The scope of Air Traffic Flow Management (ATFM) problem is to guarantee a safe and e cient air traffic flow by computing 4D trajectories, one for each flight, that satisfy all the airports and en-route sectors capacity constraints. One of the current research efforts in the ATFM domain is to investigate formulations that consider Airspace Users’ preferences, as recommended by the SESAR (Single European Sky ATM Research) programme. This is a quite challenging task because Airspace Users’ preferences, which depend on many factors, e.g., costs, duration, geometry etc., are not always fully known. We consider a path-based integer programming formulation that assigns 4D trajectories to flights, and we develop a data-driven heuristic that i) learns Airspace Users’ preferences and reduces the set of variables accordingly, and ii) solves the mathematical model. Using flight trajectories queried from Eurocontrol DDR2 data repositories, the learning phase determines relevant trajectories via clustering, and applies data analytics tools, mainly tree classifiers, support vector machines and multiple regression, to explore the relation between trajectories and potential preference determinants. As a result, a set of trajectories and information on related Airspace Users’ preferences is computed for each flight. These sets feed the optimization model for the assignment phase.
A heuristic for the Air Traffic Flow Management Problem based on Data Analytics and Integer programming
Luigi De Giovanni
;
2019
Abstract
The scope of Air Traffic Flow Management (ATFM) problem is to guarantee a safe and e cient air traffic flow by computing 4D trajectories, one for each flight, that satisfy all the airports and en-route sectors capacity constraints. One of the current research efforts in the ATFM domain is to investigate formulations that consider Airspace Users’ preferences, as recommended by the SESAR (Single European Sky ATM Research) programme. This is a quite challenging task because Airspace Users’ preferences, which depend on many factors, e.g., costs, duration, geometry etc., are not always fully known. We consider a path-based integer programming formulation that assigns 4D trajectories to flights, and we develop a data-driven heuristic that i) learns Airspace Users’ preferences and reduces the set of variables accordingly, and ii) solves the mathematical model. Using flight trajectories queried from Eurocontrol DDR2 data repositories, the learning phase determines relevant trajectories via clustering, and applies data analytics tools, mainly tree classifiers, support vector machines and multiple regression, to explore the relation between trajectories and potential preference determinants. As a result, a set of trajectories and information on related Airspace Users’ preferences is computed for each flight. These sets feed the optimization model for the assignment phase.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
