On-line fuzzy neural control of satellite attitude based on Q-learning

A fuzzy neural control approach applied to the three-axis stabilized satellite is presented. In order to solve the problems of online learning and tuning of the fuzzy neural network parameters, the method of Q-learning combined with BP neural network is proposed and studied so that the training samples for the self-learning controller are not needed. Simulation results showed that the proposed control method with Q reinforcement learning architecture could not only improve the accuracy, stability and robustness of the system, but also deal with uncertainties and external disturbance efficiently.