A robust intelligent fault detection scheme for magnetorquer type actuators of satellites

In this paper, the problem of robust fault detection for general nonlinear systems subject to state and sensor uncertainties and disturbances is considered. A nonlinear observer-based strategy is proposed where a recurrent nonlinear-inparameters neural network (NLPNN) is employed to identify the general unknown fault. The neural network weights are updated based on a modified dynamic backpropagation scheme. The proposed fault detection scheme does not rely on the availability of all state measurements. The ultimate boundedness of the state estimation error, neural network weights errors, and neural network gradients in the presence of an unknown fault as well as plant and sensor uncertainties is shown using Lyapunov´s direct method. The performance of the proposed fault detection strategy is evaluated via simulations performed on a satellite attitude control systems consisting of magnetorquer type actuators.