Fault detection and estimation for nonlinear systems with linear output structure

Abstract

In this paper, a new formulation of fault detection and estimation algorithm has been presented for a class of known nonlinear dynamic systems with a linear output structure. Under certain assumptions on the nonlinear dynamics of the system and its model uncertainty, an adaptive observer-based approach is established so as to construct several effective residual signals that can be used to perform the required fault detection and estimation. A parameter dependent Lyapunov function is used to formulate a set of adaptive tuning rules for the time-varying parameters involved in both the adaptive observer and the fault estimation error. It has been shown that the algorithms can be applied to estimate both constants and slow-drifting faults with convergent residual signals. A simple simulation example is included to illustrate the use of the proposed methods and encouraging results have been obtained. Copyright © 2004 John Wiley & Sons, Ltd.