A machine learned traffic flow coordination framework for flow-centric airspace

Abstract

Air traffic flow coordination at major flow inter- sections is a key enabler for flow-centric airspace concepts. This paper develops a flow-centric air traffic flow coordination framework to improve air traffic flow efficiency through flow identification, prediction, and re-routing at the Nominal Flow Intersections (NFIs). To identify the NFIs, a graph-based flow pattern consistency approach is proposed to model and analyze daily air traffic flow patterns. To predict future traffic demands at the identified NFIs, a transformer encoder-based neural network is adopted to learn the relations among the flow of flights at the NFIs. The acceptable flow limits at the NFIs are then determined by phase transitions of the flow efficiency versus the traffic demand. Finally, to avoid the predicted demand exceeding the identified flow limit and improve the flow efficiency, a reinforcement learning-based flow re-routing agent is trained to dynamically assign alternative routes to air traffic flows based on the evolving flow states. The agent’s performance is quantified by the flight time reduction in the flows without exceeding the flow limits. The re-routing model is trained and tested on a busy NFI that handles cross-border flows between Bordeaux and Madrid/Barcelona control centers, using ADS-B data for Dec 2019 in European airspace. Results show that, compared with the originally planned flows, the travel time of each flight is reduced by 322.168 seconds on average on a 2-hour basis.