一类基于species机制的多目标进化算法分析

《一类基于species机制的多目标进化算法分析》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、万方数据东北大学硕士学位论文AbstractResearchesonaClassofSpecies-—BasedEvolutionaryAlgoriOptimizationProblemsAbstractAlotofapplicationproblemsinthescienceandengineeringapplicationsusuallyinvolvemultipleobjectiveswhichoftenconflictwitheachother．Thesemulti-objectiveoptimiza

2、tionproblems(MOOPs)areratherdifferentfromthetraditionaloptimizationproblemswithsingleobjectiveduetoinexistenceoftheglobaloptimalsolutionwhichcanoptimizeallobjectivesimultaneously．Obviously,suchthiscomplexfeaturewouldposeagreatchallengetothesolutionalgorithm

3、forMOOPs．Evolutionaryalgorithms(EAs)havebeenwidelyappliedintoaddressingMOOPssincethepopulation—basedmechanismenablesanEAachievemultipleParetooptimumsinitssinglerunningcourse，whichisjustthemajorreasonthatevolutionarymulti—objectiveoptimizationhasbecomeanewly

4、-generatedfocusfromthecommunityofevolutionarycomputationinrecentyears．ItisnoticeablethatsearchingmultipleoptimalsolutionsinparallelisnotsuchachallengeonlywhenanEAisusedtoaddressMOOPs．Whenaddressingmulti—modaloptimizationproblems，EAisalsorequiredtoachieveasm

5、anyaspossibleglobaloptimumevensomelocaloptimalsolutions．Therefore，itbecomesaveryinterestingresearchissuetointroducethealgorithmmechanism，whichwasusedinEAsformulti-modalproblems，intothedesignofEAsforMOOPs．Basedonthemechanismofsystemengineering，thisthesiswill

6、investigateandstudyaclassofnewmulti—objectiveEAsthatareinspiredfromtheSpecies-basedscheme，whicharefirstlyproposedforEAsinmulti-modaloptimization．Themaincontentsofthisthesiscanbesummarizedasfollows．Firstly,twotypicalEAs，thatis，geneticalgorithm(GA)andparticle

7、swarmoptimization(PSO)，areintroducedtheiralgorithmprinciplesinbriefandtherelevantresearchworksonGAandPS0forMOOPsarereviewed．Secondly,anewspecies—basedmulti-objectiveGA(speMOGA)，whichcombinesthe—III—万方数据东北大学硕士学位论文Abstractspecies·-basedschemeandthemechanismof

8、NSGA·-IIthatwasawell·-knownmulti—objectiveGA，isproposedforM00Ps．Intheproposedalgorithm，aspeciesseeddeterminationmethodbasedontwonewfeaturesofindividualsandaspeciesconstructionmethodwhereaspeciesisadapt