粒子群及遗传算法在相对论返波管中的应用.pdf

计 算 物 理 Vo1．31．No．4 第31卷 第4期 2Ol4年 7月 CHINESEJOURNALOF COMPUTATIONAL PHYSICS Ju1．．2014 ArticleID：1001—246X(2014)04-0479-07 ParticleSwarm Optimization andGeneticAlgorithm f0raRelativisticBackwardW aveOscillator WANG Huihui， LIU Dagang， MENG Lin， LIU Laqun (SchoolofPhysicalElectronics，UniversityofElectronicScienceandTechnologyofChina，Chengdu610054，China) Abstract： Basedonplatform ofthree—dimensionalparticle-in—cell(PIC)，CHIPIC，modulesofparticleswarm optimization(PSO)andgeneticalgorithm (GA)aredesignedtooptimizearelativisticbackwardwaveoscillator (RBWO)，respectively．ComparisonsofPSO andGA areimplementedinthreekindsofparametersofRBWO： Continuousparameter，discreteparameter，andmixparameters．ItshowsthatperformancesofPSO arebetterthan thatofGA．PSO hashigheroptimizationaccuracyandconvergenceratethanGA． Keywords： particleswarm optimization；geneticalgorithm ；relativisticbackwardwaveoscillator；particlein—cell CLC number：0462；TN128；0224 Documentcode：A O Intr0ducti0n A relativisticbackwardwaveoscillator(RBWO) isoneofthemostimportantCherenkov radiati0nsource．especiallyinXband(8～12GHz) -53．Withrapiddevelopmentofcomputer， themethodofpartic1e．in—cellsimulation(PIC)hasbeenadoptedbymorescholarsinthedesignof RBW OE一 Severalrandom PIC simulationsarealmostimpossibletogetglobalbestsolution，as ． theef1-ectofslow-wavestructure (SWS)onRBWO performanceisnolinearanddifficulttobe estimatedaccuratelvbvtheory．Therefore，itismeaningfultoadoptan efficientglobaloptimization algorithmt0oDtimizeaRBWO assimulatedwithPIC．Geneticalgorithm (GA)L8Jandparticle swarm0ptimization(PSO)9 areefficientglobalalgorithmsusedinmanyfields¨-21]．Inthis w0rk，wecompareperformancesofGA andPSO forRBW O．We design aPSO moduleandaGA m0du1ebasedonPIC platform (CHIPIC ) t