面向在轨加注的组合体姿态SAC智能控制

In order to ensure the attitude stability, the soft Actor-Critic (SAC) intelligent control method for the attitude control of on-orbit refueling combination is established. First, the equivalent model of liquid fuel sloshing-transfer coupling motion is proposed, and the time-varying inertia attitude dynamics environment is constructed for training the intelligent controllers. Then, the modified SAC algorithm is proposed. In the training process, random actions are used to enhance the optimization efficiency. In practice, deterministic actions are applied to enhance the control stability. On this basis, the SAC attitude intelligent controller is designed to achieve the attitude convergence. At the same time, the SAC refueling pressure intelligent controller is designed to reduce the influence of the disturbance on the attitude by automatic adjusting the refueling pressure. The simulation results show that, compared with the twin delayed deep deterministic policy gradient ( TD3 ) algorithm, the proposed SAC attitude intelligent controller is more robust to the random disturbance caused by time-varying inertia. The SAC refueling pressure intelligent control can suppress the refueling disturbance while ensuring the refueling efficiency.