A ROBUST ITERATIVE LEARNING OBSERVER BASED FAULT DIAGNOSIS OF TIME DELAY NONLINEAR SYSTEMS

Abstract

An Iterative Learning Observer (ILO) updated successively and iteratively by immediate past system output error and ILO input is proposed for a class of time-delay nonlinear systems for the purpose of robust fault diagnosis. The proposed observer can estimate the system state as well as disturbances and actuator faults so that ILO can still track the post-fault system. In addition, the observer can attenuate slow varying output measurement disturbances. The ILO fault detection approach is then applied to automotive engine fault detection and estimation. Simulations show that the proposed ILO fault detection and estimation strategy is successful.