Taxiing Speed Intelligent Management of Aircraft Based on DQN for A-SMGCS

Abstract

In view of ever-increasing air traffic, much attention has been paid to air traffic management research to improve efficiency. Normally, once two taxiing aircrafts may conflict in the hotspot area, one of them must wait in the holding point until the conflict is resolved. However, this method reduces the efficiency though it ensures sufficient safe distance between aircrafts. In this paper, we discussed the feasibility of the dynamic control of the taxiing speed to improve the efficiency under the A-SMGCS(Advanced Surface Movement Guidance and Control Systems). The conflict samples were trained based on actual operation data and AirTOp simulation operation data. Action spaces, state spaces and reward function were designed and DQN(Deep Q Network) model was used to make speed regulation decisions. The experimental results show that the model can adopt different strategies in different environments, which greatly improved the intelligent decision-making level of aircraft taxiing speed adjustment and the operating efficiency of the scene. The case study showed that the time of conflict solving and the main taxiway occupation length can be reduced by 13.16% and 47.9% separately when the conflict is released.