差分进化算法研究及其应用

922200911167121819(2009)2226673204ScienceTechnologyandEngineeringVol19No122Nov.2009Z2009Sci1Tech1Engng1(,510650),,,TP183;A2009818,,[13],,,,,,,[4],StornRPriceK1995(Dif2ferentialEvolution,DE)[56],,1996IEEE,,DE,DE[7,8]DE,DE,,,1(DifferentialEvolution,DE)Storn1995,,DEDE,DE(PSO,),,DE,,,,,,,1.1DE1.1.1DENPD,:(i=1,2,,NP)(1)(1):i;G;NP,NP,DE,xLjxjxUj,:xji,0=rand[0,1](xUj-xLj)+xLj,(i=1,2,,NP;j=1,2,,D)(2)(2):rand[0,1][0,1],,1.1.2xi,G,i=1,2,,NP,DE:vi,G+1=xr1,G+F(xr2,G-xr3,G)(3)(3),r1,r2r3,r1,r2r3iNP4F[0,2],1.1.3,:ui,G+1=(u1i,G+1,u2i,G+1,,uDi,G+1)(4)uji,G+1=vji,G+1,if(randb(j)CR)orj=rnbr(i);xji,G+1,else,(i=1,2,,NP,j=1,2,,D)(5)(5):randb(j)[0,1]j;rnbr(i)1,2,,D,ui,G+1vi,G+1;CR,[0,1]1.1.4ui,G+1,DExi,G,DE,1.1.5,,uji,G+1xLjuji,G+1xUj,:uji,G+1=rand[0,1](xUj-xLj)+xLj,(i=1,2,,NP;j=1,2,,D)(6)(6),,,1.2DE(1)DE,DE:;(2),k=l;(3),;(4),,;,;(5),,;(6),;(7),;(8)k=k+1,(4)4766921z=f(x,y)=(x2-2x)e-x2-y2-xy,|x|3,|y|2,Matlab[9](1(a));(1(b)),,11,,2(-1,0.5)(0.8,-0.3).,,DE,DE,DE,,(),,,,,,NP=60,CR=0.3,F=0.1,G=50,DE,22DE1,DE(0.61105,-0.30552),-0.6414242z=f(x,y)=3(1-x)2e-x2-(y+1)2-10(15x-x3-y5)e-x2-y2-13e-(x+1)2-y2,|x|3,|y|31,DE,33DE2576622:,DE(0.22828,-1.6255),-6.5511333,,[10],,,,,1..:,19892..:,20013,..:,19974,,..,2007;31(2):1931965StornR,PriceK.Differentialevolutionasimpleandefficienta2daptiveschemeforglobaloptimizationovercontinuousspaces.Berke2ley:UniversityofCalifornia,20066LampinenJ.Abibliographyofdifferentialevolutionalgorithm.ht2tp://(6):6636828ChengS,HwangC.Optimalapproximationoflinearsystemsbyadif2ferentialevolutionalgorithm.IEEETransonSystems,ManandCy2bernetics:A,2001;31(6):6987079.MATLAB5.X.:,200010..:,2001ResearchandApplicationBasedonDifferentialEvolutionAlgorithmWANDong(DepartmentofComputer,GuangdongCommunicationPolytechnic,Guangzhou510650,P.R.China)[Abstract]Aimedatarecentlydevelopedevolutionalgorithm,differentialevolutionalgorithm,theprinciple,processandparametersofdifferentialevolutionalgorithmareintroducedandanalyzedsystematically,thendifferen2tialevolutionalgorithmisestablishedforextremumproblemsofmultivariatefunction.Thismethodhasadvantageofchoosinginitialpointrandomly;thereliabilityandefficiencyofthisalgorithmaredemonstratedbysomenumericalexperiments.[Keywords]differentialevolutionalgorithmextremumvalueofmultivariatefunctionminimumvalue67669