Intelligent Maneuver Decision Method of UAV based on Reinforcement Learning and Neural Network

The efficiency of UAV maneuver decision-making based on the traditional maneuver control quantity is the key to restrict the improvement of UAV autonomous air combat game ability in complex battlefield environment. By refining pilots' combat training and air combat thinking experience, the intelligent maneuver decision-making method for UAV autonomous air combat can effectively improve the UAV autonomous air combat efficiency. Aiming at the problem of maneuver decision-making in continuous state space, this paper designs an autonomous maneuver decision-making model based on actor critical reinforcement learning theory, adopts NRBF neural network as the value function approximator, the action controller outputs continuous control variables, and introduces Gaussian random action variables to balance the problem of "exploration utilization" in strategy learning A state space adaptive adjustment method based on relative entropy distance is proposed to simplify the network structure and enhance the learning ability of the network. The simulation results show that the proposed method has the ability of air combat confrontation, the output control quantity is smooth, and the strategy learning efficiency is higher.