TIME VARYING ACCELERATION COEFFICIENTS WITH DOMINANT SOCIAL COMPONENT PARTICLE SWARM OPTIMIZATION FOR INTERCONNECTED POWER SYSTEM

Transcription

1 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. TIME VARYING ACCELERATION COEFFICIENTS WITH DOMINANT SOCIAL COMPONENT PARTICLE SWARM OPTIMIZATION FOR INTERCONNECTED POWER SYSTEM S. Surendran 1 and S.Thangael 1 Department of Electrcal and Electroncs Engneerng, Tagore Insttute of Engneerng and Technology, Inda Department of Electrcal and Electroncs Engneerng, K.S.Rangasamy College of Technology, Inda E-Mal: suree_be98@yahoo.co.n ABSTRACT In ths paper, Tme Varyng Acceleraton Coeffcents wth Domnant Socal Component (TVACDSC), Tme Varyng Acceleraton Coeffcents (TVAC) and Fxed Acceleraton Coeffcents (FAC) Partcle Swarm Optmzaton (PSO) technques are proposed to optmze the proportonal, ntegral and derate gans of PID controller for two equal area nterconnected power system. The performance ndex s consdered as mnmzaton of Integral of Tme weghted Absolute alue of Error (ITAE). Man purpose of ths nterconnected power system s to prode better qualty of power to consumers. For ths reason, the responses are analyzed and compared wth the responses of Gratatonal Search Algorthm (GSA) n terms of rse tme, settlng tme, peak oershoot, peak undershoot and peak tme. Fnally, suggested the better optmzaton technque to prode better qualty of supply to consumers n nterconnected power system. Keywords: nterconnected power system, PID controller, PSO, acceleraton coeffcent, cognte, socal components. INTRODUCTION The mportance of nterconnected power system s to meet contnuous power supply wthout deatng frequency and te lne power flow thereby all the consumers wll hae been obtanng contnuous power supply. The nterconnected power system conssts of seeral control areas. If any sudden perturbatons occurred n the load of nterconnected power system then frequency deatons occurred n all areas. For acheng better qualty of power system, the frequency must be mantaned at constant. If any deatons happened n frequency of nterconnected power system, t wll lead to reduce the relablty and effcency of system. So, frequency deatons must be eradcated as soon as possble. If the frequency deaton s not eradcatng wth control acton then hgh frequency deaton arose for eery next load perturbatons. Durng hgh frequency deatons occurred n nterconnected power system, t wll dsgrace the performance of load, dstress the performance of system protecton schemes and harm equpment. These hgh frequency deatons leads to the nterconnected power system crumble. Dscrepancy n frequency harmfully affects the operaton and speed control of nducton and synchronous motors. In domestc area, when the power supply frequency retards, refrgerator s effcency goes down; teleson and ar condtoners reacte power consumpton ncreases consderably. For ths reason, t s ery sgnfcant to mantan the frequency at rated alue or wthn tolerable range. The adjustment of generaton for contnuous load change, the system frequency has to mantan wthn adequately small tolerance leel. So, Load Frequency Control (LFC) plays a ery tal role n an nterconnected power system for supplyng electrc power wth better qualty. The real power flow between the dfferent control areas are regulated by the Load Frequency Control (LFC) at constant frequency. The ancent method to control the balance between the net generaton and load demand by usng one generatng unt desgned as the regulated unt and t was manually adjusted to Control. At present, seeral unts are noled n load frequency control and t leads to mproe the regulaton, effcency and economy n oerall system. Intally, speed goernor adjusts the nput to the turbne, wll try to reduce the frequency deaton and te lne power flow deaton to zero when load s subjected to changes tself. Results were not tolerable one when the hgh obstacles mpled on the frequency constancy. In ths crcumstance t s ery essental to attan much better frequency constancy than obtaned by the speed goernor system. The addtonal control technques hae been