资源描述：
《Data-driven fuzzy clustering based on maximum entropy》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、Availableonlineatwww.sciencedirect.comExpertSystemswithApplicationsExpertSystemswithApplications36(2009)625–633www.elsevier.com/locate/eswaData-drivenfuzzyclusteringbasedonmaximumentropyprincipleandPSOa,b,*aDebaoChen,ChunxiaZhaoaSchoolofComputerScienceandTechnology,Nanjin
2、gUniversityofScienceandTechnology,Nanjing210094,ChinabPhysicalDepartment,HuaibeiCoalIndustryTeachersCollege,Huaibei235000,ChinaAbstractToidentifytheoptimumfuzzyrulebaseisthemajordifficultyindesigningfuzzymodel.Todesignoptimumfuzzyrulebase,whichistraditionallyachievedbytedio
3、ustrialanderrorprocess,fromnumericaldata,anoveldata-drivenfuzzyclusteringmethodbasedonmax-imumentropyprinciple(MEP)andparticleswarmoptimization(PSO)isproposed.Inthisalgorithm,themembershipsofoutputvari-ablesareinferredbymaximumentropyprinciple,andthecentersoffuzzyrulebase
4、areoptimizedbyPSO.ComparingwiththemethodthatdesigningfuzzyrulebaseonlybyPSOorotherevolutionarycomputationmethods,thenumberofparameterstobeoptimizeddecreasedgreatly,andthecomputationcostdeclined.Tochecktheeffectivenessofthesuggestedapproach,threeexamplesformodelingareexamin
5、edcomparingwiththemethodonlyusingPSO.Theperformanceoftheidentifiedfuzzymodelsisdemonstrated.PublishedbyElsevierLtd.Keywords:Particleswarmoptimization(PSO);Maximumentropyprinciple(MEP);Nonlinearsystem;Fuzzymodeling1.Introductionnumericaldatahavebeenproposedinthebibliography
6、toovercometheproblemofknowledgeacquisition(Dick-Fuzzymodelingisusuallyusedtodealwithnonlinearerson&Kosko,1996;Jang,Sun,&Mizutani,1997;Liusystemwhentheprecisemathematicsmodelisdifficultto&Hu,1992).Theinputdomainisdividedintoafixedcon-setup.Inpracticalapplications,amajordifficul
7、tyforfuzzyfigurationofmembershipfunctions.Theparametersformodelingishowtofindanappropriatefuzzyrulebaseforfuzzyrulesarecomposedofantecedentpartandconse-agiventask.Atrialanderrormethodisusedinmostcasesquencepart,antecedentpartofthefuzzyrulesistobepre-butwithaheavycomputation
8、burdenandlowefficiency.definedandthentheconsequencepart.Inthesemethods,Toderivefuzzyrulesautomatica
