%0 Journal Article
%T 一类基于数据的解释性模糊建模方法的研究
%A 邢宗义
%A 贾利民
%A 张永
%A 胡维礼
%A 秦勇
%J 自动化学报
%P 815-824
%D 2005
%X ？Anapproachtoidentifyinterpretablefuzzymodelsfromdataisproposed.Interpretability,whichisoneofthemostimportantfeaturesoffuzzymodels,isanalyzedfirst.Thenumberoffuzzyrulesisdeterminedbyfuzzyclustervalidityindices.Amodifiedfuzzyclusteringalgorithm,combinedwiththeleastsquaremethod,isusedtoidentifytheinitialfuzzymodel.Anorthogonalleastsquarealgorithmandamethodofmergingsimilarfuzzysetsarethenusedtoremovetheredundancyofthefuzzymodelandimproveitsinterpretability.Next,inordertoattainhighaccuracy,whilepreservinginterpretability,aconstrainedLevenberg-Marquardtmethodisutilizedtooptimizetheprecisionofthefuzzymodel.Finally,theproposedapproachisappliedtoaPHneutralizationprocess,andtheresultsshowitsvalidity.
%K Fuzzymodeling
%K interpretability
%K fuzzyclustering
%U http://www.aas.net.cn/CN/abstract/abstract15937.shtml