資源描述：
《多agent主從粒子群分布式計算框架_鄭宇軍》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、軟件學(xué)報ISSN1000-9825,CODENRUXUEWE-mail:jos@iscas.ac.cnJournalofSoftware,2012,23(11):3000?3008[doi:10.3724/SP.J.1001.2012.04305]http://www.jos.org.cn?中國科學(xué)院軟件研究所版權(quán)所有.Tel/Fax:+86-10-62562563?多Agent主從粒子群分布式計算框架1+121鄭宇軍,陳勝勇,凌海風(fēng),徐新黎1(浙江工業(yè)大學(xué)計算機科學(xué)與技術(shù)學(xué)院,浙江杭州310023)2(解放軍理工大學(xué)機械工程系,江蘇南京21000
2、7)Multi-AgentBasedDistributedComputingFrameworkforMaster-SlaveParticleSwarms1+121ZHENGYu-Jun,CHENSheng-Yong,LINGHai-Feng,XUXin-Li1(CollegeofComputerScienceandTechnology,ZhejiangUniversityofTechnology,Hangzhou310023,China)2(DepartmentofMechanicalEngineering,PLAUniversityofScien
3、ceandTechnology,Nanjing210007,China)+Correspondingauthor:E-mail:yujun.zheng@computer.org,http://www.zjut.edu.cnZhengYJ,ChenSY,LingHF,XuXL.Multi-Agentbaseddistributedcomputingframeworkformaster-slaveparticleswarms.JournalofSoftware,2012,23(11):3000?3008(inChinese).http://www.jo
4、s.org.cn/1000-9825/4305.htmAbstract:Toeffectivelysolvelarge-scaleoptimizationproblems,thepaperproposesadistributedagentcomputingframeworkbasedontheparallelparticleswarmoptimization(PSO).Theframeworkusesamasterswarmforevolvingcompletesolutionsoftheproblem,andusesasetofslaveswar
5、msforevolvingsub-solutionsofthesubproblemsconcurrently.ThemasterswarmandslaveswarmsalternativelyimplementthePSOproceduretoimprovetheproblem-solvingefficiency.Usingtheasynchronousteambasedagentarchitecture,amaster/slaveswarmconsistsofdifferentkindsofagents,whichshareapopulation
6、ofsolutionsandcooperatetoevolvethepopulation,suchasinitializingsolutions,movingparticles,handlingconstraints,anddecomposing/synthesizingsub-solutions.Theframeworkcanbeusedtosolvecomplicatedconstainedandmultiobjectiveoptimizationproblemsefficiently.Experimentalresultsdemonstrat
7、ethatthisapproachhassignificantperformanceadvantageovertwootherstate-of-the-artalgorithmsonatypicaltransportationproblem.Keywords:agent;particleswarmoptimization(PSO);master-slavemodel;cooperativeevolution;distributedcomputing摘要:面向大規(guī)模復(fù)雜優(yōu)化問題,提出了一個基于并行粒子群優(yōu)化的分布式Agent計算框架.框架中使用一個主
8、群(masterswarm)來演化問題的完整解,并使用一組從群(slaveswarm)來并行優(yōu)化一組子問題的解,主群和從群
