D.Godwn Immanuel et al. / Internatonal Journal of Engneerng and Technology (IJET) A Mult Obectve Hybrd Dfferental Evoluton Algorthm ated Genetc Algorthm Approach for Optmal Reactve Power and oltage Control D.Godwn Immanuel, G.Selvekumar, C.Chrtober Ar Raan 3 Reearch Scholar, Sathyabama Unverty, Chenna, Taml Nadu, Inda ce Prncpal, Excel Engneerng College, Salem, Taml Nadu, Inda 3 Aocate Profeor, Pondcherry Engneerng College, Pondcherry, Inda dgodwnmmanuel@gmal.com gk7070@gmal.com 3 ar_70@hotmal.com Abtract To provde a qualty and ecured power upply n all the load centre a challengng tak. Due to the ncreae n load and varou emergency condton the voltage level n load centre may vary beyond the acceptable level whch lead to voltage collape and ncreae the ytem loe. To mantan the voltage level wthn the acceptable range n all load centre reactve power upport very eental. The control centre ha to provde the optmal reactve power where ever neceary to prevent thee crtcal tuaton. To provde the controllable amount of reactve power and voltage n the needy regon effcent power flow analy are requred. Many tradtonal method are avalable but t ha lmtaton and converged n poor optmal oluton. In th paper an effectve hybrd Dfferental Evoluton Algorthm ated Genetc Algorthm approach wth mult obectve functon are condered. In th approach the bet feature of Dfferental Evoluton Algorthm and Genetc Algorthm are combned together and developed. In IEEE 30bu the effectvene and the uperorty of th method verfed. Keyword-Newton Raphon Power Flow, Genetc Algorthm, Dfferental Evoluton, olt Ampere Reactve, oltage Stablty Index I. INTRODUCTION Due to the tremendou ncreae n demand of electrcty, mantanng the voltage tablty very dffcult tak. Becaue of the uncertanty of ncreae n load and emergency tuaton, the ytem may loe t tablty and the voltage devate acro the acceptable level. Th can be remeded by proper necton of reactve power optmally wherever needed. Optmal reactve power control play a vtal role n ecured and economc operaton of power ytem. It non-lnear problem wth a combnaton of both dcrete and contnuou varable. Th optmal necton of reactve power n the crtcal tuaton can mnmze the voltage devaton a well a mnmze the ytem loe. Once the voltage devaton mnmzed, the tablty of the ytem can be mantaned. To mantan the ytem tablty, the control varable taken nto conderaton are generator bu voltage, tranformer tap ettng and wtchable AR ource. Several conventonal method are avalable for reactve power control problem. The quadratc programmng for optmal reactve power control uggeted n []. In [] the reactve power plannng wth non-lnear programmng mplemented. Mxed nteger programmng developed and mplemented n [3] for reactve power and voltage control. All thee technque have lot of problem uch a converged n mnmal oluton; more number of teraton etc. Thee problem can be overcome by the ntroducton of ntellgent technque. In [4], the bogeography wate optmal AR control dcued. The general quantum algorthm ued n [5]. In [6], the optmal reactve power dpatch baed on dfferental evoluton algorthm mplemented. The popular work n th area mult-obectve optmzaton approache. In [7], the mult-obectve optmzaton of reactve power control baed on fuzzy ytem uggeted. The hybrd real coded genetc algorthm and dfferental evoluton algorthm appled for optmal power flow n [8]. Mot of the work n th area, havng only two obectve functon condered. In th paper, a hybrd ntellgent algorthm for mult-obectve functon mnmzaton of real power loe, mnmzaton of voltage devaton and mnmzaton of voltage tablty ndex are condered. ISSN : 0975-404 ol 6 No Feb-Mar 04 99
D.Godwn Immanuel et al. / Internatonal Journal of Engneerng and Technology (IJET) II. PROBLEM FORMULATION A. Obectve Functon The bac tratege of reactve power control problem are to dentfy the optmal control varable whch can mnmze the obectve functon. In th hybrd formulaton the followng obectve functon are condered. ) Mnmzaton of real power loe The obectve functon mnmzaton of real power lo can be calculated by nl PLo = gk[ k = + nl : number of lne g k : conductance of the k th co( δ δ lne & : voltage magntude at the bue & δ & δ : voltage phae angle at the bue & ) Mnmzaton of voltage devaton To provde ecurty and ervce qualty the voltage devaton ha to be mnmzed. Then only the voltage tablty can be mantaned. The obectve functon mnmzaton of voltage devaton can be calculated by D = NL = )].0 () NL : number of load bue : voltage magntude of bu 3) Mnmzaton of voltage tablty ndex The tablty of the ytem can be mproved by reducng the value of voltage tablty