面向变构型飞行器的强化学习位置姿态一体化控制方法  

Reinforcement Learning-based Integrated Position and Attitude Control Method towards Morphing Flight Vehicles

作　　者：路坤锋[1,2] 贾晨辉 黄旭 刘晓东 柳嘉润[1,2] 王昭磊 LU Kunfeng;JIA Chenhui;HUANG Xu;LIU Xiaodong;LIU Jiarun;WANG Zhaolei(Beijing Aerospace Automatic Control Institute,Beijing 100854,China;National Key Laboratory of Science and Technology on Aerospace Intelligent Control,Beijing 100854,China)

机构地区：[1]北京航天自动控制研究所,北京100854 [2]宇航智能控制技术全国重点实验室,北京100854

出　　处：《宇航学报》2024年第7期1100-1110,共11页Journal of Astronautics

基　　金：国家自然科学基金(U21B2028)。

摘　　要：针对变构型飞行器在飞行过程中由于构型发生改变导致其质心、气动力、转动惯量和气动力矩以及飞行器的抗扰能力等参数或特性发生变化,对飞行器飞行控制品质产生较大影响的问题,提出一种基于强化学习的变构型飞行器一体化位置姿态控制方法,通过孪生延迟深度确定性策略梯度(TD3)强化学习算法训练神经网络控制律,实现变构型飞行器的一体化位置姿态控制。算法通过数学仿真与飞行试验进行了验证,仿真结果与飞行试验结果表明,该算法所设计的神经网络控制律能够实现变构型飞行器的一体化位置姿态控制,并对于外界干扰具有较强的适应能力。Aiming at the problem that the change of parameters or characteristics of morphing flight vehicle,such as center of mass,aerodynamic forces,moments of inertia,aerodynamic moments,and disturbance rejection capability of the vehicle during flight,which may significantly affect the flight control quality of the vehicle,an integrated position and attitude control algorithm based on reinforcement learning is proposed.The twin-delayed deep deterministic policy gradient(TD3)reinforcement learning algorithm is utilized to train the neural network control laws to achieve integrated position and attitude control for morphing aircraft.The algorithm has been verified through mathematical simulation experiment and the flight test.The results of simulation experiments and flight tests show that the network control law designed with the proposed algorithm can achieve integrated position and attitude control for morphing aircraft,and has strong adaptabilities to external disturbances.

关 键 词：强化学习 变构型飞行器 位置控制 姿态控制 

分 类 号：V249.1[航空宇航科学与技术—飞行器设计] V448.2