基于多智能体遗传算法的约束优化方法-研究

《基于多智能体遗传算法的约束优化方法-研究》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、AbstractIIIAbstractEvolutionaryAlgorithmprovidesanewwaytosolvecomplexoptimizationproblems.Becauseofitsintelligence,universality,robustnessandglobalsearchability,EAshavebeenwidelyusedinthisfieldandhaveagreatsuccessinrecentlyseveraldecades.Itiscommontofaceanumberofoptimization

2、problemsinmanyareasoftherealworld,especiallyinthescienceandengineeringfields.However,theseproblemsareoftenconstrained.Becauseofthedifferentfeaturesoftheseproblems,thetraditionalmethodsarehardtosolvetheseproblemseffectively.Asroustpopulation-basedglobalsearchmethods,Evolution

3、aryAlgorithms(EAs)areverypromisingtosolvetheconstrainedoptimizationproblems.TheaimofthisdissertationistoexplorethetheoriesandmechanismsofEAs,andtodothecorrespondingtheoreticandexperimentalanalyses.Themainresearchworkinthisdissertationconsistofthefollowingaspects.(1)Weextendt

4、hemultiagentgeneticalgorithm(MAGA)tosolveconstrainedoptimizationproblems(COPs)(MAGA_COPs)bycombiningtheneighborhoodcompetitionoperatorwithanefficientconstrainthandlingtechnique.Thismethodcanmakegooduseoftheinformationofinfeasiblesolutionswhichisaimatguidingthesearchtowardthe

5、globaloptimaofCOPs.Thisalgorithmistestedon12benchmarkfunctions,andtheresultshowsthat12benchmarkfunctionscanfindglobaloptima.(2)AnapproximationstrategyforfeasibleregionalsousedinMAGA_COPs,thismethodmakethesolutionoffunctionstoapproachtheglobaloptimalsolutionandeffectiveinprev

6、entingthealgorithmtrappingintolocaloptima.Thealgorithmistestedon12benchmarkfunctions,andtheresultshowsthatthealgorithmisoutperformsotherscomparedwithsomeotherstate-of-the-artalgorithms.(3)WeimproveMAGA_COPsbycombiningMAGA_COPswithtraditionalmethods.Hybridapproachbasedonmulti

7、agentgeneticalgorithmisgiven,inordertoovercometheslowerconvergenceofthestandardgeneticalgorithm,andinwhichthelocalsearchisweak.Thealgorithmistestedon12benchmarkfunctions,andtheresultshowsthatthealgorithmisanefficientandconvergenthybridgeneticalgorithm.(4)Weimprovethemultiage

8、ntgeneticalgorithm(MAGA)tosolvelayoutoptimizationbycombiningtheneighborhood