基于主元分析的多变量统计过程的故障辨识技术

Faults Distinguish Technology of Multivariate Statistical Process Based on Principle Component Analysis

为了更好的进行故障检测与诊断,介绍了主元分析理论,给出了基于主元分析的过程故障辨识机理及策略。仿真实例表明,利用此方法建立的故障诊断模型,能够在不依赖过程机理的前提下高效抽取原始数据空间的主要变化信息,对过程的非正常变化做出反应,同时还能较正确地找出发生故障的原因以及相应环节。

In this paper, the Principle Component Analysis theory is introduced. The mechanism and strategy of fault diagnosis based on this theory is presented. A simulation instance indicates a fault diagnosis model constructed with this approach can abstract efficiently the main variable information of original data set independent of process mechanism, and can respond to abnormal change of the process. It can find the cause of a fault and locate the fault exactly.