Sensor fault estimation in finite-frequency domain for nonlinear time-delayed systems by T–S fuzzy model approach with local nonlinear models

Abstract

This paper investigates the sensor fault estimation problem in finite-frequency domain for nonlinear time-delayed systems via T–S fuzzy approach with local nonlinear models. First, to estimate the fault, an augmented system is constructed, where the auxiliary state vector consists of the states of the system and the auxiliary filter. Second, an unknown input observer with finite-frequency specifications is designed to achieve fault estimation. Then, by utilising the Parseval's theorem, the sufficient conditions of the presented observer are established. Different from some existing methods, the state estimation error dynamics are decoupled from the local nonlinear dynamics via the designed observer such that the observer synthesis is simplified and the conservatism can be reduced. Meanwhile, compared with some conventional methods in the entire-frequency domain, a less restrictive analysis condition could be derived by the developed observer scheme. Finally, simulation results are given to illustrate the merit of the proposed approach.