基于可分性测度的模糊隶属函数确定方法  

Determination method of fuzzy membership function based on separability measure

作　　者：董炜[1] 陈卫征[1,2] 徐晓滨[3] 吉吟东[1] 

机构地区：[1]清华信息科学与技术国家实验室,北京100084 [2]清华大学自动化系,北京100084 [3]杭州电子科技大学自动化学院,杭州310018

出　　处：《控制与决策》2014年第11期2089-2093,共5页Control and Decision

基　　金：国家自然科学基金项目(61104019;61004070);国家科技支撑计划课题项目(2009BAG12A08);清华大学自主科研计划项目

摘　　要：隶属程度的思想是模糊数学的基本思想,应用模糊数学方法解决实际问题的关键在于建立符合实际的隶属函数,然而,如何正确地确定隶属函数仍是至今尚未完全解决的问题.鉴于此,提出一种基于可分性测度的隶属函数确定方法,利用类间在各个特征上的可分性确定模糊集的划分,进而确定描述该模糊集的隶属函数.通过轨道电路故障诊断实验表明了所提出方法的有效性.The idea of the degree of membership is the basic idea of fuzzy mathematics.The key to solve practical problems using fuzzy mathematical method lies in establishing the actual membership function.However,how to correctly determine the membership function is still not yet fully resolved so far.Therefore,a method for determining the membership function based on the separability measure is presented.This method can determine the division of fuzzy sets by using the separability on various features between any two categories,and then determine the membership function to describe each fuzzy set.After that,an experiment is carried out in the track circuit fault diagnosis,and the result shows the effectiveness of the proposed method.

关 键 词：可分性测度 隶属函数 模糊集 模糊推理 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]