deeloped n accompanyng wth speed goernng mechansm to mnmze the frequency deaton and make te lne power flow deaton to zero. Many control strateges for Load Frequency Control n nterconnected power systems hae been proposed by researchers as follows. O.I.Elgerd and C.E.Fosha [] deeloped a dynamc system model of mult area nterconnected power system. Integral gan (K ) and frequency bas (B) parameters dscoered from classcal optmzaton theory n the astuteness of mnmzng the Integral Square Error (ISE) crteron. C.E.Fosha and O.I.Elgerd [3] agan deeloped state arable model of the mult area nterconnected power system MW - frequency control problem. Modern control technques were used to deelop optmal controllers for nterconnected power 1936

2 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. system that sgnfcantly mproed the transent response for correspondng load perturbatons. C.T.Pan and C.M.Law [4] presented an adapte controller for load frequency control. Aleksandar M.Stankoc et al., [5] presented a physcally motated augmentaton of the standard ntegral controller n LFC. Janardan Nanda et al., [6] nestgated contnuous dscrete mode nterconnected hydro thermal system usng conentonal ntegral and proportonal ntegral controllers (system works n contnuous mode, controller works n dscrete mode). Seyed Abbas Taher and Reza Hematt [7] presented quanttate feedback theory for load frequency control. I. A. Chdambaram and B. Paramasam [8] desgned a Genetc Algorthm (GA) based controllers wth Integral Square Error crteron for the decentralzed load frequency control of two area nterconnected thermal reheat power systems wth and wthout Redox Flow Batteres (RFB) consderng Thyrstor Controlled Phase Shfter (TCPS) n the Te-lne. H. A. Shayanfar et al., [9] descrbed a Mult Input Mult Output (MIMO) desgn technque based on the Characterstc Loc (CL) method appled to load frequency control of nterconnected power system. Wen Tan [1] dscussed a unfed PID tunng method for load frequency control of nterconnected power system. Gayadhar Panda et al., [11] presented a modfed genetc algorthm based optmal selecton of ntegral gan and frequency bas constant for load frequency control of mult area nterconnected power system. K. P. Sngh Parmar et al., [1] presented output feedback controller desgn for two area nterconnected power system. Armn Ebrahm Mlan and Babak Mozafar [13] presented new genetc algorthm based method for acheng optmal gans n two area nterconnected power system. Serhat Duman and Nuran Yorukeren [14] presented GSA method for determnaton of optmal PID parameters n two area nterconnected power system. Rta San et al., [15] presented Bacteral Foragng Optmzaton (BFO) method for two area nterconnected power system. Mostly, many researchers followed the Integral, PI and PID control strateges. The domnance of Integral, PI and PID control strateges hae a smple structure, long-standng, successful desgn technque, hgh ablty of solng many practcal control problems. Among these strateges, PID control strategy ges better response than Integral and PI control strateges. So, the PID controller parameters are tuned by usng dfferent tunng technques. The Zegler Nchols technque s one of the most wellknown ancent PID tunng technques. Ths tunng approach works farly well for a wde range of practcal processes. Howeer, sometmes t does not prode good tunng and tends to produce a ast oershoot. Therefore, ths method usually needs retunng before appled to load frequency control. Seeral ntellgent approaches such as Genetc Algorthm [8, 11, 13], Partcle Swarm Optmzaton [, 1,, 3], Bacteral Foragng Optmzaton Algorthm [15] and etc. hae been suggested to enhance the capabltes of tradtonal PID parameter tunng technques. Rana Hassan et al., [16] nestgated the superorty of PSO oer GA and statcally proed that 99% confdence leel n 7 out of 8 test problems. Emad Elbeltag et al., [17] nestgated that the PSO method was generally found to perform better than genetc algorthm, memetc algorthm, ant colony optmzaton and shuffled frog leapng algorthm n terms of success