Adaptive Extended Kalman Filter for Actuator Fault Diagnosis

Abstract

This paper presents an algorithm for actuator fault diagnosis of nonlinear systems. The method is derived under classical uniform complete observability, controllability, and persistent excitation condition. To this end, the fault is modeled as a constant or a piecewise constant parameter vector. The diagnosis algorithm is based on the Extended Kalman Filter (EKF) with an explicit update law for the actuator fault estimation. From a practical point of view, the proposed algorithm can be used for general nonlinearity. To illustrate the effectiveness of the diagnosis algorithm, we present two numerical examples using the models of an autonomous car and a gantry crane.