A nonlinear observer-based approach to fault detection, isolation and estimation for satellite formation flight application

Abstract

In this paper, the problem of fault diagnosis for a satellite formation flight is considered. A scheme is investigated for fault detection, isolation, and estimation of a class of nonlinear systems. In this scheme, we consider the model uncertainty, input and environmental disturbances. For fault detection, a nonlinear observer is designed to minimize the uncertainty within ℋ∞ framework. The Linear Matrix Inequality (LMI) formulation is used to obtain the observer gain matrices. In the next step, a bank of nonlinear robust unknown input observers is designed to isolate the faulty actuator. Fault isolation is achieved based on the generalized observer strategy. The proposed observer can simultaneously estimate faults and states; it can also decouple the unknown input disturbances and attenuate the effect of model uncertainty and external disturbances. In order to fulfill these goals, a Lipschitz formulation and Linear Parameter Varying (LPV) method are used in all of the proposed observers, which lead to less conservative LMI condition. Then, the mentioned approach is applied to fault diagnosis of a formation of satellites. Moreover, a distributed fault detection, isolation, and estimation scheme based on above observers is proposed. In this scheme, each satellite not only can diagnose its own faults but also it is able to diagnose its neighbors faults.