An Integrated Fault Detection and Diagnosis Using Kaman Filter and Eigen Structure Assignment - Application to Three Tank System

Abstract

An integrated fault detection and diagnosis approach against partial actuator failure for discrete-time stochastic systems has been proposed. When the dynamics of the process under control are changed due to the component (actuator) faults, the performance of the system degrades. The aim of proposed scheme is to detect and diagnoses the fault present and also to predict the effectiveness of the actuator. The scheme is based on a kalman filter for the parameter estimation via the Eigen structure assignment. This method also guarantees the stability of the system by analyzing the cost effectiveness factor of the actuator. The proposed approach it is demonstrated in a three tank benchmark system and the result illustrates its robustness.