摘要
针对多无人机编队系统中如何实现高效率自主飞行、提高编队飞行控制精度的问题,提出了一种基于近端策略优化(PPO)与领导-跟随(leader-follower)策略结合的无人机编队飞行控制方法。通过定义状态空间、动作空间建立无人机的观测模型和控制模型;设计MLP-GRU-ACTLayer的策略网络架构,对观测数据进行特征提取及动作输出,并在训练过程中通过分布熵和动作评估方法对策略进行优化;设计奖惩机制引导无人机学习航向、高度和速度的控制策略,实现编队飞行的稳定性和适应性;同时将基于三变量(航向、高度、速度)的领导-跟随策略应用到编队飞行中,实现整体编队飞行控制。仿真结果表明,该方法能够高效实现无人机编队飞行控制,较传统方法将高度最大差值减少了7 m,速度最大差值减少了5 m/s。
To address the challenges of achieving high-efficiency autonomous flight and improving formation flight control accuracy in multi-UAV formation systems,this paper proposes a UAV formation flight control method that integrates proximal policy optimization(PPO)with a leader-follower strategy.First,the observation and control models of UAVs are established by defining state and action spaces.Subsequently,a policy network architecture,MLP-GRU-ACT Layer,is designed to extract features from observation data and output actions.During the training process,the policy is optimized through distribution entropy and action evaluation methods.A reward-punishment mechanism is designed to guide UAVs in learning control strategies for heading,altitude,and velocity,ensuring the stability and adaptability of formation flight.Additionally,a leader-follower strategy based on three variables(heading,altitude,and velocity)is applied to achieve overall formation flight control.Simulation results demonstrate that the proposed method can efficiently achieve UAV formation flight control.Compared to traditional methods,it reduces the maximum altitude deviation by 7 m and the maximum velocity deviation by 5 m/s.
作者
霍琳
高永霖
赵锐
李泽铎
HUO Lin;GAO Yonglin;ZHAO Rui;LI Zeduo(Shenyang Aerospace University,Shenyang 110136,China;Liaoning General Aviation Academy,Shenyang 110136,China)
出处
《兵器装备工程学报》
北大核心
2025年第S1期180-187,共8页
Journal of Ordnance Equipment Engineering
关键词
无人机编队
近端策略优化
领导-跟随策略
三变量
编队保持
UAV formation
proximal policy optimization
leader-follower strategy
three variables
formation maintenance
作者简介
霍琳(1984-),女,博士,副教授,硕导,E-mail:huolin@sau.edu.cn;通信作者:高永霖(1997-),男,硕士,E-mail:2838815717@qq.com。
