基于DDPG-LQR的高超声速飞行器时间协同再入制导  

Time Cooperative Reentry Guidance for Hypersonic Vehicle based on DDPG-LQR

作　　者：宋志飞 吉月辉[1,2] 宋雨 刘俊杰[1,2] 高强 SONG Zhifei;JI Yuehui;SONG Yu;LIU Junjie;GAO Qiang(School of Electrical Engineering and Automation,Tianjin University of Technology,Tianjin,300384;Tianjin Key Laboratory of Control Theory and Application for Complex Systems,Tianjin,300384)

机构地区：[1]天津理工大学电气工程与自动化学院,天津300384 [2]天津市复杂系统控制理论与应用重点实验室,天津300384

出　　处：《导弹与航天运载技术(中英文)》2025年第1期57-64,共8页Missiles and Space Vehicles

基　　金：国家自然科学基金(No.62203331);天津理工大学研究生教育教学研究与改革项目(ZDXM2202,YBXM2204)。

摘　　要：针对多高超声速飞行器协同作战的特点,提出一种基于深度策略性梯度和线性二次型调节器(Deep Deterministic Policy Gradient-Linear Quadratic Regulator,DDPG-LQR)的时间协同再入制导方案。首先,采用序列凸优化方法生成满足多个约束的时间协同再入轨迹及其相应的稳态控制量,并且采用Radau伪谱法离散运动学方程,以提高轨迹优化离散精度。其次,采用线性二次型调节器(Linear Quadratic Regulator,LQR)跟踪时间协同再入轨迹。为了提高协同制导精度和制导效果,采用深度策略性梯度(Deep Deterministic Policy Gradient,DDPG)在线优化LQR的权重矩阵系数。在DDPG算法中,通过引入合适的奖励函数来提高算法的优化性能。仿真结果表明,在初始状态误差和不确定性的情况下,通过与传统的LQR控制器相比,本文所提出的协同制导方案具有更好的协同制导精度和制导效果。Aiming at the characteristics of multiple hypersonic vehicles cooperative combat,a time cooperative reentry guidance scheme based on deep deterministic policy gradient and linear quadratic regulator(DDPG-LQR)is proposed.Firstly,the sequential convex programming method is used to generate the time cooperative reentry trajectory satisfying multiple constraints and its corresponding steady-state control quantity.The Radau pseudospectral method is used to discretize the motion equations to improve the discretization accuracy of trajectory optimization.Secondly,the linear quadratic regulator(LQR)is used to track the time cooperative reentry trajectory.In order to improve the cooperative guidance accuracy and guidance effect,the deep deterministic policy gradient(DDPG)is used to optimize the weight matrix coefficients of the LQR online.In the DDPG algorithm,the optimization performance of the algorithm is improved by introducing an appropriate reward function.The simulation results show that the cooperative guidance scheme proposed has better cooperative guidance effect and guidance accuracy than the traditional LQR controller in the case of initial state error and uncertainty.

关 键 词：多高超声速飞行器 协同制导 序列凸优化 深度策略性梯度 线性二次型调节器 

分 类 号：V412[航空宇航科学与技术—航空宇航推进理论与工程]