons to hgh power adjustable speed drves (ASDs). Ths power electronc converter generates harmoncs whch ncludes fundamental, thrd, ffth etc and other hgher harmoncs. These harmonc current may cause power qualty degradaton, transformer overheatng, malfunctonng of medcal facltes, destructon of electrc power components, pollute the power system and rotary machne vbraton etc. Many power qualty standards are proposed, such as IEC and IEEE etc. To mantan the harmonc level and the harmoncs can be suppressed by a passve or actve power flter. However the passve flters are sufferng from the dsadvantages such as senstve to the varaton of frequency, system mpedance, and possblty of seres/parallel resonance and fxed flter frequency. Because of seres/parallel resonance t may cause the damage to nductor and capactor of passve power flters. Performance of passve power flter can be affected by the system mpedance and APF s are used to resolve passve flters problems. APF S are shunt actve and seres actve power flter. The shunt actve power flters (APFs) are used to elmnate current harmoncs, load balancng, power factor correcton of three-phase four wre dstrbuton system and the seres actve flters are used to elmnate the voltage Manuscrpt receved Aprl 18, T.Gunasekar, Research Scholar, Anna Unversty, Chenna., Inda Dr.R.Anta, Professor, Department of EEE, IRTT, Erode,TamlNadu, Inda Phone harmoncs. DC Capactors are used to gve the supply for both flters. It s controlled by usng PI controller. Some recently developed evolutonary algorthms, notably genetc algorthms (GAs), Partcal Swarm Optmzaton, make optmal tunng of PI controllers. In ths paper genetc algorthm and Partcle Swarm Optmzaton s used for tunng of PI controller and the Comparson s made. The hysteress band s mplemented wth fuzzy logc to mprove the system performances. The effectveness of the proposed controller s dscussed below. II. BLOCK DIAGRAM DESCRIPTION Fg 1 shows the block dagram of Unfed Power Qualty Condtoner. It conssts of a three-phase source, whch s connected to non-lnear load. The UPQC s connected before the load to make the source and the load voltage free from any dstortons. The UPQC, carred out by usng two VSIs, s shown n Fg. 1.one VSI acts as the shunt APF and the other as the seres APF. The shunt APF s realzed usng a three-phase, three-leg VSI, and the seres APF s carred out usng a three-phase, three-leg VSI. Both APFs share a common dc lnk between them. Fg 1. Block Dagram of UPQC The Seres Actve Power Flter s coupled usng couplng transformer and the shunt actve flter s connected n parallel wth the phases. The proposed control strategy ams to generate reference sgnals for both shunt and seres APFs of the UPQC. The seres actve flter s controlled to elmnate voltage harmoncs and the shunt actve flter s controlled to allevate current from harmoncs and load balancng. The nverter can be mplemented by IGBTs operatng wth the fuzzy hysteress controller for the flterng functon

2 Desgn of UPQC by Optmzng PI Controller usng GA and PSO for Improvement of Power Qualty III. FUNDAMENTAL COMPONENT EXTRACTION There are dfferent control strateges beng used for the calculaton of reference currents n actve power flter namely Instantaneous Reactve Power Theory (p-q theory), Unty Power Factor method, One Cycle Control, Fast Fourer Technque etc. In ths paper, SRF theory s used to extract the three-phase reference currents and voltages used by the actve power flters. Fg.2 shows the block dagram of three-phase SRF-theory, used for harmonc component extracton. The synchronous reference frame theory s used to extract the fundamental component n the supply voltage or current. It s based on the transformaton of the currents or voltages n synchronously rotatng d-q frame. If θ s the transformaton angle, then the current and voltage transformaton from α-β to d-q s defned as n the Fg 2 In ths method, the source currents and voltages are frst detected and transformed nto two-phase statonary frame (αβ-0) from the three-phase statonary frame (a-b-c), as per equaton (1) a b o c (1) ca T cb abc cc o (4) Fnally the current from two phase statonary frame αβ0 s transformed back nto three-phase statonary frame abc as per equaton (4) and the compensaton reference currents ca, cb and cc are obtaned for the shunt actve flter and as same voltage sgnals are obtaned by gvng the voltage sgnal as the nput. IV. FUZZY LOGIC CONTROLLER Fuzzy set theory exhbts mmense potental for effectve solvng of the uncertanty n the problem. It s an outstandng mathematcal tool to handle the uncertanty arsng due to vagueness. Fuzzy logc control s dvded nto Fuzzfcaton, nference and defuzzfcaton whch s shown n Fg.3. Now, the two phase current quanttes α and β of statonary αβ-axes are transformed nto two-phase synchronous (or rotatng) frame (d-q-axes) usng equaton (2), where cosθ and snθ represents the synchronous unt vectors whch can be generated usng phase-locked loop system (PLL). d