进化多目标优化算法研究.pdfVIP

ISSN1000-9825，CODENRUXUEWE—mail：jos@iscas．ac．ca

JournalofSoftware，Vo1．20，No．2，February2009，pp．271-289http：／w／ww．jos．org．cn

doi：l0．3724／SPJ．1001．2009．03483Te1／Fax：+86．10

@byInstituteofSofwtare．theChineseAcademyofSciences．Allrightsreserved．

进化多目标优化算法研究

公茂果l2+，焦李成1,2，杨咚咚1,2马文萍1,2

(西安电子科技大学智能信息处理研究所，陕西西安710071)

(西安电子科技大学智能感知与图像理解教育部重点实验室，陕西西安710071)

ResearchonEvolutionaryMulti—ObjectiveOptimizationAlgorithms

GONGMao—Guo’，JIAOLi—Cheng’，YANGDong．Dong，，MAWen．Ping，

(InstituteofIntelligentInformationProcessing，XidianUniversity，Xi’an710071，China)

(KeyLaboratoryofIntelligentPerceptionandImageUnderstandingfortheMinistryofEducation，XidianUniversiyt，Xi’an710071

China)

+Correspondingauthor：E-mail：gong@ieee．org

GongMG，JiaoLC，YangDD，MaWP．Researchonevolutionarymulti-objectiveoptimizationalgorithms．

JournalofSoftware，2009，20(2)：271—289．hup：／／www．jos．org．cn／1000—9825／3483．htm

Abstract：Evolutionarymulti—objectiveoptimization(EMO)，whosemaintaskistodealwithmulti-objective

optimizationproblemsbyevolutionarycomputation，hasbecomeahottopicinevolutionarycomputation

community．AftersummarizingtheEMOalgorithmsbefore2003briefly,therecentadvancesinEMOarediscussed

indetails．Thecurrentresearchdirectionsareconcluded．Ontheonehand，morenewevolutionaryparadigmshave

beenintroducedintoEMOcommunity,suchasparticleswarmoptimization，artificialimmunesystems，and

estimationdistributionalgorithms．Ontheotherhand，inordertodealwithmany·objectiveoptimizationproblems，

manynewdominanceschemesdifferentfromtraditionalPareto—dominancecomeforth．Furthermore，theessential

characteristicsofmulti-objectiveoptimizationproblemsaredeeplyinvestigated．Thispaperalsogivesexperimental

comparisonofseveralrepresentativealgorithms．Finally,severalviewpointsforthefutureresearchofEMOare

p