Reconfigurable fault-tolerant attitude tracking for spacecraft with unknown nonlinear dynamics using neural network estimators with learning-type weight updating

This study investigated the problem of robust and reconfigurable attitude-tracking control with fault-tolerant capability for spacecraft under nonlinear inertia uncertainties, disturbance torques, and actuator faults. To improve the accuracy of reconstructing actuator faults, we proposed a nonlinear learning neural network estimator that combines the radial basis function neural network (RBFNN) model with an iterative learning algorithm, enabling the arbitrary precision of actuator fault reconstruction. A P-type iterative learning algorithm successively updates the RBFNN’s weight with a low computational load. Moreover, to ensure fast and robust spacecraft attitude fault-tolerant tracking, the learning RBFNN was integrated into a sliding mode control (SMC) scheme, leading to a learning neural-network SMC (LNNSMC), designed using the separation principle. The learning RBFNN was utilized to approximate and compensate for unknown nonlinear attitude dynamics online. Finally, the superiority of the presented method was demonstrated through a numerical example.