摘要
在传统动态核主元分析(DKPCA)中引入循环时间算法(CTA),提出了基于循环时间算法与动态核主元分析(CTA-DKPCA)的故障诊断方法。该方法通过引入循环时间逻辑,对动态数据矩阵分段捕捉故障信息,提升了故障诊断精度与速度,再使用基于重构贡献(RBC)的方法来识别故障变量、诊断故障原因。将该方法应用于田纳西伊斯曼(TE)过程进行故障诊断分析,与PCA及其衍生算法和2阶SVM算法比较,结果表明CTA-DKPCA模型对故障的检测与变量识别具有较高的敏感度。
Introduce the cycle temporal algorithm(CTA)into the traditional dynamic kernel principal component analysis(DKPCA),and propose a fault diagnosis method based on the cycle temporal algorithm and dynamic kernel principal component analysis(CTA-DKPCA).By introducing cycle temporal logic,this method captures fault information in sections from the matrix of dynamic data,improves the accuracy and speed of fault diagnosis.Then the reconstruction-based contribution(RBC)method is used to identify fault variables and diagnose fault causes.The proposed method is applied to the fault diagnosis of Tennessee Eastman(TE)process,and compared with the PCA and its derivative algorithms and the 2-class SVM algorithm.The results show that the CTA-DKPCA model is highly sensitive to fault detection and variable identification.
作者
张佳鑫
罗文嘉
戴一阳
ZHANG Jia-xin;LUO Wen-jia;DAI Yi-yang(School of Chemistry and Chemical Engineering,Southwest Petroleum University,Chengdu 610500,China;School of Chemical Engineering,Sichuan University,Chengdu 610065,China)
出处
《控制工程》
CSCD
北大核心
2021年第5期844-850,共7页
Control Engineering of China
基金
国家自然科学基金资助项目(21706220)。
关键词
故障诊断
动态核主元分析
重构贡献
变量识别
Fault detection
dynamic kernel principal component analysis
reconstruction-based contribution
variable identification
