The invention discloses a satellite group collaborative task planning method based on mixed expert experience playback, and relates to the technical field of reinforcement learning, satellite scheduling and multi-agent, and the method comprises the following steps: S1, building an optimization model for constellation emergency task planning; s2, establishing a Markov decision model oriented to constellation task planning, and designing a depth Q-Learning task planning algorithm based on mixed expert experience playback; and S3, performing online decision making, and performing real-time task planning by using the trained network. According to the satellite group cooperative task planning method based on mixed expert experience playback, the influence of user requirements, unique observation and satellite observation resources is considered, the task completion rate, the task delay time and the satellite resource balance degree are taken as indexes, a decision-making model oriented to ground emergency observation is established, and the satellite group cooperative task planning method based on mixed expert experience playback is established by utilizing expert experience data. And designing a depth Q-Learning constellation task planning algorithm based on expert experience, obtaining an optimal online task planning strategy, and planning an observation window of a satellite to a target.
本发明公开了一种基于混合专家经验回放的星群协同任务规划方法,涉及强化学习、卫星调度和多智能体技术领域,包括以下步骤:S1:建立面向星座紧急任务规划的优化模型;S2:建立面向星座任务规划的马尔科夫决策模型,设计基于混合专家经验回放的深度Q‑Learning任务规划算法;S3:在线决策,利用训练好的网络进行实时任务规划。本发明采用上述的一种基于混合专家经验回放的星群协同任务规划方法,考虑用户需求、唯一性观测和星上观测资源的影响,以任务完成率、任务延迟时间及星上资源均衡度为指标,建立了面向对地紧急观测的决策模型,利用专家经验数据,设计基于专家经验的深度Q‑Learning星座任务规划算法,获得最优在线任务规划策略,规划卫星对目标的观测窗口。
Star group collaborative task planning method based on mixed expert experience playback
一种基于混合专家经验回放的星群协同任务规划方法
2023-11-17
Patent
Electronic Resource
Chinese
Data-based collaborative awareness event playback method and device
European Patent Office | 2023
|Task-based programming and sequence planning for human-robot collaborative assembly
BASE | 2019
|