rate and soluton qualty. M.A.Panduro et al., [18] nestgated that the partcle swarm optmzaton and dfferental eoluton algorthm gae a better performance n terms of the sde lobe leel wth respect to the genetc algorthm under equal computatonal tme. Rega Rajendra and Dlp K. Prathar [19] nestgated the performance of PSO algorthm and t ges better results due to ts nherent ablty to carry out the global and local searches smultaneously. On the other hand, the GA s a potental tool for global search, although t may not be so much powerful n local search. From the aboe nestgatons, PSO s better optmzaton technque than GA. Some adantages of PSO algorthms are derate free algorthm, easy to mplementaton, lmted number of parameters, smple calculaton and smple concept compared from other optmzaton algorthms lke BFO algorthm [15]. PSO Algorthm can be appled to the tunng of PID controller gans (K p, K and K ) to ensure optmal control performance. The best alues of PID controller parameters can yeld a best system response and ts alues are obtaned from the performance ndex. Many number of performance ndces are used n load frequency control. Fernando G. Martns [4] nestgated about ITAE crteron based PID parameters tunng gae better performance than Zegler-Nchols tunng. Serhat Duman and Nuran Yorukeren [14] demonstrated that ITAE crteron gae better performance of conergence than ISE crteron. Gayadhar Panda et al., [11] found that superorty n responses are obtaned from mnmzng the ITAE crteron compared from ISE crteron. R. Krshna kumar [5] obsered that the performance ndex ITAE crteron gae better dynamc performance compared from ISE crteron. Jeethaenkatachalam and Rajalaxm. S [6] demonstrated the superorty of the ITAE crteron n the dampng and settlng of the transent responses compared to ISE crteron. From the aboe nestgatons, t s concluded that the mnmzaton of ITAE crteron ges better performance compared from mnmzaton of ISE crteron and Zegler-Nchols tunng. In ths paper, TVACDSC, TVAC and FAC PSO technques are proposed for tunng the proportonal, ntegral and derate gans of PID controller n two equal area nterconnected power system wth 1% step load perturbaton n area 1. The proposed TVACDSC PSO and TVAC PSO technques ncreases the acceleraton coeffcents from ntal alue to fnal alue (constrcton factor s decreased) and FAC PSO algorthm acceleraton coeffcents are fxed (constrcton factor s constant). These proposed PSO technque responses are compared wth responses obtaned from GSA n ref. [14] and 1937

3 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. analyzed n terms of rse tme, 5% of settlng tme, peak oershoot, peak undershoot and peak tme. MODELING OF TWO AREA INTERCONNECTED POWER SYSTEM Frequency deaton ( ) The net surplus power n the area followng a dsturbance P D equals P G P D MW, and the power wll be absorbed by the system n three ways: a) By ncreasng the area knetc energy W at the rate dw kn d f Wkn d W kn Δf (1) dt dt f f dt b) By ncreased load consumpton: All typcal loads (because of the domnance of motor load) experence an ncrease D P D / f MW/Hz wth speed or frequency. Ths D parameter can be found emprcally; c) By ncreasng the export of power, a te lnes, wth the total amount P MW defned poste out from the area. From the aboe aspects, the power equlbrum equaton n th area s gen n equaton (). Fgure-1. Block dagram model of control area. Incremental generated power ( P G ) Real power generaton n a synchronous machne s controlled by changng the prme moer torque. Ths change n torque s based on openng or closng the steam ale n a steam turbne. Process of openng or closng the steam ale s based on the functonal dagram. Functonal dagram s shown n Fgure-. W f d dt Δf DΔf te kn G D () From equaton (), all terms dmensons are n MW. These dmensons are conerted nto per unt representaton by ddng the total rated area power (P ) expressed n MW n th area. Equaton (3) represents the dmensons n per unt. H d f dt Δf DΔf te G D (3) Where, Inerta constant H = W /P n sec. ndcates the nomnal alues Fgure-. Functonal dagram of turbne control