基于强化学习的在轨目标逼近

On-Orbit Target Approach Based on Reinforcement Learning

从强化学习的角度,对在轨目标逼近问题进行研究,设计了一种整合制导与控制的端到端的算法。首先对在轨目标逼近问题进行数学建模;然后对强化学习算法原理进行简介,根据问题特点分析不同强化学习框架的优劣,确定以DDPG作为算法框架,并设计了基于强化学习的在轨目标逼近算法;最后通过仿真验证,分析了基于强化学习逼近算法的优劣性。

From the perspective of reinforcement learning,the approaching problem of on-orbit targets is studied,and an end-to-end algorithm that integrates guidance with control is designed.Firstly,the on-orbit target approach is modelled.Then,the principle of reinforcement learning algorithm is introduced,the advantages and disadvantages of different reinforcement learning frameworks are analyzed according to the characteristics of the problem,and DDPG is used as the algorithm framework and an on-orbit target approach algorithm based on reinforcement learning is designed.Finally,the advantages and disadvantages of the algorithm based on reinforcement learning are analyzed through simulation.