简介:Acceleratingtheconvergencespeedandavoidingthelocaloptimalsolutionaretwomaingoalsofparticleswarmoptimization(PSO).TheverybasicPSOmodelandsomevariantsofPSOdonotconsidertheenhancementoftheexplorativecapabilityofeachparticle.Thusthesemethodshaveaslowconvergencespeedandmaytrapintoalocaloptimalsolution.Toenhancetheexplorativecapabilityofparticles,aschemecalledexplorativecapabilityenhancementinPSO(ECE-PSO)isproposedbyintroducingsomevirtualparticlesinrandomdirectionswithrandomamplitude.Thelinearlydecreasingmethodrelatedtothemaximumiterationandthenonlinearlydecreasingmethodrelatedtothefitnessvalueofthegloballybestparticleareemployedtoproducevirtualparticles.TheabovetwomethodsarethoroughlycomparedwithfourrepresentativeadvancedPSOvariantsoneightunimodalandmultimodalbenchmarkproblems.ExperimentalresultsindicatethattheconvergencespeedandsolutionqualityofECE-PSOoutperformthestate-of-the-artPSOvariants.