Over-the-Horizon Air Combat Environment Modeling and Deep Reinforcement Learning Application

Abstract

As we all know, over-the-horizon air combat has become one of the important fight forms that determine the trend of modern warfare. The biggest challenge in the confrontation process is how to make aircrafts cooperative-decision to lock, launch and avoid operations. To this end, this paper investigates the deep reinforcement learning application on the over-the-horizon air combat environment to enhance the ability of multi-aircraft cooperative decision-making and intelligent optimization. First, a novel over-the-horizon air combat environment is constructed as a training environment for deep reinforcement learning, which could provide an easy-to-calculate simulation environment with higher precision. Then, we propose the proximal policy optimization combined with the long short-term memory network to deal with incomplete information and realize intelligent decision optimization at the same time. Finally, the effectiveness of the proposed algorithm is verified by simulation experiments.