ndex n load bue. The voltage tablty ndex ha to be calculated for all the load bue. The load bu whch havng the hghet value of SI the mot vulnerable bu. The obectve functon SI can be calculated a ng SI = F = ng +,..., n = & : voltage magntude at the bue & n : number of bue ng : number of generator bue B. Problem Contrant ) Equalty Contrant The equalty contrant real and reactve power balance equaton at all the bu bar are formulated from Newton Raphon power flow analy. P Q g g P d Q d nb = nb = = = Y Y () (3) co( δ δ θ ) (4) n( δ δ θ ) (5) nb : number of bue Y : mutual admttance n between the node and δ&δ : angle of bu voltage of bu and bu repectvely θ : admttance angle between the bue and Pg&Qg : real and reactve power generaton at bu Pd&Qd : real and reactve power demand at bu ) Inequalty Contrant The nequalty contrant of the control varable are formulated by P mn P P max (6) ISSN : 0975-404 ol 6 No Feb-Mar 04 00
D.Godwn Immanuel et al. / Internatonal Journal of Engneerng and Technology (IJET) Q g g mn Q Q max (7) g g mn max (8) g g T mn T T max (9) Qc mn Qc Qc max (0) Sl mn Sl Sl max () P mn and P max : mnmum and maxmum lack bu real power. Qg mn and Qg max : mnmum and maxmum reactve power generaton. g mn and g max : mnmum and maxmum value of generator voltage. T mn and T max : mnmum and maxmum range of tap changng tranformer. Qc mn and Qc : mnmum and maxmum allowable output of reactve power compenaton equpment. Sl mn and Sl max : mnmum and maxmum lne flow lmt III. HYBRID GENETIC ALGORITHM AND DIFFERENTIAL EOLUTION ALGORITHM In th hybrd algorthm, the bet feature of Genetc Algorthm and Dfferental Evoluton Algorthm combned. The bet factor electon and cro-over are taken from Genetc Algorthm and the bet factor mutaton taken from Dfferental Evoluton Algorthm are combned together and gve the effcent oluton. The tep by tep proce of th algorthm dcued below. A. Intalzaton Create a populaton wth n number of chromoome. Intalze random value for all chromoome wthn the lmt of control varable. B. Selecton The proce of electng bet ftne chromoome from the populaton called a electon. The Roulette Wheel electon approach ued n th hybrd algorthm. C. Cro-over It the proce of generatng new off-prng by exchangng the nformaton among the bet chromoome. In th proce, the ngle pont cro-over technque appled n th hybrd algorthm. D. Mutaton It the proce of mutatng all the chromoome and generate a donor vector. In th algorthm, the mot commonly ued bet mutaton trategy DE/rand/ appled to mprove the mutaton proce. I. IMPLEMENTATION OF PROPOSED HYBRID ALGORITHM The propoed hybrd algorthm ha been developed and appled ung MATLAB oftware. In th hybrd algorthm, a populaton of 30 chromoome are created. In each chromoome, t ha 4 gene. There are 5 gene of generator bue, 4 gene of tap changng tranformer and 5 gene of tatc capactor. Run the NR power flow analy for each chromoome and get the dependent and ndependent varable. All the contrant are checked for thee chromoome, whether t wthn the lmt are not. The chromoome whch are not n the lmt are reected from the populaton. The mnmum and maxmum lmt of generator voltage are 0.95 and. p.u. The tranformer tap ettng range n between 0.9 and. p.u. and the wtchable AR ource vare n between 0 and 5 wth each tep of. Calculate the obectve functon for all chromoome n the populaton and elect the bet chromoome from the mattng pool by Roulette Wheel approach. The cro-over contant ued n th hybrd algorthm 0.4. Generate a random number to each chromoome and check f t le than 0.4. If t le, then t elected for mattng. The ngle pont cro-over ued n th algorthm. After the cro-over proce, mutaton can be done for all chromoome n the populaton. The DE/rand/ rule appled n mutaton proce and the calng factor 0.9 condered. A target vector elected for mutaton. The ame proce electon, cro-over and mutaton are repeated untl the toppng crtera reached. The toppng crtera elected n th hybrd algorthm 00 generaton.. RESULTS AND DISCUSSIONS The propoed hybrd algorthm ha been verfed n IEEE 30 bu ytem. In th ytem t havng lack bu, 5 generator bue, 4 load bue, 4 on load tap changng tranformer, 5 tatc compenator and 4 tranmon lne. The propoed hybrd algorthm appled n the above ytem wth the treed condton of 5% of load. The mot crtcal bue are frt dentfed by the voltage tablty ndex. The bue have the hghet value of SI the mot crtcal bu. The bue 30,9,6,5 and 4 are dentfed a the mot crtcal ISSN : 0975-404 ol 6 No Feb-Mar 04 0
