A method of air traffic control anti-collision based on k-time control deep reinforcement learning

Abstract

The invention discloses an air traffic control anti-collision method based on K-time control depth reinforcement learning, which includes the following steps: firstly, setting the number of aircraft in a sector in a use scene, and setting the control times K in the anti-collision process; then, In the training mode, carry out K times of control, in the first K-1 times of control, through the function of the neural network, according to the method of two-dimensional normal distribution, determine the next position point, and according to the method of reinforcement learning, the parameters of the neural network To update, in the Kth control, the destination is used as the next location point, and this cycle completes the training of the neural network; finally, in the application mode, using the trained neural network, the shortest path without conflicts can be obtained . The method of the invention can be applied in the existing air traffic management system, and can obtain the shortest path to the destination without conflicting with other aircraft in the sector, which has practical significance for air traffic control path planning.

Description

technical field [0001] The invention relates to the field of air traffic management, in particular to an air traffic control anti-conflict method based on K-time control depth reinforcement learning. Background technique [0002] In recent years, civil aviation has developed rapidly, and the continuous development has brought serious air traffic congestion, which has brought great pressure to air traffic controllers. When an aircraft flies from one sector to another, it needs to plan its trajectory and give correct guidance to avoid conflicts with existing aircraft in the sector. Existing algorithms can generate optimal or sub-optimal flight paths and guide aircraft, but the calculation efficiency is low and cannot meet the real-time requirements of real-world air traffic control, and further research is still needed. Deep reinforcement learning has high execution efficiency and flexible use. After improvement, it can be applied in the air traffic control system to quickly ...

AI Technical Summary

Problems solved by technology

Existing algorithms can already generate optimal or suboptimal flight paths and guide aircraft, but the calculation efficiency is low and cannot meet the real-time requirements in real-world air traffic control, and further research is still needed

Images