Possibilities of quantification of factors influencing the aircraft ground handling process and TOBT prediction

Abstract

The subject of this paper is to summarize current research in the area of aircraft departure delay prediction based on machine learning algorithms and to confirm the relevancy of the identified variables (factors) whose implementation into predictive models could improve their accuracy and thus the ability to accurately predict the Target Off Block Time (TOBT) at Collaborative Decision Making (CDM) airport. In order to predict delays, several prediction models have been developed. One of the large categories of mathematical models are machine learning methods. The article includes a comprehensive literature review focused on machine learning algorithms confirming that none of those approaches used data from aircraft ground handling to predict aircraft departure delays, mainly due to ground handling data availability and scope of the research. The paper describes variables that could extend the existing machine learning prediction models. This research is supported with the real operational data from Václav Havel Airport Prague. The case study at Prague airport verifies a correlation of proposed variables with TOBT time. In several cases, a strong correlation between the proposed variables and TOBT was confirmed.