A novel decision-making algorithm for beyond visual range air combat based on deep reinforcement learning

In this paper, a novel decision-making algorithm based on deep reinforcement learning(DRL) is proposed for the decision-making problem in beyond visual range(BVR) air combat. Firstly, the relative kinematics model of 1 vs 1 air combat is established, and the state space and action space of the fighter are designed. Then, a new reward function is designed according to the situation of the BVR air combat, which is suitable for a wider range of BVR confrontation scenarios, and the construction of a decision-making method based on DRL is completed. Finally, several sets of experimental data are given to verify the effectiveness of the algorithm.