arrangement. From ths Fgure- ncremental generated power s obtaned from change n speed changer poston and speed goernor. Ths functonal dagram of Fgure- s modfed nto block dagram model. The block dagram model of turbne control arrangement s shown n Fgure- 3. The equaton (3) s smplfed nto equaton (4). K p [ G (s) D (s) te (s)] ΔF (s) (4) 1 st p Where, K p 1 D and Hz/puMW H Tp sec. f D Block dagram model of equaton (4) s shown n Fgure-1. Fgure-3. Block dagram model of turbne control arrangement. In ths block dagram has been represented by two tme constants T g and T t. The former represents the tme constant of the goernor, and the latter represents the tme lag of the turbne n th area. 1938

4 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. Incremental te-lne power ( ) The total real power exported from th area (P ) s equal to sum of all out flowng lne powers (P ) n the lnes connectng th area wth th areas,.e., P te P te If the te-lne losses are neglected, then the nddual te-lne powers expressed n mathematcal form s shown n equaton (6). (5) The total ncrement n exported power from th area n two area system s obtaned from equaton (11) s expressed n equaton (1). te T (s) [ΔF (s) ΔF (s)] (1) s Dagrammatc representaton of the equaton (1) s shown n Fgure-4. P te V V sn(δ δ) Pte max sn(δ δ) (6) X P r Where, V = V e δ, V = V e δ are the termnal bus oltages of the te-lne and X s reactance between th area and th area. The maxmum alue P x represents the maxmum real power (here expressed n per unt of the rated area power (P )) that can be transmtted a the telne. The te-lne s termed as weak because P x P. δ and δ are phase angle deaton from ther nomnal alues δ and δ. If the phase angles deate from ts nomnal alues, then the te-lne power flow s also deate from ts nomnal alue. Deaton of te-lne power flow s mathematcally expressed n equaton (7). te P V V te ( δ δ ) cos(δ δ )( δ δ ) (7) (δ δ ) X P The relatonshp between the phase angle deaton and area frequency deaton s expressed n equaton (8). δ π Δfdt and δ π Δfdt (8) From equatons (7) and (8), te r T ( Δf dt Δf dt) (9) Where, V V T π X P r cos(δ δ ) Takng Laplace Transform of equaton (9) Fgure-4. Block dagram model of ncremental te-lne power exported from th area. The total ncrement n exported power from th area n two equal area nterconnected power system s obtaned from total ncrement n exported power from th area s expressed n equaton (13). te Pr (s) te (s) te (s) (13) P r Because, two equal area nterconnected power system, the rated area powers P and P are equal. Area Control Error (ACE) When systems are nterconnected, te-lne power flows as well as frequency must be controlled. The sum of te-lne and frequency errors can be expressed as Area Control Error (ACE) n equaton (14). ACE b ΔF (14) te Dagrammatc representaton of equaton (14) s shown n Fgure-5. te T (s) [ΔF (s) ΔF (s)] (1) s The total ncrement n exported power from th area s fnally obtaned from equaton (1) s expressed n equaton (11). Fgure-5. Block dagram model of Area Control Error. te 1 (s) T [ΔF (s) ΔF (s)] s (11) 1939

5 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. Two area nterconnected power system model s obtaned from the block dagram models of Fgures-1, 3, 4 and 5 s shown n Fgure-6. Fgure-6. Block dagram representaton of uncontrolled two area nterconnected power system. PID CONTROLLER AND PERFORMANCE INDEX The classcal PID controller can be expressed n two area nterconnected power system as gen n equatons (15) and (16). K c1(s) (K p Kds)(ACE 1) (15) s K c (s) (K p Kds)(ACE ) (16) s K p, K and K d are the proportonal, ntegral and derate gans respectely. Inddual performance of each controller has some nddual characterstcs.e., ntegral controller nullfy the steady state error and reduces the rse tme, proportonal controller reduces the rse tme and steady state error and derate controller mproes the transent response, reduces the settlng tme and oershoots. These PID gan parameters are optmally tuned by usng partcle swarm optmzaton technque. The optmzaton process s based on the performance