Multi-aircraft adaptive cooperative route planning based on deep reinforcement learning
Aiming at the problem of limited penetration and combat capability of a single aircraft,a multi-aircraft adaptive cooperative route planning method is proposed.By introducing the multi-agent deep reinforcement learning algorithm,the multi-aircraft route planning decision-making framework is constructed,and the online route planning instructions of each aircraft are solved.On this basis,an optimized DL-MADDPG algorithm is proposed to guide the aircraft for disturbance learning and enhance the adaptability of aircraft in complex environments.At the same time,the cooperative reward and individual reward are respectively set in the reward function,which can effectively ensure the cooperation of multi-aircraft system strategy and the effectiveness of individual strategy of each aircraft.Simulation results show that the proposed multi-aircraft cooperative route planning method based on deep reinforcement learning has good adaptability and robustness,and can help multi-aircraft to realize online decision-making of cooperative route planning in complex multi-mission scenarios.
NETL
NSTL
万方数据
