Multiple-fault diagnosis for spacecraft attitude control systems using RBFNN-based observers

In this paper, a novel multiple-fault diagnosis (MFD) scheme using radial basis function neural network (RBFNN)-based observers is presented for a spacecraft attitude control system (ACS) in the presence of external disturbances and nonlinear uncertainties. Based on dynamic and kinematic models, robust fault detection observers (FDOs) are designed to detect the simultaneous occurrence of actuator, gyro, and star sensor faults. Then, a series of RBFNN-based fault isolation observers (FIOs) are designed to decouple the faults of different components completely. This complete decoupling will guarantee that the diagnosis result of one component is not affected by the faults of other components; thus, multiple faults can be diagnosed simultaneously. To improve the accuracy of fault detection and reconstruction, disturbance compensation observers (DCOs) based on the RBFNN are also designed to compensate for the external disturbances. It is worth noting that the developed fault diagnosis scheme can be used to detect and isolate small faults. Finally, simulation results are presented to show the effectiveness and feasibility of the proposed method.