基于DICA的故障检测和诊断  

Fault Detection and Diagnosis Based on DICA

作　　者：郭金玉 王乐 李元 GUO Jinyu;WANG Le;LI Yuan(College of Information Engineering,Shenyang University of Chemical Technology,Shenyang 110142,China)

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《沈阳大学学报（自然科学版）》2022年第4期290-297,共8页Journal of Shenyang University：Natural Science

基　　金：辽宁省教育厅项目(L2019007)。

摘　　要：针对在大多数的工业过程中,观测到的统计值具有动态和非线性的问题,提出了一种基于DICA的故障检测和诊断方法。首先,构造一个具有时滞变量的增广矩阵。然后,对处理后的数据应用独立成分分析(ICA)方法,提取在统计上独立的非高斯信号源。最后,计算I^(2)、I_( e)^(2)和Q统计量监控样本状态。为了识别故障变量,基于灵敏度分析的思想构造用于监测统计的贡献图,能够有效准确地定位故障变量。将该方法应用于田纳西-伊斯曼过程,仿真结果表明,动态独立成分分析(DICA)能有效地检测多变量动态过程中的故障。同时也表明,所提出的故障诊断方法在定位故障变量上是有效的。A fault detection and diagnosis method based on DICA was proposed for the dynamic and nonlinear characteristics observed in most industrial processes.First,an augmented matrix with time-delay variables was constructed,the augmented matrix was normalized.Then independent component analysis(ICA)was applied to the processed data to extract statistically independent non-Gaussian signal sources.Finally,I^(2)、I_( e)^(2) and Q statistics were calculated to monitor the sample state.To identify the fault variables,a contribution graph based on sensitivity analysis was constructed for monitoring statistics,which could locate the fault variables effectively and accurately.The proposed method was applied to the Tennessee-Eastman process.Simulation results show that dynamic independent component analysis(DICA)can effectively detect the fault in the multivariable dynamic process.At the same time,the simulation results show that the proposed fault diagnosis method is effective in locating fault variables.

关 键 词：独立成分分析 动态独立成分分析 故障检测 故障诊断 田纳西-伊斯曼过程 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]