Online Target Assignment Method for Aircraft Cluster Based on Prior Information Guidance

This paper proposes a new model for online target allocation of aircraft clusters based on prior information guidance for deep reinforcement learning DQN algorithm, aiming to address the issues of low training efficiency, unstable training, and poor solution credibility encountered by existing deep reinforcement learning algorithms when solving target allocation problems for aircraft clusters. By setting the flight maneuverability constraints as the prior information for solving the target allocation problem of the aircraft cluster, the model takes into account the maneuverability of the aircraft while compressing the model's action space and conducting guided training. Finally, the effectiveness of the designed work in this paper is verified, and the simulation results show that by introducing the prior information of flight maneuverability constraints, on the one hand, it can effectively accelerate model training and improve solution effectiveness, and on the other hand, it can ensure that the allocated results are all maneuverable and reachable, improving the solution credibility of the target allocation problem.