cos sn sn cos q (2) The d-q currents thus obtaned comprses of AC and DC parts. The fundamental component of current s represented by the fxed DC part and the AC part represents the harmonc component. Ths fundamental component can be easly extracted usng a Low Pass Flter (LPF), as mplemented n Fgure 2. Now nverse transformaton s performed to transform the currents from two phase synchronous frame d-qnto two-phase statonary frame α-β as per equaton(3) Fg 2. SRF Method Fg 3. Fuzzy Inference System The knowledge base s composed of a data base and rule base and s desgned to obtan good dynamc response under uncertanty n process parameters and external dsturbances. The data base consstng of nput and output membershp functons, provdes nformaton for the approprate fuzzfcaton operatons, the nference mechansm and defuzzfcaton. The nference mechansm uses a collecton of lngustc rules to convert the nput condtons nto a fuzzfed Output. Fnally, defuzzfcaton s used to convert the fuzzy outputs nto control sgnals. In desgnng of a fuzzy control system, the formulaton of ts rule set plays a key role n mprovement of the system performance. The mamdan type fuzzy logc controller s used, the max-mn nference method s appled n ths study. cos sn d sn cos q (3) Table 1: Fuzzy Rule Table V. FUZZY HYSTERESIS CONTROLLER The hysteress band current control (HBCC) technque s used for pulse generaton n VSIs shown n Fg

3 The control method offers good stablty, gves a very fast response, provdes good accuracy and has got a smple operaton. It conssts of a hysteress band surroundng the generated error sgnal. The error s obtaned by subtractng the actual sgnal from the reference sgnal. The reference sgnal used here s obtaned by the SRF method. The error sgnal s then fed to the fuzzy and then fed to relay wth the desred hysteress band to obtan the swtchng pulses for the nverter. A genetc algorthm s typcally ntalzed wth a random populaton consstng of between ndvduals. Ths populaton (matng pool) s usually represented by a real valued number or a bnary strng called a chromosome. How well an ndvdual performs a task s measured s assessed by the objectve functon. The objectve functon assgns each ndvdual a correspondng number called ts ftness. The ftness of each chromosome s assessed and a survval of the fttest strategy s appled. The flowchart for GA s shown n Fg 5. The objectve functon s gven as, Fg 4. Hysteress Controller VI. DC LINK VOLTAGE The dc sde voltage of APF should be controlled and kept at a constant value to mantan the normal operaton of the nverter. Because there s energy loss due to conducton and swtchng power losses assocated wth the dodes and IGBTs of the nverter n APF, whch tend to reduce the value of V dc across capactor C dc. A feedback voltage control crcut needs to be ncorporated nto the nverter for ths reason. The dfference between the reference value, V ref and the feedback value (Vdc), an error functon frst passes a PI regulator and the output of the PI regulator s subtracted from the d axs value of the harmonc current components. The PI controller values are tuned usng Genetc Algorthm and Partcle Swam Optmzaton. The DC capactor voltage can be found by usng the equaton. Where C dc, V dc V s, I c L c,r c I co,v dco dc-sde capactor and ts voltage ac-sde voltage and current of PWM nverter Flter nductance and ts resstance steady-state operatng ponts of Ic and Vdc V 2 2 LL V dc 3 (5) VII. GENETIC ALGORITHM The basc prncples of Genetc Algorthm (GA) were frst proposed by Holland. It s nspred by the mechansm of natural selecton where stronger ndvduals would lkely be the wnners n a competng envronment. In ths Genetc Algorthm approach s used to determne the optmzed value of PI controller parameters namely Kp and K. The genetc algorthm starts wth no knowledge of the correct soluton and depends entrely on responses from ts envronment and evoluton operators (.e. reproducton, crossover and mutaton) to arrve at the best soluton. By startng at several ndependent ponts and searchng n parallel, the algorthm avods local mnma and convergng to sub optmal solutons. In ths way, GA has been shown to be capable of locatng hgh performance areas n complex domans wthout experencng the dffcultes assocated wth hgh dmensonalty, as may occur wth gradent descend technques or methods that rely on dervatve nformaton. Fg 5. Flow chart for Genetc Algorthm 198

