A Hybrid Fault Detection and Diagnosis System Based on KPCA and DDAG

In order to improve the capability of the fault diagnosis, this paper introduces the Decision Directed Acyclic Graph (DDAG) algorithm and establishes a new detection and diagnosis system combing the DDAG with the Kernel Principal Component Analyses (KPCA) method. The hybrid system uses KPCA and DDAG to detect and identify the fault. A specific description of the principles and procedures about how to use KPCA method and DDAG is given. The new detection and diagnosis system has an excellent performance in the fault detection and diagnosis of the Tennessee-Eastman (TE) process. This paper gives a new way to research the fault detection and diagnosis in industrial nonlinear system.