DISURBANCE DECOUPING SEQUENTIAL MONTE CARLO FILTERING AND APPLICATION IN NONLINEAR ROBUST FAULT DIAGNOSIS

Abstract

Fault detection and isolation for nonlinear stochastic systems is an important and difficult problem, since the noises especially the measurement noises are unavoidable in practice. In this paper, a new robust sequential Monte Carlo filtering approach for nonlinear systems with unknown disturbances is proposed from the recursive Bayesian estimation theory. Based on this new filtering technique, a robust fault detection and isolation strategy for nonlinear stochastic systems is presented, which is also illustrated by simulations on a DTS200 three-tank model.