中国科技核心期刊

中文核心期刊

CSCD来源期刊

空间控制技术与应用 ›› 2022, Vol. 48 ›› Issue (3): 39-48.doi: 10.3969/j.issn.1674 1579.2022.03.005

• 论文与报告 • 上一篇    下一篇

GEO在轨服务任务建模与强化学习服务序列规划

  

  1. 国防科技大学航天科学与工程学院
  • 出版日期:2022-06-27 发布日期:2022-07-15
  • 基金资助:
    国家自然科学基金资助项目(12102460)

On orbit service mission modeling and reinforcement learning service sequence planning in GEO


  • Online:2022-06-27 Published:2022-07-15
  • Supported by:

摘要: 面向地球同步轨道卫星故障修复和功能维护的在轨服务系统是我国正在建设发展的重要航天系统工程之一.针对地球同步轨道多目标服务任务规划问题,提出了一种在轨服务任务建模与强化学习服务序列规划方法.推导了航天器轨道动力学模型和霍曼兰伯特四脉冲交会模型,针对几种典型的地球同步轨道在轨服务任务建立了任务模型,基于强化学习设计了目标卫星服务序列规划方法,并通过数值仿真验证了任务规划方法的有效性.仿真结果表明,该方法能够真实全面地反映服务卫星在目标交会和任务执行过程中的轨道参数改变以及速度增量和时间消耗,规划得到的最优服务序列更具工程实用性.

关键词: 地球同步轨道, 在轨服务, 任务建模, 服务序列规划, 强化学习

Abstract: The on orbit service system for fault repair and function maintenance of geosynchronous satellites is one of the important aerospace projects that are being developed in China. Aiming at the multi target service mission planning problem in geosynchronous earth orbit, an on orbit service mission modeling and reinforcement learning service sequence planning method is proposed in this paper. The spacecraft orbit dynamic model and the Hohmann Lambert four pulses rendezvous model are derived. The mission models are established for several typical on orbit service missions in geosynchronous earth orbit. A target satellite service sequence planning method is developed based on reinforcement learning. Numerical simulations are carried out to verify the effectiveness of the mission planning method. The results illustrate that the proposed method can comprehensively reflect the change of orbit parameters and the consumption of velocity increment and time of the service satellite in both the target rendezvous process and the mission execution process. The optimal service sequence obtained by planning is more applicable in the engineering.

Key words: geosynchronous earth orbit, on orbit service, mission modeling, service sequence planning, reinforcement learning

中图分类号: 

  •