Research on Game Penetration Strategy of High-Speed Vehicle in Midcourse Guidance Phase
摘要
Aiming at the problem that it is difficult to realize the penetration of interceptor by high-speed vehicle with traditional program maneuver, this paper proposes an intelligent game penetration strategy based on Proximal Policy Optimization-clip (PPO-clip) algorithm. Through the analysis of the influencing factors of interceptor miss distance, the neural network is used to generate the penetration strategy in the dynamic game environment online. Firstly, using the adjoint analysis method, the analytical expression of the miss distance is derived and the game goal that maximizes the miss distance of the interceptor is established. Secondly, the PPO-clip deep reinforcement learning algorithm is used to construct the observation space and action space of the vehicle intelligent game strategy. Based on the principle of minimum energy and the principle of maximum miss distance, the reward function of the penetration strategy is designed. Finally, combined with typical confrontation scenario training, the intelligent game penetration strategy set of high-speed vehicles is obtained. The simulation results show that when the maneuver ability of the penetration side is limited, the proposed game penetration strategy can effectively realize the midcourse game penetration of the interceptor.