Research on the Application of Reinforcement Learning Algorithm in Decision Support of Beyond-Visual-Range Air Combat

Abstract

In order to solve problems of the action selection space and the difficulty of convergence of traditional proximal policy optimization algorithm in air combat simulation, proximal policy hierarchical optimization algorithm is proposed. The framework of intelligent decision model of air combat based on reinforcement learning is constructed, and the antagonistic experiment is carried out and visualized. The experimental result shows that proximal policy hierarchical optimization algorithm could drive the agent to produce indirect attack and other tactical behaviors in the process of confrontation. The purpose of improving the performance of the traditional algorithm and decision-making efficiency of air combat is achieved.

Keywords

• |beyond-visual-range air combat|intelligent decision|artificial intelligence|reinforcement learning|proximal policy optimization algorithm|hierarchical reinforcement learning