ndex. The performance ndex s consdered as the Integral of the Tme multpled Absolute alue of Error (ITAE). The performance ndex (J) commonly used for optmzng the PID gan settng of the supplementary controller and t s based on the frequency deatons of two areas and te lne power flow deaton between two areas. Performance Index s gen n equaton (17). J t(te 1 Δf1 Δf )dt (17) PARTICLE SWARM OPTIMIZATION (PSO) ALGORITHM Partcle swarm optmzaton (PSO) algorthm was nented by Dr. James Kennedy and Dr. Russell Eberhart n 1995 and ts elementary dea was orgnally nspred by mock of the socal behaor of brds flockng. It s based on the natural process of group communcaton to share nddual knowledge. When a group of brds search food n a searchng space, although all brds do not know where the best poston s. But from the nature of the socal behaor, f any member can fnd out a desrable path to go, the rest of the members wll follow quckly. In PSO, each member of the populaton s consdered as partcle and the populaton s consdered as swarm. All partcles are ntated randomly and mong n randomly chosen drectons, each partcle goes through the searchng space to fnd the personal best (best poston of each partcle) and the poston yeldng the lowest alue amongst all the personal best that s global best (best poston of partcle n the entre swarm). Partcles elocty s updated by usng personal and global best postons and then each partcle postons are updated by the present elocty. The process s repeated up to the stoppng crteron whle predetermned n adance. Postons of partcle are updated by usng equaton (18). x tr 1 tr tr 1 (18) Velocty of partcle s updated by usng equaton (19). tr 1 tr tr 1 1 c r [P tr best tr tr ] c r [G tr best tr ] (19) Where, x = Poston of partcle at teraton tr. = Velocty of partcle at teraton tr. P = Personal best poston of partcle at teraton tr. G = Global best poston at teraton tr. c and c = Cognte and Socal parameters respectely. r and r = Random numbers n between and 1 at teraton tr. In ths optmzaton technque, updated elocty s ncreased oer teraton. The hgh alue of updated elocty respond poor optmzaton result (skpped the optmal alue). Due to ths reason, elocty clampng, nerta weght factor and constrcton factor are ntroduced. Velocty clampng s lmtng the total elocty only. Inerta weght s only concentrate the preous teraton elocty but unable to concentrate the acceleraton components. Constrcton factor concentrates the acceleraton components and lmts the total elocty. For ths reason, constrcton factor s taken as man role of ths paper. Ths algorthm s dscussed n to three ways. Fxed Acceleraton Coeffcents (FAC) Constrcton factor χ s extremely mportant to control the exploraton and explotaton tradeoff, to ensure 194

6 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. conergence behaor, and also to exclude the nerta weght ω and the maxmum elocty x. Velocty updates equaton of the partcle s modfed from equaton (19) and s gen n equaton (). tr 1 ( tr tr 1 1 c r [P tr best tr tr ] c r [G tr best Where, χ, c1 c 4 tr ]) () From the aboe expressons, constrcton factor s obtaned from acceleraton coeffcent. If φ < 4, then all partcles would slowly spral toward and around the best soluton n the searchng space wthout conergence guarantee. If, then all partcles conerge quckly and guaranteed. For ths reason, c and c are taken as and more than. Ths paper suggests the alues for c and c as.5. For ths reason, 5% updated elocty s controlled by each teraton. Tme Varyng Acceleraton Coeffcents (TVAC) Tme (or Iteraton) aryng acceleraton coeffcents are to enhance the global search n the begnnng of the optmzaton and to moe forward the partcles to conerge towards the best optmum soluton at the end of search. The arable acceleratng coeffcents are obtaned from the equatons (1) and () and dscoer the updated elocty usng equaton (). tr c1 c1 (c1f c1 ) (1) tr c max tr c (cf c) () tr max Where, c and c are ntal and fnal alues of cognte acceleraton factors, c and c are ntal and fnal alues of socal acceleraton factors, tr x s the maxmum number of teratons, tr