隐马尔可夫模型的TE过程故障诊断  

Hidden Markov Model Based on Fault Diagnosis for TE Process

作　　者：刘小雍[1] 周淑芳[2] 熊中刚[1] 张慧 LIU Xiao-yong;ZHOU Shu-fang;XIONG Zhong-gang;ZHANG Hui(College of engineering and technology, Zunyi Normal College, Zunyi 563002, China)

机构地区：[1]遵义师范学院工学院,贵州遵义563002 [2]遵义医学院附属医院医学检验科,贵州遵义563003 [3]上海新时达机器人有限公司,上海201801

出　　处：《遵义师范学院学报》2016年第5期92-94,130,共4页Journal of Zunyi Normal University

基　　金：遵义师范学院博士基金资助项目(遵师BS(2015)04号);贵州省教育厅基金资助项目(黔教合KY字[2015]457号);贵州省科技厅基金资助项目(黔科合LH字[2015]7054号)

摘　　要：对基于隐马尔可夫模型(HMM,Hidden Markov Model)的故障诊断方法在田纳西-伊斯曼(TE)过程中的应用进行了研究,实现了故障模式的识别以及退化状态的评估。利用主元分析法对观测变量的维数进行降维,并结合滑动窗口技术,提高了诊断的实时性,同时在保存原数据主要信息的同时大大减少了计算量。The fault diagnosis method based on Hidden Markov Model (HMM) and its application in Tennessee Eastman (TE) process are studied, which is to identify the fault pattern and assess the current degradation state. Principle component analysis (PCA) is adopted to reduce the dimension of observed variables in the fault diagnosis based on HMM, which maintains the main feature of the original data and relieves the computational burden greatly. Moving window is employed to track the dynamic data and simplify the calculation in the test of the fault diagnosis.

关 键 词：故障诊断 隐马尔可夫模型 TE过程 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]