Distributed Vibration Control of Large Flexible Satellite Solar Panel Via Reinforcement Learning

This paper proposes a distributed vibration control strategy based on reinforcement learning to solve the vibration control problem of large flexible satellite solar panel. The main contents are as follows: Firstly, a multiple substructures model for satellite solar panel is established, and a interaction topology is constructed to describe the information flow among substructures. Secondly, a distributed vibration controller based on reinforcement learning is designed, and the optimal control is obtained by using the strategy iteration algorithm. It is proved theoretically that the iterative results converge to the solution of the HJB equation. The strategy algorithm is implemented by means of neural network, and the gradient of the output and value function of the system is sampled to update the network, which can achieve better convergence performance. In this process, the weight of the network reaches the ideal value and the approximate optimal distributed vibration control protocol is obtained. Finally, simulation results are given to prove that the distributed control algorithm in this paper can effectively suppress the vibration of flexible solar panel.