面向模糊多级特征的F-MEWMA控制图研究  

F-MEWMA control chart for fuzzy hierarchical quantitative characteristics

作　　者：周娟[1] 余忠华[1] 侯智[1] 

机构地区：[1]浙江大学浙江省先进制造技术重点研究实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2016年第7期1373-1380,共8页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(50835008;71071138)

摘　　要：围绕难以定量测量、但可以分级量化的质量特性,在实施统计过程控制中所遇到的检出力受限问题展开讨论,提出模糊多变量指数加权移动平均控制图(F-MEWMA)方法.该方法借助于模糊理论,对分级量化特征进行模糊化处理.针对不同α截集包含信息量多寡的问题,提出应用加权α截集模糊区间值来构造统计量,分别针对模糊单变量和模糊多变量质量特性进行数学表征与比较分析.采用Matlab仿真的方式确定不同权重系数和不同维数下的控制限,完成F-EWMA、F-MEWMA的设计.采用Matlab仿真的方法,以识别变异的概率为评价指标对FMEWMA的监控效果进行分析.以电能表潜动和起动的质量特性为例,对提出的方法进行应用,取得了较好的应用效果.Fuzzy multivariate exponentially weighted moving average control chart （F-MEW MA） was proposedin order to solve the problem of lower power when implementing statistical process control to thequality characteristic which was difficult to quantify but could be described in hierarchical levels.F-MEWMA tried to blur the hierarchical quantitative characteristics based on fuzzy theory. Since differenta cut sets containing different amount of information, interval value of weighted fuzzy a cut set was proposed.Mathematical characterization and comparison of the quality characteristics for fuzzy univariate dataand fuzzy multivariate data were conducted respectively, and the control limits for different weight coefficientsand different dimensions were determined by Matlab simulation. Then F-EWMA and F-MEWMAwere designed. The effect of F-MEWMA was analyzed by calculating probability of identifying variation byMatlab simulation. The implementation of the F-MEWMA on the electric energy meter running and startingquality characteristics showed good result.

关 键 词：模糊多级特征 模糊理论 模糊多变量指数加权移动平均控制图(F-MEWMA) 加权α截集模糊区间值 识别变异的概率 

分 类 号：O213[理学—概率论与数理统计]