A novel guidance law based on proximal policy optimization

In this paper, a new guidance law based on deep reinforcement learning is proposed for the high-speed maneuvering target attack problem. Firstly, the missile-target kinematic model is established, and the action space and state space of reinforcement learning are designed. Then, according to the missile strike process, the reward function suitable for this scenario is proposed. The proximal policy optimization (PPO) based guidance law construction is completed. Finally, the strike effect in multiple sets of experiments verifies the effectiveness of the method proposed in this paper.