微小卫星集群协同观测构形调整规划
Configuration adjustment planning for cooperative observation of microsatellite cluster
摘要点击 231  全文点击 334  投稿时间:2019-06-29  修订日期:2019-11-05
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.90500
  2019,36(12):2046-2053
中文关键词  微小卫星集群  构形调整  协调图  强化学习  Max-plus 算法
英文关键词  microsatellite cluster  configuration adjustment  coordination graph  reinforcement learning  max-plus algorithm
基金项目  国家重点研发计划(2016YFB0500803), 研究生创新基地(实验室)开放基金(kfjj20180321), 江苏省高校优势学科建设工程资助项目资助.
作者单位E-mail
汪志军 南京航空航天大学 jesse341@163.com 
赵艳彬 上海卫星工程研究所  
程月华 南京航空航天大学 chengyuehua@nuaa.edu.cn 
廖鹤 南京航空航天大学  
姜斌 南京航空航天大学  
中文摘要
      以微小卫星集群实现小行星探测为背景,研究局部信息交互的空间目标观测任务构形调整。针对难以直接求取集群构形调整的全局最优解问题,利用通信协调图,将全局协调决策分解成多个局部求解问题,并引入强化学习机制实现求解。针对集群全局协调决策问题,通过设计基于Max-plus算法的全局协调决策算法来实现全局协作;针对单星局部优化问题,设计基于神经网络的局部Q学习算法来实现单星动作调整规划。仿真结果表明,本文所提的协作规划算法能自主有效地将集群调整至期望构形,实现协同观测任务。
英文摘要
      This paper investigates the problem of configuration adjustment of spatial target observation tasks for local information interaction during the deep space exploration mission of the microsatellite cluster. In order to solve the problem of directly solving the global optimal difficulty of cluster configuration adjustment, this paper uses the communication coordination diagram to decompose the global coordination strategy into multiple local solution problems, and the reinforcement learning mechanism is introduced to realize solution. Aiming at the global coordination decision-making problem of cluster, the cooperative interaction in the local learning process is realized by designing the global coordinated decision algorithm based on Max-plus algorithm. Aiming at the local optimization problem of individual satellite, a local Q learning algorithm based on neural network is designed to realize the adjustment action planning of each satellite. The simulation results show that the proposed planning algorithm can independently and effectively adjust the cluster to the desired configuration to realize the observation task collaboratively.