Exploring Supervised Learning to Predict Air Traffic Delay in ATM Network Resiliency

Abstract

Tactical Air Traffic Flow Management (ATFM) highly relies on situational awareness at European ATM Network (EATMN). The study is establishing the link between ATFM and system resiliency. A basic model for EATMN with the purpose of understanding network resiliency is described. Also the data flow and key role of Large-scale capacity regulations (also known as ATFCM regulations) as a feedback loop is demonstrated. Then the complexity of EATMN and available data is reviewed to declare the advantages of Machine Learning (ML) to predict Network delay. additionally the literature on ML approaches in delay prediction is addressed to explain the contribution of our methodology. Furthermore the study explores different ML classes in a two step approach. First the applicability of ML for delay prediction is checked and secondly, in search of a baseline for prediction quality, different supervised learning methods are applied.