基于2DPCA理论的间歇过程故障诊断方法  

Fault Diagnosis Method for Batch Process based on 2DPCA Theory

作　　者：王国柱 李晶晶 付嘉良 杜志勇 许玲 WANG Guo-zhu(Henan Institute of Technology,Xinxiang 453003,China)

机构地区：[1]河南工学院

出　　处：《河南机电高等专科学校学报》2019年第4期1-5,共5页Journal of Henan Mechanical and Electrical Engineering College

基　　金：河南省科技攻关项目基金资助项目（182102210033)

摘　　要：提出了一种基于2DPCA理论的间歇过程故障诊断方法,该方法采用各个批次数据协方差矩阵的平均值进行PCA建模,避免了三维数据二维化转换后维数过高的影响,能够更加准确地检测到故障的存在,在一定程度上解决了过程故障漏报的现象。针对RTA间歇反应过程,分别运用MPCA和2DPCA方法进行了大量的仿真实验,证明2DPCA方法在间歇过程故障诊断方面具有一定的优势。In this paper,a novel fault diagnosis method for batch process based on two-dimensional principal component analysis(2DPCA)theory is proposed.This method uses the average value of co-variance matrix of each batch data to build the PCA model,avoids the influence of high dimension after converting three-dimensional data into two-dimensional data,and can detect faults more accurately.In this paper,the MPCA and DPCA methods are used to simulate RTA batch reaction process,which proves that 2DPCA method has certain advantages in fault diagnosis of batch process.

关 键 词：间歇过程 故障诊断 多向主元分析 二维主元分析 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]