基于分解的多目标进化算法分析-analysis of multi-objective evolutionary algorithm based on decomposition.docxVIP

AbstractTheoptimizationproblemsinreallifetendtocopewithmultipletargets,forexample,time,cost,quality,quantity,etc.Evolutionaryalgorithmsareaclassofstochasticsearchalgorithm,bysimulatingthenaturalbiologicalevolutionandnaturalselectiontoguideanevolutionaryprocessofindividualsandtoaccumulatethedominantgenes.Evolutionaryalgorithmsareveryrobustandstrongadaptive,whichcanhelpustoavoidmuchlimitationswhensolvingproblemsdirectlyonlywithsomeclassicmathematicapproaches,thusevolutionaryalgorithmscandealwiththecomplexproblemsmoreeffectivelythantraditionaloptimizationmethodsfeeldifficulttosolve.What’smore,bigsuccessesofNSGA-II[1],anexcellentevolutionaryalgorithmbasedontheParetodominancemethod,giveresearchersafalseimpressionthattraditionalmathematicmethodsinoptimizationdomainseemoutofdate.Fortunately,inrecentyears,methodsbasedonscalarfunctionsanddecompositionapproachesbegintochangethissituationandbecomeoneofmostnovelaswellasattractivetrendsoriginatingfromtraditionalmathematictheory,thenaffectsfurthertotheentiremulti-objectiveoptimizationdomain.Asresult,tofindinspirationsamongtraditionalprinciplesandapproacheshasbecomeanewpath.Usually,thesub-objectivesamongonemulti-objectiveoptimizationproblemareconflictivebetweeneachother.Inthemeanwhile,optimizationalgorithmisaimingtofindasetofcompromisedsolutionstotheeachsub-problem,i.e.,thesolutionsnamedParetooptima.Supposedtobegivenanaggregativefunctionandsomeorderswithseveralpreferences,amulti-objectiveproblemcanbeconvertedtoasingle-objectiveoptimizationproblemthatmeansmanysuccessfulsingle-objectiveoptimizationalgorithmscanbeeasilyadoptedtoexploreoptimalsolutions.However,anidealtrade-offrelationforeachsub-objectivecannotbepredetermined,thereforeanapproachcanprovidethebestcandidatesolutionsandchoosepreferencesamongthemisnecessary.Decomposition-basedmulti-objectiveevolutionaryalgorithm(MOEA/D[2])transformsamulti-objectiveoptimizationproblemintoasetofsingle-objectiveandassignssuitableindividualstotheeachsub-problemproblemsviaasetofpre-generatedunif