A Terminal Airspace Flight Path Planning Algorithm Based on Improved Deep Q-Network

Due to the deteriorated weather and the activities of other airspace users, the control job at terminal airspace become complex and changeable.This situation brings more work burden for the air traffic controllers to conduct aircraft path planning.The conventional aircraft path planning methods are not consistent with the actual operation situation of air traffic control facility.In this paper, a terminal airspace flight path planning algorithm based on improved Deep Q-Network (DQN) is proposed.It use the results generated by classical path planning algorithms as priority experience to feedback the DQN model.The experimental results show the proposed algorithm is able to reduced the distance of the path planning and training time compared with the traditional path planning algorithm.It greatly improves the model training efficiency.