Fault Detection of DCS Central Control Hardware Devices Based on RTE-PCA

Abstract

DCS systems have been widely used in industrial production, but often the fault data of their central control systems are very complex and not conducive to troubleshooting and monitoring. Therefore, an excellent data dimensionality reduction method can accelerate the efficiency of fault diagnosis without losing accuracy. In this paper, we use Random Trees Embedding (RTE) algorithm to firstly expand the data with features to make the highly coupled data linearly separable, and then use PCA to reduce the dimensionality of the data with excellent results. Then, the effect of RTE-PCA algorithm with different parameters is discussed.