简介:Particleswarmoptimizer(PSO),anewevolutionarycomputationalgorithm,exhibitsgoodperformanceforoptimizationproblems,althoughPSOcannotguaranteeconvergenceofaglobalminimum,evenalocalminimum.However,therearesomeadjustableparametersandrestrictiveconditionswhichcanaffectperformanceofthealgorithm.Inthispaper,thealgorithmareanalyzedasatime-varyingdynamicsystem,andthesufficientconditionsforasymptoticstabilityofaccelerationfactors,incrementofaccelerationfactorsandinertiaweightarededuced.Thevalueoftheinertiaweightisenhancedto(fi1,1).Basedonthededucedprincipleofaccelerationfactors,anewadaptivePSOalgorithm-harmoniousPSO(HPSO)isproposed.FurthermoreitisprovedthatHPSOisaglobalsearchalgorithm.Intheexperiments,HPSOareusedtothemodelidentificationofalinearmotordrivingservosystem.AnAkaikeinformationcriteriabasedfitnessfunctionisdesignedandthealgorithmscannotonlyestimatetheparameters,butalsodeterminetheorderofthemodelsimultaneously.TheresultsdemonstratetheeffectivenessofHPSO.
简介:TherearesomeadjustableparameterswhichdirecdyinfluencetheperformanceandstabilityofParticleSwarmOp-ttimizationalgorithm.Inthispaper,stabilitiesofPSOwithconstantparametersandtime-varyingparametersareanalyzedwithoutLipschitzconstraint.Necessaryandsufficientstabilityconditionsforaccelerationfactorψandinertiaweightwarepresented.Exper-imentsonbenchmarkfunctionsshowthegoodperfomanceofPSOsatisfyingthestabilitycondition,evenwithoutLipschitzcon-straint.Andtheinertiaweightwvalueisenhancedto(-1,1).