s the number of current teratons. In ths paper, the suggested ntal alues of cognte and socal parameters and fnal alues of cognte and socal parameters are.5. Because at the ntal stage the constrcton factor s hgh and then gradually reduced to reach low alue for fnal stage. Ths strategy controls the updated elocty. Tme Varyng Acceleraton Coeffcents wth Domnant Socal Component (TVACDSC) Compared wth the socal component, hgh alue of the cognte component wll result n excesse wanderng of ndduals through the search space. In contrast, a relately hgh alue of the socal component may lead partcles to hurry prematurely toward a local optmum. For ths reason, proper control acton of these two components s ery mportant to dscoer the optmum soluton exactly and effcently. So elocty update equaton s slghtly modfed from equaton () and s gen by equaton (3). tr 1 c (r ( tr tr frn c (r tr 1 1 )[G tr best frn 1 tr )[P ]) tr best tr ] (3) Where, frn and frn - fractonal alues ncorporated n cognte and socal components. In ths paper, the suggested fractonal alues 1 and are.5. These alues domnate the socal component compared from cognte component. The updated elocty s found from equatons (1), () and (3). The ntal and fnal alues of cognte and socal parameters are same n TVAC. SIMULATION RESULTS AND DISCUSSION The Smulaton was carred out usng MATLAB programmng and smulnk toolbox. In ths smulaton 5 teratons are taken. Populaton sze of partcle swarm optmzaton s taken as 1. The system parameters are gen n appendx. The proportonal, ntegral and derate gans of PID controller wth occurrence of 1% of step load perturbaton n area 1 of two equal area nterconnected power system are optmzed by usng proposed partcle swarm optmzaton technques wth ITAE crteron. Fxed Acceleraton Coeffcents (FAC) Cognte and socal parameters c and c are consdered as.5. Constrcton coeffcent s obtaned as.5 from the alues of cognte and socal parameters. Incase c and c are consdered as, the constrcton coeffcent s obtaned as 1. Performance Index No. of teratons Fgure-7. Ftness functon for FAC PSO wth 1% step load perturbaton n area 1. If χ s 1, then the current elocty ncreases oer teraton and partcles can hardly change ther drecton to moe back towards optmum, and the swarm derges. If χ<, then the current elocty ncreases as a fractonal 1941

7 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. alue of preous elocty and then quckly changes ts drecton to moe towards optmum. For ths reason, χ s set to.5; ths stuaton ges the elocty update 5% only. So the optmal performance s obtaned moderately. Fgure-7 shows ftness functon for 1% step load perturbaton n area 1. Tme Varyng Acceleraton Coeffcents (TVAC) Cognte and socal parameters c and c are obtaned by usng ntal and fnal alues of cognte and socal parameters. The ntal alues of cognte and socal parameters c and c are. The fnal alues of cognte and socal parameters c and c are.5. If the ntal alues c and c are and fnal alues c and c are.5, then the constrcton coeffcent s obtaned as.868 at frst teraton and.38 at 5 th teraton. The constrcton coeffcent s gradually reduced from.868 to.38. At 5 th teraton, the constrcton coeffcent s.5. For the frst 5 teratons, the constrcton coeffcent s gradually reduced from.868 to.5. Ths alue s hgher than that of FAC. For the next 5 teratons, the constrcton coeffcent s gradually reduced from.5 to.38. Ths alue s lower than that of FAC. Fgure-8 shows ftness functon for 1% step load perturbaton n area 1. Performance Index No. of teratons Fgure-9. Ftness functon for TVACDSC PSO wth 1% step load perturbaton n area 1. The responses obtaned from TVACDSC PSO, TVAC PSO and FAC PSO technques are compared wth the responses obtaned from gratatonal search algorthm n ref. [14]. Fgures-1, 11 and 1 show the comparate responses of frequency deatons n area 1, frequency deatons n area and te lne power flow deatons n two equal area nterconnected non-reheat thermal power system. Performance Index No. of teratons delf1 n Hz Tme n sec. Fgure-1. Frequency deaton n area 1. Fgure-8. Ftness functon for TVAC PSO wth 1% step load perturbaton n area 1. Tme Varyng Acceleraton Coeffcents wth Domnant Socal Component (TVACDSC) The fractonal number s taken as.5. Ths alue reduces the effect of cognte component and ncreases the effect of socal component. Among the three terms (preous teraton elocty, cognte component and socal component), socal component s more domnant compared from other two terms. Ths domnance effect ges better response. Fgure-9 shows ftness functon for 1% step load perturbaton n area 1. delf n Hz Tme n sec. Fgure-11. Frequency deaton n Area. 194

