UIO-based Fault Estimation for a Class of Time-varying Systems with Event-triggered Transmissions

Abstract

This paper addresses the fault estimation problem for a class of discrete time-varying systems by resorting to the unknown input observer (UIO) technique. The system is assumed to be subject to unknown inputs and stochastic additive disturbances. The sensor can decide whether to send the current measurement to the filter according to the difference between the current measurement and the last transmitted one. A set of filters is obtained such that the effects of unknown inputs on the estimation results can be integrated into uncertainties with known bounds and each filter corresponds one possible fault. Upper bounds of the state and fault estimation error covariances are calculated in the simultaneous presence of the unknown inputs and event-triggered measurements, and then the filter gains are determined recursively to minimize the bounds. The residual matching method is employed to isolate the faults and the outputs of the filters can be used to realize fault estimation. The effectiveness of the proposed method is demonstrated by a simulation example.