Adaptive attitude tracking control for spacecraft based on input normalized neural network

To deal with the attitude tracking control problem of spacecraft using a variable speed control moment gyro (VSCMG) cluster as the actuator in the presence of inertia and actuator uncertainties, an adaptive attitude tracking controller is developed based on input normalized neural networks. A nonlinear feedback controller is designed as a baseline attitude controller, and an adaptive input normalized neural network, used to estimate and eliminate the unknown uncertainties is added as a compensator, thus a large number of estimates of parameters needed in some standard adaptive control methods are avoided. The complicated stability analysis of the closed-loop system based on common neural networks is simplified by using the input data normalization. Lyapunov stability theory is considered to guarantee the stability of the closed-loop dynamics and convergence of attitude tracking errors. Simulation results show the proposed control method can satisfy the high-accuracy attitude tracking control requirements for spacecraft in the presence of inertia and actuator uncertainties as well as external disturbances.