资源描述:
《基于协同进化的动态多目标优化算法》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、摘要关键词:协同进化动态多目标优化进化算法粒子群优化预测模型万方数据ABSTRACTABSTRACTInreallife,therearemanyfieldsinvolvingalotofstaticanddynamicoptimizationproblems,suchasengineeringdesign,networkcommunication,capitalbudgeting,urbantransportation,datamining,andsoon.Meanwhile,theseoptimizati
2、onproblemshavemulti-objectivefunctions.Atpresent,therearealotofmaturemulti-objectiveevolutionaryalgorithms.Untilrecentyears,therearemoreandmoreresearchersinterestingindynamicoptimizationproblems.Incontrast,thenumberofresearchesondynamicmulti-objectiveoptimi
3、zationproblems(DMOP)ismuchsmaller.Basedontheabovebackground,thispaperproposesdynamicmulti-objectiveoptimizationalgorithmsbasedoncoevolution.Thespecificworkisarrangedasfollows:1.CoevolutionaryTechniquebasedMulti-swarmParticleSwarmOptimizerforDynamicMulti-obj
4、ectiveOptimization.Intheproposedalgorithm,thenumberofparticleswarmsisdeterminedbythenumberoftheobjectivefunctions,whichmeansthatonefunctioncorrespondoneparticleswarm.Thenmultipleparticleswarmssolveproblemcooperativelyusinginformationsharingstrategy.Inordert
5、oimprovetheperformanceofthealgorithmunderthedynamicenvironment,anovelvelocityupdateandspecificboundaryconstraintsareadoptedtomakesurethefastconvergencespeed.Thesimilaritydetectiveoperatorisusedtodetectachangeinenvironmenteffectively,andthensuitabledynamicme
6、chanismrespondstothechangeappropriatelytoimprovethediversity.Furthermore,thisalgorithmiscomparedagainstthreeotherstate-of-the-artdynamicmulti-objectiveoptimizationalgorithms.TheexperimentalresultsshowthatthisalgorithmisveryexcellenttosolveDMOP.2.ANovelCoope
7、rativeCoevolutionaryDynamicMulti-objectiveOptimizationAlgorithmUsingaNewPredictiveModel.Intheproposedalgorithm,thebasicideaofdecompositionisinaccordancewiththesearchspaceofdecisionvariables,whichdividesaproblemsolutionintodifferentsubpopulations.Throughshar
8、ingthebenefit,restricting,coordinatingandevolvingwitheachother,subpopulationscanfindParetooptimalsettogether.Thiswayimprovestheconvergenceperformanceofthealgorithm.Thenthesimilaritydetectiveoperatorisu