Nonlinear Adaptive Observer Based Component Fault Diagnosis of Nonlinear Systems in Closed-Loops

摘要

Designing Adaptive observers for MIMO nonlinear time-varying deterministic systems is an open problem. In this paper we give a novel solution to this problem by use of a modified strong tracking filter(STF). First, the modified STF is presented,then some technical points of view to use the STF as a nonlinear adaptive observer are discussed. Later on, by combining with a modified Bayes classification algorithm, the proposed nonlinear adaptive observer is adopted to constitute a fault detectionand diagnosis algorithm, and is successfully used for the component fault diagnosis of a continuous stirred tank reactor, a nonlinear deterministic system controlled in closedloops. Computer simulation results demonstrate the effectiveness of the proposed schemes