•  
  •  
 

Corresponding Author ORCID Identifer

Madan Singh https://orcid.org/0000-0003-3879-3213

Section

Regular Issue

Abstract

Modern air traffic management faces growing challenges, particularly at busy airports that must handle high flight volumes in limited airspace corridors. The study is especially pertinent for such airports, where daily operations must navigate intricate air traffic networks. To overcome these challenges, innovative approaches are needed to predict and manage congestion effectively which remains a major challenge. The paper presents a Spatio-temporal analytical framework for air traffic flow forecasting, utilizing a dual deep-learning architecture that combines network topology evaluation with time-series analysis. The model contains two major components. The first component employs Graph Neural Networks (GNNs) to model spatial relationships between interconnected flight routes, which are represented as a graph. The second uses Recurrent Neural Networks (RNNs) to analyse temporal patterns in flight delays and traffic density variations. By integrating these approaches, the framework uniquely accounts for both geographical distribution and time-dependent fluctuations in air traffic congestion. Tests with real-world flight data confirm that the model outperforms traditional methods, delivering higher accuracy in identifying high-congestion routes and predicting peak demand periods, thereby enabling more efficient traffic flow management. These improvements enhance operational efficiency at key air traffic hubs, potentially reducing delays and optimizing resource allocation across airspace sectors.

Download

Share

COinS