摘要
在现代生产制造系统中,快速检测制造过程产品质量特性的偏移,对于确保过程稳定运行具有重要作用。基于实时对比(Real Time Contrasts,RTC)方法,以随机森林作为分类器,分别构造了多元过程监控的休哈特(Shewhart)类控制图和指数加权移动平均(Exponentially Weighted Moving Average,EWMA)控制图。最后通过仿真实验,对10维和100维多元正态分布过程和2维的非正态过程监控的平均运行链长(Average Run Length,ARL)进行了比较分析,结果表明,基于RTC方法的EWMA控制图在监控高维复杂数据偏移上具有一定的优势。
Detection of product quality characteristic changes is significant to ensure the stable operation of multivariate process in modern manufacturing industries. This paper uses RTC method and applies random forest as classifying tool to construct Shewhart-type control chart and EWMA control chart of multivariate process monitoring respectively. Then, numerical examples are used to calculate and analyze ARLs of 10＆100 dimensional normal data and 2 dimensional non-normal data. Results show that the method of EWMA control chart combining RTC statistic is effective and better than other methods in monitoring high dimensional complex data.
出处
《组合机床与自动化加工技术》
北大核心
2014年第1期57-60,63,共5页
Modular Machine Tool & Automatic Manufacturing Technique
关键词
多元过程控制
实时对比
随机森林
平均运行链长
EWMA控制图
multivariate statistical process control, real time contrasts, random forest, average run length, EWMA control chart