8 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. delpte1 n p.u Tme n sec. Fgure-1. Te lne power flow deaton. Fgures-13 and 14 show the comparate responses of Area Control Error n Area 1 and Area. Fgures-15 and 16 show the comparate responses of PID Controller Output n Area 1 and ts closest ew. Fgures- 17 and 18 show the comparate responses of PID Controller Output n Area and ts closest ew. ACE 1 n p.u ACE n p.u Tme n sec. Fgure-13. Area Control Error n Area 1. x Tme n sec. Area 1 PID output n p.u Area 1 PID output n p.u Area PID output n p.u Tme n sec. Fgure-15. PID Controller Output n Area Tme n sec. Fgure-16. Closest ew of PID Controller Output n Area Tme n sec. Fgure-17. PID Controller Output n Area. Fgure-14. Area Control Error n Area. 1943

9 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. Area PID output n p.u Tme n sec. Fgure-18. Closest ew of PID Controller Output n Area. Table-1 shows the frequency deaton n area 1 comparate responses of 5% tolerance of settlng tme, peak oershoot, peak undershoot and peak tme n the TVACDSC PSO, TVAC PSO, FAC PSO and Gratatonal Search Algorthm technques for two equal area nterconnected non-reheat thermal power system. From Fgure-1 shows TVACDSC, TVAC and FAC PSO reduce the frequency deaton wthn one second than Gratatonal Search Algorthm. Gratatonal Search Algorthm takes more than seconds. From Fgures-1, 11 and 1 show the FAC PSO mproes the response than GSA. TVAC PSO slghtly mproes the response than FAC PSO. TVACDSC PSO drastcally mproes the response than TVAC PSO. From Fgures-1, 11, 1 and Table-1 show the reducton of rse tme, settlng tme, peak oershoot and peak tme n the order of GSA, FAC PSO, TVAC PSO and TVACDSC PSO. From Fgures-15, 16, 17 and 18 show the Speed Changer Poston nput from PID Controller output based on ACE n area 1 and of Fgures-13 and 14. CONCLUSIONS Optmzaton of PID controller for Interconnected Power System usng Tme Varyng Acceleraton Coeffcents wth Domnant Socal Component Partcle Swarm Optmzaton technque ges better performance compared from Tme Varyng Acceleraton Coeffcents PSO, Fxed Acceleraton Coeffcents PSO and Gratatonal Search Algorthm n terms of rse tme, 5% of settlng tme, peak oershoot, peak undershoot and tme taken to reach peak oershoot. TVACDSC, TVAC and FAC PSO technques are gen better response than Gratatonal Search Algorthm. Tme Varyng Acceleraton Coeffcents wth Domnant Socal Component Partcle Swarm Optmzaton technque s better optmzng technque for PID controller parameter optmzaton n Interconnected Power system. Controller types TVACDSC PSO PID Table-1.Frequency deaton responses for Area 1. Settlng tme n sec. Peak oershoot Peak undershoot Peak tme n sec APPENDIX Two areas nterconnected power system parameters P = P = MW H = H = Sec. D = D = 8. X pumw/hz T = T =. 8 Sec. T = T =. Sec. T P = T P = Sec. K P = K P = Hz/puMW R = R =. Hz/puMW P x = MW T =. pumw/hz b = b =. pumw/hz REFERENCES [1] O.I. Elgerd Electrc Energy Systems Theory an Introducton. McGraw Hll. [] O.I.Elgerd and C.E.Fosha Optmum Megawatt- Frequency Control of Mult area Electrc Energy Systems. IEEE Trans. on PAS.PAS-89(4): [3] C.E.Fosha and O.I.Elgerd The Megawatt- Frequency Control Problem: A New Approach a Optmal Control Theory. IEEE Trans. on PAS. PAS- 89(4):