D.Godwn Immanuel et al. / Internatonal Journal of Engneerng and Technology (IJET) bue. Connect the tatc AR compenaton n thee bue. Apply the hybrd algorthm and get the optmal oluton. Sl. No. TABLE-I CONTROL ARIABLES AND OPTIMAL SOLUTION OF IEEE30 BUS TEST SYSTEM Control varable Mn lmt Max lmt Intal Settng Optmal ettng ung GA Optmal ettng ung DE Optmal ettng ung GA-DE g.95..05.0500.0500.050.050 g.95.05.0400.056.08.037 3 g3.95.05.000.0063.008.06 4 g4.95.05.000 0.9895.08.00 5 g5.95.05.0500.0584.030.000 6 g6.95.05.0500.0806.030.005 7 T.9. 0.9780.0500 0.97 0.978 8 T.9. 0.969 0.9000 0.935 0.969 9 T3.9. 0.930 0.950 0.900 0.93 0 T4.9. 0.9680 0.9500 0.954 0.968 Q30 0 5 0 5 4 4 Q9 0 5 0 5 5 5 3 Q6 0 5 0 5 3 4 Q5 0 5 0 5 Q30 0 5 0 3 4 P Lo (MW) 0.760 0.55 0.4 0.43 SI 0.978 0.807 0.73 0.7 oltage Devaton (p.u) - - 0.8 0.54 The lne data and bu data of IEEE 30 bu tet ytem taken from [9].Apply the propoed hybrd algorthm and get the optmal oluton of all obectve. The tet reult of the propoed hybrd algorthm are compared wth the extng method gven n [0]and []. Table- how the optmal ettng of the control varable for IEEE 30 bu tet ytem. By th hybrd algorthm the real power lo mnmzed, the voltage tablty ndex of the mot crtcal bu reduced and the voltage devaton mnmzed o that the tablty of the ytem mproved. From thee mulaton reult the hybrd ntellgent algorthm work atfactorly and how the potentalty. Fg..Convergence charactertc of voltage tablty ndex ISSN : 0975-404 ol 6 No Feb-Mar 04 0
D.Godwn Immanuel et al. / Internatonal Journal of Engneerng and Technology (IJET) The Fg. how the convergence curve of voltage tablty ndex of the mult obectve functon.it converged wth a global oluton of 0.7. Fg..Convergence charactertc of voltage devaton The Fg. how the convergence curve of voltage devaton of the mult obectve functon. The voltage devaton mnmzed to an optmal value of 0.54 p.u. Fg. 3.Convergence charactertc of real power lo The Fg.3 how the convergence curve of real power lo mnmzaton of mult obectve functon. The real power lo mnmzed to a global value of 0.43 MW I. CONCLUSION In th paper a hybrd Dfferental Evoluton Algorthm baed Genetc Algorthm ha been developed and appled uccefully to olve reactve power control problem. Th problem ha been formulated wth the mult obectve functon mnmzaton of real power loe, mnmzaton of voltage devaton and mnmzaton of voltage tablty ndex. The lmtaton of the ndvdual Dfferental Evoluton Algorthm and Genetc Algorthm are overcome n th hybrd algorthm. The propoed hybrd algorthm ha been teted on IEEE 30 bu tet ytem whch gve the optmal oluton and atfe all the equalty and nequalty contrant. The mulaton reult how that the propoed algorthm uperor to the method compared wth the tet reult. REFERENCES [] Grudnn. N, Reactve power optmzaton ung ucceve quadratc programmng method. IEEE Tran Power Sytem 998, 3,pp.9-5.. [] La LL, Ma JT, Applcaton of evolutonary programmng to reactve power plannng-comparon wth nonlnear programmng approach. IEEE Tran Power Sytem, 997;: pp.98-06. [3] Km DH, Lee JH, Hong SH, Km SR. A mxed nteger programmng approach for the lnearzed reactve power and voltage control-comparon wth gradent proecton approach, Internatonal conference on energy management and power delvery, proceedng of EMPD98, vol.;998, pp.67-7. [4] P.K.Roy, S.P.Ghohal, S.S.Thakur, optmal var control for mprovement n voltage profle and for real power lo mnmzaton ung bogeography baed optmzaton, Electrcal Power and Energy Sytem,43(0) pp.830-838. [5] John G. lachogann, Jacob Otegaard, Reactve power and voltage control baed on general quantum genetc algorthm. Expert Sytem wth Applcaton36,(009) pp.68-66. [6] A.A.Abou El Ela, M.A.Abdo, S.R.Spea, Dfferental evoluton algorthm for optmal reactve power dpatch; Electrc Power Sytem Reearch 8(0) pp.458-464. [7] Wen Zhang, Yutan Lu, Mult obectve reactve power and voltage control baed on fuzzy optmzaton trategy and fuzzy adaptve partcle warm, Electrcal Power and Energy Sytem, 30,(008),pp.55-53.. [8] C.N.Rav, G.Selvakumar, C.Chrtober Ar Raan, Hybrd real coded genetc algorthm-dfferental evoluton for optmal power flow, Internatonal Journal of Engneerng and Technology, vol.5,no.4,03 pp.3404-34.. [9] Alac O, Scott B, Optmal load flow wth teady tate ecurty, IEEE Tranacton on power Apparatu ytem, 974, pp. 87-0. [0] D.Devara, J. Preetha Roelyn, Genetc algorthm baed reactve power dpatch for voltage tablty mprovement, Electrcal Power and Energy Sytem,00, vol 3, pp. 5-56. [] Godwn Immanuel, Dr. G. Selva Kumar, Dr. C. Chrtober Ar Raan, Dfferental Evoluton Algorthm Baed Optmal Reactve Power Control for oltage StabltyImprovement, Appled Mechanc and Materal, olume. 448-453, October 03,Page No.357-36. ISSN : 0975-404 ol 6 No Feb-Mar 04 03