Scholarlink

In this paper, we propose an approximate optimal control strategy for drag-free satellites using reinforcement learning, where state constraints related to the position of test masses are taken into consideration. To handle six-degrees-of-freedom position constraints, a barrier function is embedded in the performance cost function. Then, one critic neural network is utilized to learn cost function online and get the optimal control law. Meanwhile, in order to relax excitation condition, historical data are introduced into the weight adaptive law. The Lyapunov method is utilized to ensure that the weight error and state error of the drag-free satellites are uniformly ultimately bounded. Numerical simulations confirm the efficacy of the proposed method.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Learning-Based Drag-Free and Attitude Control of Spacecraft with State Constraints

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Haoran Li,
  • Xiaodong Shao,
  • Qinglei Hu,
  • Yonghe Zhang,
  • Pengcheng Wang

摘要

In this paper, we propose an approximate optimal control strategy for drag-free satellites using reinforcement learning, where state constraints related to the position of test masses are taken into consideration. To handle six-degrees-of-freedom position constraints, a barrier function is embedded in the performance cost function. Then, one critic neural network is utilized to learn cost function online and get the optimal control law. Meanwhile, in order to relax excitation condition, historical data are introduced into the weight adaptive law. The Lyapunov method is utilized to ensure that the weight error and state error of the drag-free satellites are uniformly ultimately bounded. Numerical simulations confirm the efficacy of the proposed method.