The Fault Detection And Diagnosis Of Sensors Based On PCA

Research of fault detection based-on principal component analysis (PCA) is one ofthe important branches in fault detection domain. This article first introduced the theoryof principal component analysis, and then analyzed the characteristic of the power plantprocess control. Based on this characteristic the article analyzed how to use PCA tomonitor and diagnosis the sensors of the power plant process. Use the practicalapplication result of the traditional PCA, this article has analyzed the flaw of traditionalmethod and the improvement direction, and then proposed a new strategy for faultisolation. In the end use the comparison of this method and the traditional method provesthe validity and advance of the new method. The second part of this article discuss theflaw of the traditional method when use it monitor the nonlinear system. Then proposedtwo methods based on PCA to monitor the nonlinear system and use the simulation toprove the validity of the two methods, the end summarize the characteristic of the twomethods.