A Data-Driven Framework for Modelling Complexity in Terminal Manoeuvring Area
Résumé
This paper presents an objective, data-driven framework for quantifying air traffic complexity in the Terminal
Manoeuvring Area (TMA) using historical ADS-B data from Singapore TMA. The motivation for developing this framework
stems from the limitations of traditional subjective measures, which are often influenced by individual perceptions and can vary
significantly between air traffic controllers. Subjective measures may also fail to capture real-time operational demands, especially
in complex, high-density environments such as Singapore TMA. By focusing on operational outcomes—specifically vectoring and
holding patterns—the framework provides a more accurate reflection of real-time complexity. Principal Component Analysis
(PCA) and k-means clustering are employed to classify complexity levels based on trajectory features such as arc lengths,
curvatures, and holding durations. The results show that total arc lengths and curvatures are significant complexity factors,
with extensive vectoring contributing more to TMA complexity than holding patterns. The significance of this work lies in
its data-driven and objective approach to measuring air traffic complexity, offering a more accurate reflection of real-time demands compared to traditional subjective methods. Quantitative evaluations across multiple real-world scenarios validate the
framework’s effectiveness, showing that TMA complexity is more strongly associated with vectoring intensity and holding patterns
than with flight density alone. This current framework can be extended to incorporate vertical profiles of arrival and departure
flights and develop predictive models with practical, actionable lookahead times for real-time air traffic management.
Domaines
Intelligence artificielle [cs.AI]| Origine | Fichiers produits par l'(les) auteur(s) |
|---|