Skip to main content

Predicting Airport ATFM Regulations using Deep Convolutional Neural Networks

Paper ID

SIDs-2022-082

Conference

SESAR Innovation Days

Year

2022

Theme

Machine Learning I

Project Name

CRONOS, SESAR 2020 IR Wave 2 project PJ04-W2 TAM

Keywords:

Airport, ATFM Delays, ATFM Regulations, Convolutional neural networks, Time series

Authors

Olivier Lattrez, Rocío Barragán Montes and Mateusz Michalski

DOI

Project Number

874472

Abstract

Airport ATFM regulations are a source of inefficiencies in European aviation. Providing a timely and accurate regulation prediction to the NMOC will contribute to better situational awareness. It will allow to anticipate collateral issues and help to coordinate preventive measures that can avoid ATFM regulations from being implemented. In this paper, a Deep Convolutional Neural Network that was trained to predict the probability of an ATFM regulation at the airport level is presented. The model, which showed promising results, has been put in operation in trial mode since the summer of 2022 and has been providing valuable insights during the pre-tactical and tactical phases on a daily basis.