上海交通大学学报

上海交通大学学报（自然版）

• 机械工程 • 上一篇    下一篇

基于路径图的多元统计过程诊断

陈文多，苗瑞，赵言正，江志斌
  

  1. (上海交通大学 机械与动力工程学院，上海 200240)
  • 收稿日期:2009-11-04 修回日期:1900-01-01 出版日期:2010-12-31 发布日期:2010-12-31

Multivariate Statistical Process Diagnosis Based on Path Diagram

CHEN Wenduo ,MIAO Rui ,ZHAO Yanzheng,JIANG Zhibin   

  1. (School of Mechanical Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2009-11-04 Revised:1900-01-01 Online:2010-12-31 Published:2010-12-31

摘要： 针对多元统计过程控制图中失控信号的诊断问题，提出了基于路径图的多元统计过程诊断方法.利用路径图给出的变量之间定量化的因果关系，对所有变量分别建立对应的线性回归方程，以得到联立方程模型.通过联立方程模型建立了统一的T2统计量分解方式，并构建了分解式中所有分解项的判定界限，将其作为判定引起失控信号产生的根本原因的标准.通过对三元系统仿真，验证了其有效性.

关键词: 路径图, 多元统计过程, 诊断, T2分解

Abstract: In view of the diagnosis of outofcontrol signals in multivariate statistical process control charts, a method of multivariate process diagnosis based on path diagram was proposed. Utilizing the quantitative causation among variables given by the path diagram, a simultaneous equation model could be obtained by establishing the linear regression equations for all variables independently. Then, the simultaneous equation model was applied to derive a uniform decomposition for T2 statistic. And the control limits for all decomposition terms were given as the standards of judging the root causes contributing to the outofcontrol signals. Simulation studies for a system consisting of three variables were carried out to demonstrate that the proposed method is reasonable.

中图分类号: