基于集群作战的事件触发强化学习分布式跟踪控制  

Event-triggered Reinforcement Learning for Distributed Tracking Control in Swarm Operations

作　　者：李华青 李骏 郑李逢 王慧维 石亚伟 LI Huaqing;LI Jun;ZHENG Lifeng;WANG Huiwei;SHI Yawei(Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing,College of Electronic and Information Engineering,Southwest University,Chongqing 400715,China)

机构地区：[1]西南大学电子信息工程学院非线性电路与智能信息处理重庆市重点实验室,重庆400715

出　　处：《指挥与控制学报》2025年第1期79-86,共8页Journal of Command and Control

基　　金：国家自然科学基金(62173278)资助。

摘　　要：针对有人/无人集群系统协同飞行场景,提出一种基于强化学习的分布式动态事件触发跟踪控制方法。该方法以未来信息化空战为背景,将路基/舰载战斗机、预警机等高成本有人机视为领航者,作为“忠诚僚机”的高速低成本无人机视为跟随者,实现有人/无人集群系统的分布式实时跟踪控制。为提高集群隐身性能,设计了一种动态事件触发强化学习算法,无人作战单元仅依赖局部信息可自适应地调整通信触发阈值,有效地减少有人/无人集群通信传输频率。数值仿真验证了该方法的可行性。Aiming at the cooperative flight scenarios of manned-unmanned swarm systems,a distributed event-triggered tracking control approach based on reinforcement learning is proposed.To achieve distributed tracking control of manned-unmanned swarm systems in real time,the approach is based on the background of future information-based air warfare,in which high-cost manned aircraft,such as load-based/ship-based fighters and early warning aircraft,are considered as navigators and high-speed low-cost unmanned aircraft as "loyal wingmen" are considered as followers.The design of a dynamic event-triggered reinforcement learning algorithm allows the unmanned combat unit to adaptively adjust the trigger threshold by only relying on local information,thereby reducing the frequency of manned-unmanned swarm communication transmission and enhancing swarm stealthy performance.Finally,the feasibility of the approach is verified by the numerical simulation.

关 键 词：集群作战 强化学习 动态事件触发 分布式跟踪控制 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计]