资源描述：
《外文翻译-微分优化算法与对于全球数值优化的适应策略》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、NORIliWtXrERNVUbVTEOiNK.UIMVERSiTYMINGDEC()1.LE(；E英文翻译系别专业班级学生姓名学号指导教师DifferentialEvoIutionAlgorithmWithStrategyAdaptationforGIobaINumericalOptimizationA.K.Qin,V.L.Huang,andP.N.SuganthanAbstract—Differentialevolution(DE)isanefficientandpowerfulpopulation-basedstochast
2、icsearchtechniqueforsolvingoptimizationproblemsovercontinuousspace,whichhasbeenwidelyappliedinmanyscientificandengineeringfields.However,thesuccessofDEinsolvingaspecificproblemcruciallydependsonappropriatelychoosingtrialvectorgenerationstrategiesandtheirassociatedcont
3、rolparametervalues・Employingatrial-and-crrorschemetosearchforthemostsuitablestrategyanditsassociatedparametersettingsrequireshighcomputationalcosts.Moreover,atdifferentstagesofevolution,differentstrategiescoupledwithdifferentparametersettingsmayberequiredinordertoachi
4、evethebestperformance・Inthispaper,weproposeaself-adaptiveDE(SaDE)algorithm,inwhichbothtrialvectorgenerationstrategicsandtheirassociatedcontrolparametervaluesaregraduallysell-adaptedbylearningfromtheirpreviousexperiencesingeneratingpromisingsolutions.Consequently,amore
5、suitablegenerationstrategyalongwithitsparametersettingscanbedeterminedadaptivelytomatchdi氓rentphasesofthesearchprocess/evolution.TheperfonnanceoftheSaDEalgorithmisextensivelyevaluated(usingcodesavailablefromP.N.Suganthan)onasuiteof26bound-constrainednumericaloptimizat
6、ionproblemsandcomparesfavorablywiththeconventionalDEandseveralstate-of-the-artparameteradaptiveDEvariants.IndexTerms一Differentialevolution(DE),globalnumericaloptimization,parameteradaptation,self-adaptation,strategyadaptation.I.INTRODUCTIONEVOLUTIONARYALGORITHMS(EAs),
7、inspiredbythenaturalevolutionofspecies,havebeensuccessfullyappliedtosolvenumerousoptimizationproblemsindiversefields.However,whenimplementingtheEAs,usersnotonlyneedtodeterminetheappropriateencodingschemesandevolutionaryoperators,butalsoneedtochoosethesuitableparameter
8、settingstoensurethesuccessofthealgorithm,whichmayleadtodemandingcomputationalcostsduetothetime-consumingtrial-and-errorparam
