基于强化学习的航天器姿态预设性能容错控制  

Fault-tolerant control of spacecraft attitude with prescribed performance based on reinforcement learning

作　　者：金磊[1] 杨绍龙 JIN Lei;YANG Shaolong(School of Astronautics,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学宇航学院,北京100191

出　　处：《北京航空航天大学学报》2024年第8期2404-2412,共9页Journal of Beijing University of Aeronautics and Astronautics

基　　金：中央高校基本科研业务费专项资金(YWF-22-L-801)。

摘　　要：针对惯量不确定性和执行机构故障的航天器姿态控制问题,提出了一种基于强化学习的预设性能容错控制方法。采用预设性能方法设计航天器的姿态控制器,以保证控制过程的暂态响应。为在线补偿惯量不确定,在预设性能控制器的基础上引入强化学习算法,使用评判网络近似代价函数,用于评估系统性能,同时使用动作网络产生前馈补偿控制,用于处理惯量不确定;设计自适应补偿控制,补偿执行机构故障和外扰动对航天器姿态的影响。基于Lyapunov稳定性理论证明整个闭环系统的稳定性。仿真结果表明:所提容错控制方法能够实现航天器执行机构故障情况下的稳定控制。A fault-tolerant control method with prescribed performance based on reinforcement learning was proposed for spacecraft attitude control with inertia uncertainties and actuator faults.In order to ensure the transient response of the control process,the attitude controller of the spacecraft was designed by using the prescribed performance method.A reinforcement learning algorithm was introduced based on the prescribed performance controller to compensate for the inertia uncertainty online.The critic network was used to approximate the cost function to evaluate the performance of the system,and the actor network was used to generate feedforward compensation control and deal with the inertia uncertainty.Then,an adaptive compensation control law was designed to compensate for the effect of actuator faults and external disturbance on spacecraft attitude.According to Lyapunov stability theory,the stability of the whole closed-loop system was proved.The simulation results show that the proposed fault-tolerant control method can realize the stability control of spacecraft with actuator faults.

关 键 词：强化学习 容错控制 预设性能 航天器 姿态控制 

分 类 号：V448.22[航空宇航科学与技术—飞行器设计] TP302.8[自动化与计算机技术—计算机系统结构]