欢迎来到天天文库
浏览记录
ID:55570466
大小:1.24 MB
页数:7页
时间:2020-05-18
《位科技与创新管研讨会.pdf》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、2008數位科技與創新管理研討會論文編號:155數位科技組粒子群最佳化演算法求解分散式系統可靠度邱錦清德明財經科技大學資訊管理系摘要本論文提出一個粒子群最佳化演算法求解具容量限制之k-節點集合可靠度,以減少計算時間以及與正解之絕對誤差。此方法在粒子建構其k-節點的過程中,使用一個有效率的目標函數評估其所選擇之節點。由於只須一次的可靠度計算,所以用在可靠度計算的時間極少。而且,本演算法所得到的解,其相對於正解之絕對誤差比k-樹縮減法小。計算結果顯示,本方法對於較大的分散式系統比起傳統的方法有效率。關鍵字:粒子群最佳化演算法、分散式系統、可靠度最佳化1
2、1902008數位科技與創新管理研討會論文編號:155數位科技組ParticleswarmoptimizationfordistributedsystemreliabilityChin-ChingChiuDept.ofManagementInformationSystem,TakmingUniversityofScienceandTechnologyAbstractInthiswork,wepresentaparticleswarmoptimizationtoreducethecomputationaltimeandtheabsoluteerror
3、fromtheexactsolutionforobtainingthereliabilityofak-nodesetwithcapacityconstraint.Thismethodusesanefficientobjectivefunctiontoevaluatethechosennodeswhenaparticlederivesitsk-nodesetprocess.Reliabilitycomputationisperformedonlyonce,therebyspendinglesstimetocomputethereliability.M
4、oreover,theabsoluteerrorofproposedalgorithmfromexactsolutionissmallerthanthatofk-treereductionmethod.Computationalresultsdemonstratethattheproposedalgorithmisamoreefficientsolutionforalargedistributedsystemthanconventionalones.Keywords:particleswarmoptimization,distributedsyst
5、ems,reliabilityoptimization11912008數位科技與創新管理研討會論文編號:155數位科技組1.IntroductionParticleswarmoptimization(PSO)wasfirstintroducedbyKennedyandEberhartasanoptimizationtechniqueforcontinuousspaceproblemsin1995[4].Itisakindofevolutionarycomputationtechniquemotivatedbythebehavioroforganis
6、mssuchasbirdflocking.Becauseofitssimpleandeasyimplementation,PSOhasbeenwidelyappliedtooptimizationproblemswhichhavebeentackledbyotherevolutionarycomputationtechniquessuchasgeneticalgorithmsandevolutionaryprogramming.Thenetworkreliabilityproblemwithrespecttoanetworkwithageneral
7、structureisNP-hard[9].Efficientalgorithmseasilyimplementedonacomputerareneededtoanalyzethereliabilityoflargenetworks.Inaddition,suchalgorithmsshouldyieldgoodapproximationsofthereliabilitywhenthenetworksaresolargethatthecomputationaltimebecomesprohibitive.Thisworklargelyfocuses
8、onhowtocomputenearlymaximumsystemreliabilitysubjecttothecapac
此文档下载收益归作者所有