A sliding mode observer based sensor fault detection and isolation scheme for a nonlinear satellite system

This paper proposes a method to detect and isolate gyro faults for a benchmark Mars Express (Mex) satellite subject to uncertainties and noise. A Least squares approach for an over-determined system is used to detect the presence of a fault. Only if a fault is detected is isolation carried forward. A model for each of the fault cases is developed using the remaining healthy sensors which is a unique advantage of the sensor structure. A generalized bank of nonlinear observers based on sliding modes is designed to isolate the faulty sensor. The equivalent injection signal is processed from each of the observers to generate signals, which are an indicator of the effect of the faulty sensor. The proposed design is verified on a nonlinear Mex satellite for a variety of fault scenarios. Finally the performance of the designed FDI scheme is verified using Monte Carlo simulations.