10 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. [4] C.T.Pan and C.M.Law An Adapte Controller for Power System Load Frequency Control. IEEE Trans. on Power Systems. 4(1): [5] Aleksandar M.Stankoc, Glead Tadmor and Tmoor A.Sakharuk On Robust Control Analyss and Desgn for Load Frequency Regulaton. IEEE Trans. on Power Systems. 13(): [6] Janardan Nanda, Ashsh Mangla and Sanjay Sur. 6. Some New Fndngs on Automatc Generaton Control of an Interconnected Hydrothermal System wth Conentonal Controllers. IEEE Trans. on Energy Conerson. 1(1): [7] Seyed Abbas Taher and Reza Hematt. 8. Robust Decentralzed Load Frequency Control Usng Mult Varable QFT Method n Deregulated Power Systems. Amercan Journal of Appled Scence. 5(7): [8] I.A.Chdambaram and B.Paramasam. 9. Genetc Algorthm Based Decentralzed Controller for Load Frequency Control of Interconnected Power Systems wth RFB Consderng TCPS n the Te-Lne. Internatonal Journal of Electronc Engneerng Research. 1(4): [9] H.A.Shayanfar, M.Ghazal and M.Karam. 9. Load Frequency Control usng Mult Varable Characterstc Loc Method n Power Systems. IJTPE. 1(1): [1] Wen Tan. 1. Unfed Tunng of PID Load Frequency Controller for Power Systems a IMC. IEEE Trans. on Power Systems. 5(1): [11] Gayadhar Panda, Sdhartha Panda and C. Ardl. 1. Automatc Generaton Control of Mult-Area Electrc Energy Systems Usng Modfed GA. Internatonal Journal of Electrcal and Electroncs Engneerng. 4(6): [1] K.P.SnghParmar, S.Majh and D.P.Kothar. 11. Optmal Load Frequency Control of an Interconnected Power System. MIT Internatonal Journal of Electrcal and Instrumentaton Engneerng. 1(1): 1-5. [13] Armn EbrahmMlan and Babak Mozafar. 11. Genetc Algorthm based Optmal Load Frequency Control n Two Area Interconnected Power Systems. Transacton on Power system optmzaton, Global Journal of Tech. and Optmzaton. : 6-1. [14] Serhat Duman, Nuran Yorukeren. 1. Automatc Generaton Control of the Two Area Non-reheat Thermal Power System usng Gratatonal Search Algorthm. Przeglad Elektrotechnczny (Electrcal Reew), R. 88, NR. 1a. pp [15] Rta San, Dr. Rajee Gupta and Dr.Grsh Parmar. Dec.1-Feb.13. Optmzaton of LFC usng Bactera Foragng Optmzaton Algorthm. IJETCAS. 3(): [16] Rana Hassan, Babak Cohanm and Oler de Weck. 4. A Comparson of Partcle Swarm Optmzaton and the Genetc Algorthm. Amercan Insttute of Aeronautcs and Astronautcs. pp [17] Emad Elbeltag, Tarek Hegazy and Donald Grerson. 5. Comparson among Fe Eolutonary-Based Optmzaton Algorthms. ELSEVIER, Adanced Engneerng Informatcs. 19: [18] M.A.Panduro, C.A.Brzuela, L.I.Balderas and D.A.Acosta. 9. A Comparson of Genetc Algorthms, Partcle Swarm Optmzaton and the Dfferental Eoluton Method for the Desgn of Scannable Crcular Antenna Arrays. Progress In Electromagnetcs Research B. 13: [19] RegaRajendra and Dlp K. Prathar. 11. Partcle Swarm Optmzaton Algorthm s Genetc Algorthm to Deelop Integrated Scheme for Obtanng Optmal Mechancal Structure and Adapte Controller of a Robot. Scentfc Research, Intellgent Control and Automaton. pp [] James Kennedy and Russell Eberhart Partcle Swarm Optmzaton. Proc. IEEE Int l. Conf. on Neural Networks. IV. pp [1] Asanga Ratnaweera, Saman K.Halgamuge and Harry C.Watson. 4. Self-Organzng Herarchcal Partcle Swarm Optmzer wth Tme-Varyng Acceleraton Coeffcents. IEEE Trans. on Eolutonary Computaton. 8(3):4-55. [] Danel Bratton and James Kennedy. 7. Defnng a Standard for Partcle Swarm Optmzaton. Proceedngs of the 7 IEEE Swarm Intellgence Symposum. [3] Dan Palup Rn, StMaryam Shamsuddn and St Sophyat Yuhanz. 11. Partcle Swarm Optmzaton: Technque, System and Challenges. 1945

11 ARPN Journal of Engneerng and Appled Scences 6-16 Asan Research Publshng Network (ARPN). All rghts resered. Internatonal Journal of Computer Applcatons. ( ). 14(1): [4] Fernando G. Martns. 5. Tunng PID Controllers Usng the ITAE Crteron. IJEE. 1(3): 1-7. [5] R. Krshna Kumar. 1. Performance Comparson of LFC of Two Area Interconnected Power System for the Performance Indces ISE and ITAE wth Optmum Integral Controller Gan Desgned Usng Genetc Algorthm. Internatonal Journal of Appled Engneerng Research and Deelopment. (1): [6] Jeethaenkatachalam and Rajalaxm. S. 13. Automatc Generaton Control of Two Area Interconnected Power System Usng Partcle Swarm Optmzaton. IOSR Journal of Electrcal and Electroncs Engneerng. 6(1):