4 Desgn of UPQC by Optmzng PI Controller usng GA and PSO for Improvement of Power Qualty VIII. PARTICLE SWAM OPTIMIZATION In order to mprove the power qualty further by reducng the THD to great extent, Partcle Swam Optmzaton technque s used to tunng the parameters of the PI controller. PSO s a populaton based stochastc optmzaton technque nspred by socal behavor of brd flockng or fsh schoolng. PSO learns from the scenaro and uses t to solve the optmzaton problems. In PSO, each sngle soluton s a brd n the search space whch s called as partcle. All partcles have ftness values whch are evaluated by the ftness functon to be optmzed, and have veloctes whch drect the flyng of the partcles. The partcles fly through the problem space by followng the current optmum partcles. PSO s ntalzed wth a group of random partcles (solutons) and then searches for optma by updatng generatons. In each teraton, every partcle s updated by followng two best values. The frst one s the best soluton (ftness) t has acheved so far. IX. RESULT AND DISCUSSION The smulaton results wthout flter shown n fgure 5 t shows that the voltage contans harmoncs and the load current contans harmoncs and t s unbalanced. Due to the unbalanced nature the neutral current flow s hgh. MATLAB SIMULINK model for the proposed system s shown n Fg 7.The nstallaton of UPQC compensates the harmoncs and unbalance and the neutral current magntude gets reduced effectvely. The DC lnk Capactor value s mantaned constant usng the GA and PSO. Fg.8 shows the waveform wth GA tunng and Fg.9 shows the waveform wth PSO tuned PI controller. The THD values wthout UPQC, GA tuned and PSO tuned PI controller s shown n Table.2. Ths value s called Pbest. Another best value that s tracked by the partcle swarm optmzer s the best value, obtaned so far by any partcle n the populaton. Ths best value s a global best and called gbest. PSO has the good sharp of fndng the optmal ftness functon and also has proved t effectveness n fndng the optmal K P and K parameters. The flow chart for partcle swam optmzaton s shown n Fg.6. Fg 7. Load Voltage, Load Current and Neutral current waveform wthout UPQC Fg 6. Flow chart for Partcle Swam Optmzaton Fg 8. Load Voltage, Load Current and Neutral current waveform wth GA UPQC 199

5 Load 3: R=50Ω; C = 1000µf Sngle Three phase load: R=10Ω ; C = 1000µf DC lnk voltage: 240V. DC lnk capactance value: 2500µf Rpple flter parameters: L f =1.8mH, R f =0.25Ω. Dc lnk voltage PI controller parameters: K p = , K =9.99.(Tuned usng GA) K p = ,k =10.0 (Tuned usng PSO) REFERENCES Fg 8. Load Voltage, Load Current and Neutral current waveform wth PSOs UPQC Table.2. Comparson of THD GA Vs PSO [1] A. Arabal, M. Ghofran, M. Etezad-Amol, M. S. Fadal, and Y. Baghzouz, Genetc-Algorthm-Based Optmzaton Approach for Energy Management IEEE Transactons On Power Delvery, Vol. 28, No. 1, January 2013 [2] Bhm Sngh, Venkateswarlu.P, A Smplfed Control Algorthm for Three-Phase Four-Wre Unfed Power Qualty Condtoner 2010 [3] Dr.S.Murugan, M.Tech., Ph.D and M.L.Ashly Beby, Improvement of Unfed Power Qualty Condtoner based on GA Optmzed NN Controller Internatonal Journal of Electrcal Engneerng.ISSN Volume 7, Number 1 (2014), pp [4] Fatha Mekr, Mohamed Machmoum, Nada At Ahmed, Benyounes Mazar, A Fuzzy hysteress voltage and current control of An Unfed Power Qualty Condtoner 2008 [5] Guseppe D Fatta, Frank Hoffmann, Guseppe Lo Re, and Alfonso Urso, A Genetc Algorthm for the Desgn of a Fuzzy Controller for Actve Queue Management IEEE Transactons On Systems, Man, And Cybernetcs Part C: Applcatons And Revews, Vol. 33, No. 3, August 2003 [6] Hany M. Hasanen,, and S. M. Muyeen, Desgn Optmzaton of Controller Parameters Used n Varable Speed Wnd Energy Converson System by Genetc Algorthms IEEE Transactons On Sustanable Energy, Vol. 3, No. 2, Aprl 2012 [7] Hurng-Lahng Jou, Jnn-Chang Wu, Yao-Jen Chang, and Ya-Tsung Feng, A Novel Actve Power Flter for Harmonc Suppresson 2005 X. CONCLUSION Ths paper descrbes GA and PSO tuned PI and Fuzzy based control strategy used n the UPQC, whch manly compensates the reactve power along wth voltage and current harmoncs under unbalanced load-current condtons. For proposed system the smulaton results wth PSO technques prove to be more effectve than wth GAs. In GAs, the lmts defned by the number of parameters gves the search regon whle n PSO, the search regon s ndependent of the number of parameters, gven by the dstance between the randomly selected ntal poston and the poston correspondng to optmal ftness value. The speed of computaton s determned by the velocty ntalzng the PSO algorthm wth whch t reaches to the best soluton. It s also observed that the speed of computaton n PSO s very less n comparson to GAs. [8] Jayalaxm.A Bangarraju.J, and Rajagopal.V, Shunt Actve power flter for power qualty enhancement usng reduced sensor swtchng control strategy, Ffth Internatonal Conference on Power and Energy Systems, Kathmandu, Nepal October, 2013 [9] Mukhtar Sngh, Vnod Khadkkar, Ambrsh Chandra, and Rajv K. Varma, Grd Interconnecton of Renewable Energy Sources at the Dstrbuton Level Wth Power-Qualty Improvement Features IEEE Transactons On Power Delvery, Vol. 26, No. 1, Jan 2011 [10] Metn Kesler and Engn Ozdemr, Synchronous-Reference-Frame-Based Control Method for UPQC Under Unbalanced and Dstorted Load Condtons IEEE Transactons On Industral Electroncs, Vol. 58, No. 9, September APPENDIX Supply voltage: 110V, 50Hz. Load Used: Three Sngle phase loads Load 1: R=15Ω; C = 1000µf Load 2: R=25Ω ; C = 1000µf 200