Air Combat Maneuver Decision-Making Based on Improved Monte Carlo Tree Search Method

In order to improve decision-making accuracy and efficiency, achieve manned/unmanned coordination, and investigate novel tactics, autonomous air combat with mobile decision-making at its core is crucial. First, a situation assessment model is established in this research to evaluate the benefits of both the opponent and our side. Second, to bring the Monte Carlo tree search (MCTS) method closer to the demands of air combat, limiting criteria are added. The double-rollout link is then added subsequently to address the issue of the rollout link not being able to get the firing reward right away. To increase decision efficiency, a decision tree model is then incorporated into rollout. Finally, the confrontation simulation experiment is carried out, and the effectiveness of the improved algorithm is verified by analyzing the winning rate, situation change and decision frequency compared with the improved MCTS maneuver decision-making method.