基于残差主元分析的多变量系统故障诊断方法及应用  

Multivariate System Fault Diagnosis Methods and its Applications Based on Residual Principal Component Analysis

作　　者：高会[1] 李彦平[1] 

机构地区：[1]沈阳大学装备制造综合自动化重点实验室,辽宁沈阳110044

出　　处：《沈阳大学学报（自然科学版）》2016年第2期141-146,共6页Journal of Shenyang University：Natural Science

摘　　要：通过建立多变量系统偏差自回归模型,提出一种基于残差加权主元分析的多变量系统故障诊断方法.这种方法既可消除偏差数据的自相关性,又可弥补其他主元分析所带来的缺陷.应用实例仿真结果表明,其具有良好的故障诊断效果.By establishing a self-regression model of error, a kind of fault diagnosis method is presented for multivariate system by residual weighted principal component analysis. This method can eliminate the self-correlation of error data, and makes up the defects of other principal component analysis. Simulation results of an application example show that the proposed method is effective in fault diagnosis of multivariate system.

关 键 词：多变量系统 故障诊断 偏差模型 残差加权 主元分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]