Satellite Attitude Tracking Decision Method based on Deep Deterministic Policy Gradient for Moving Target Observation

Space-based observation for moving targets has the advantages of global area, which can effectively compensate for the ground-based and air-based observation methods limited by land and airspace. However, the unknown and variability of target motion makes it difficult for satellites to make decisions and adjust their attitude in real-time to track the target. Aiming at this problem, an attitude tracking decision method based on deep deterministic policy gradient is proposed in this paper. Considering the control torque saturation and space pointing constraint, a satellite attitude tracking decision model is established and the optimization objective is designed. Then, the deep deterministic policy gradient method is used to solve this optimization problem, and the action value network and policy network is trained by transfer learning to accelerate the training process of whole model. Finally, simulation results show that the proposed method is capable of tracking the moving target autonomously and effectively.