The invention discloses a spacecraft attitude preset time cooperative control method based on reinforcement learning. Mathematical description of a multi-spacecraft attitude cooperative control problem is carried out; a preset time distributed observer is constructed, and it is ensured that a follower obtains observation information of the state of a navigator within preset time; determining a preset time performance function to quantitatively describe convergence time, transient performance and steady-state performance constraints of the collaborative tracking error; converting a cooperative tracking error system constrained by preset performance into an unconstrained system based on error transformation of a barrier function; determining a performance index function and a corresponding Hamiltonian-Jacobian-Bellman equation, and solving a partial derivative about optimal control to obtain a representation form of optimal control input about an optimal function; and designing an approximate optimal controller under an evaluation network framework by adopting a reinforcement learning method. According to the method, the spacecraft formation system can be ensured to meet the preset convergence time and transient and steady-state performance, and meanwhile, the energy consumption problem is considered.
本发明公开了一种基于强化学习的航天器姿态预设定时间协同控制方法,进行多航天器姿态协同控制问题的数学描述;构造预设定时间分布式观测器,保证跟随者在预设定时间内获得领航者状态的观测信息;确定预设定时间性能函数定量刻画协同跟踪误差的收敛时间、瞬态性能及稳态性能约束;基于障碍函数的误差变换将受预设性能约束的协同跟踪误差系统转换为无约束系统:确定性能指标函数及其相应的哈密顿‑雅克比‑贝尔曼方程,求解关于最优控制的偏导得到最优控制输入关于最优函数的表现形式;采用强化学习方法在评价网络框架下设计近似最优控制器。本发明可以保证航天器编队系统在满足预设定收敛时间、瞬态和稳态性能的同时,兼顾能量消耗问题。
Spacecraft attitude preset time cooperative control method based on reinforcement learning
一种基于强化学习的航天器姿态预设定时间协同控制方法
12.12.2023
Patent
Elektronische Ressource
Chinesisch
IPC: | B64G Raumfahrt , COSMONAUTICS |
REINFORCEMENT LEARNING FOR SPACECRAFT ATTITUDE CONTROL
TIBKAT | 2020
|Europäisches Patentamt | 2024
|