改进的PCA方法在化工过程故障诊断中的应用研究  

Application of an Improved PCA Method to Chemical Fault Diagnosis

作　　者：任伟[1] 田文德[1] 杜廷召[1] 

机构地区：[1]青岛科技大学化工学院,山东青岛266042

出　　处：《化工自动化及仪表》2010年第6期20-23,28,共5页Control and Instruments in Chemical Industry

基　　金：山东省自然科学基金资助项目(ZR2009BM033)

摘　　要：主元分析(PCA)方法作为一种基于信号处理的数据驱动方法,已广泛应用于工业过程故障诊断中。但由于PCA本身的缺陷,对故障类型的识别方面存在不足。为此,引入人工神经网络(ANN)方法,与PCA方法相结合,增强了故障识别能力。以田纳西伊斯曼过程为例,验证了改进后的PCA方法在化工过程故障识别和诊断中应用的有效性。As a data-driven dustrial processes. But there method based on signal processing, PCA had been widely used in fault diagnosis of inwere some shortcomings for the identification of fault types because of PCA＇ s own flaws. In order to enhance the ability of faults identification, ANN was combined with PCA. Taking Tennessee Eastman process as an example, the effectiveness of the improved PCA method in chemical process fault identification and diagnosis was verified.

关 键 词：故障诊断 田纳西-伊斯曼过程 PCA 神经网络 化工过程 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]