基于深度Q网络的飞行器增益调参技术研究  

Research on Gain Scheduling Based on Deep Q Network for Aircraft

作　　者：白文艳 张家铭 黄万伟[1] 张远 Bai Wenyan;Zhang Jiaming;Huang Wanwei;Zhang Yuan(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]宇航智能控制技术国家级重点实验室,北京100854

出　　处：《航天控制》2022年第5期47-52,共6页Aerospace Control

摘　　要：针对飞行器传统增益调参法依赖于人工经验繁琐费时、难以实现参数自整定的缺点,提出了利用强化学习中的深度Q网络算法与飞行环境状态的交互不断学习,实现对控制增益的自动调整动作。训练结果表明,该方法使高速飞行器能够自适应调整控制增益,稳定跟踪攻角指令,节省了人工调参步骤及时间,有效提高了控制系统自适应性。Aiming at the shortcomings of the traditional gain scheduling interpolation method which relies on the experience of engineers and is time-consuming and laborious,and is difficult to meet the real-time control effect,the interaction is porposed to be used between the deep Q network algorithm and the flight environment state to realize the automatic control gain adjustment action.The training results show that the control gain can be adjusted adaptively in the hyper-sonic vehicle,the angle of attack command can be tracked effectively and strong robustness behaves by using this method.

关 键 词：高速飞行器 姿态控制 强化学习 增益调参 

分 类 号：V448.2[航空宇航科学与技术—飞行器设计]