基于奇异值分解的故障检测技术及其应用  

Fault Detection Technology Based on Singular Value Decomposition (SVD)

作　　者：宋立辉[1] 姜兴渭[1] 

机构地区：[1]哈尔滨工业大学

出　　处：《中国机械工程》2003年第24期2090-2093,共4页China Mechanical Engineering

基　　金：国家863高技术研究发展计划资助项目(2002AA721063)

摘　　要：针对基于未知输入观测器的诊断方法在诊断多故障时具有局限性,提出了一种基于奇异值分解的诊断方法,这种方法通过奇异值分解将不同故障对系统残差的影响进行分离,给出了多故障检测与分离的方法,仿真证明这种方法对于多故障诊断有很好的效果。The limitation of fault diagnosis technolo-gy based on unknown input observer (UIO) had been ex-plored in the process of multi--fault diagnosis. In orderto solve this problem, a new method based on singularvalue decomposition (SVD) was put forward in this pa-per. It can remove the influences from residual of speci-fied faults through SVD, a fault isolation method is giv-en. It is proved that the proposed method is very effec-tive to multi--fault detection and isolation (FDI) throughdigital simulations.

关 键 词：未知输入观测器 奇异值分解 故障检测 多故障诊断 

分 类 号：TP206[自动化与计算机技术—检测技术与自动化装置] V47[自动化与计算机技术—控制科学与工程]