摘要
Ahybridapproachforfuzzysystemdesignbasedonclusteringandakindofneurofuzzynetworksisproposed.Anunsupervisedclusteringtechniqueisfirstlyusedtodeterminethenumberofif-thenfuzzyrulesandgenerateaninitialfuzzyrulebasefromthegiveninput-outputdata.Then,aclassofneurofuzzynetworksisconstructedanditsweightsaretunedsothattheobtainedfuzzyrulebasehasahighaccuracy.Finally,twoexamplesoffunctionapproximationproblemsaregiventoillustratetheeffectivenessoftheproposedapproach.
出版日期
2004年03月13日(中国期刊网平台首次上网日期,不代表论文的发表时间)