高速飞行器集群通信拓扑自适应控制方法  

Communication Topology Adaptive Control Method for High-speed Space Vehicle Swarms

作　　者：白成超 王会霞[2,3] 郭继峰 路坤锋[2,3] BAI Chengchao;WANG Huixia;GUO Jifeng;LU Kunfeng(School of Astronautics,Harbin Institute of Technology,Harbin 150001,China;Beijing Aerospace Automatic Control Institute,Beijing 100854,China;National Key Laboratory of Science and Technology on Aerospace Intelligence Control,Beijing 100854,China)

机构地区：[1]哈尔滨工业大学航天学院,哈尔滨150001 [2]北京航天自动控制研究所,北京100854 [3]宇航智能控制技术国家级重点实验室,北京100854

出　　处：《宇航学报》2023年第7期1008-1019,共12页Journal of Astronautics

基　　金：国家自然科学基金(61973101);中国科协青年人才托举工程(2021QNRC001);黑龙江省自然科学基金优秀青年项目(YQ2022F012);思源联盟开放基金(HTKJ2022KL012003)。

摘　　要：针对基于传统通信机制的高速飞行器集群控制策略鲁棒性低、所需通信量大的问题,提出一种基于深度强化学习框架的可自主调节通信数量的集群控制方法。其中,使用深度神经网络构建控制策略与通信策略耦合的集群控制策略,其输出包含控制飞行器运动的过载指令以及与邻近飞行器的通信数量。通过与任务环境的不断交互,训练出的集群控制策略能根据环境信息自主调整通信拓扑结构,保证集群控制的鲁棒性和较低的通信量。仿真结果表明,相比于集中式,分层式和分布式通信机制,所提的自适应通信机制可在较低的集群通信量下安全快速地控制飞行器集群到达目标点并且较好地保持编队队形。Aiming at the problems of low robustness and large amount of communication required for the high-speed vehicle swarm control policy based on the traditional communication mechanisms,a swarm control method based on the deep reinforcement learning framework that can independently adjust the number of communications is proposed.A swarm control policy coupled with a control policy and a communication policy is constructed using a deep neural network,and its output includes overload commands to control the movement of the space vehicle and the number of communications with adjacent aircrafts.Through continuous interaction with the task environment,the trained swarm control policy can autonomously adjust the communication topology according to the environment,ensuring the robustness of the swarm control and the low communication traffic of the high-speed vehicle swarm.The simulation results show that,compared with the centralized,hierarchical and distributed communication mechanisms,the proposed adaptive communication mechanism can safely and quickly control the vehicle swarm to reach the target point and maintain the formation topology well under the lower swarm communication traffic.

关 键 词：高速飞行器 通信拓扑 集群控制 深度强化学习 自适应通信机制 

分 类 号：V19[航空宇航科学与技术—人机与环境工程]