A Data-Driven Air Traffic Sequencing Model Based on Pairwise Preference Learning

The effective sequencing of arriving flights is the primary goal of air traffic management. Although various automated tools have been developed to support air traffic controllers, these tools do not accommodate the cognitive processes of the human controllers, which are necessary for application to actual operations. This paper proposes a new framework for predicting arrival sequences based on a preference learning approach that emulates the sequencing strategies of human controllers by learning from historical data. The proposed algorithm works in two stages: it first learns the pairwise preference functions between arrivals using a binomial logistic regression, and then it induces the total sequence for a new set of arrivals by comparing the score of each aircraft, which sums the pairwise preference probabilities. The proposed model is validated using historical traffic data at Incheon International Airport, and its performance is evaluated using Spearman's rank correlation and a dynamic simulation analysis.