Gaussian Mixture Model-Based Online Anomaly Detection for Vectored Area Navigation Arrivals

Abstract

Identifying anomalous flight trajectories is crucial in airspace operations, as they can potentially lead to safety risks. One of the challenges in identifying abnormal aircraft trajectories in the vectored area navigation (RNAV) terminal airspace is distinguishing between anomalies and trajectories vectored from the structured procedures. Applying existing trajectory pattern identification algorithms to vectored trajectories could create patterns with wide variation within; hence, they cannot effectively discern anomalies from nominal vectored trajectories. In addition, most existing anomaly detection algorithms are developed to detect anomalies in historical air traffic surveillance data, and thus, they cannot be readily applicable for online implementation. To address these problems, an online anomaly detection algorithm for vectored flights in RNAV terminal airspace based on the Gaussian mixture model (GMM) is proposed, which can deal with highly complex airspace operations by incorporating the GMM with dynamic trajectory pattern classification and hybrid trajectory prediction. The proposed algorithm is demonstrated with real air traffic surveillance data at Incheon International Airport (ICN), South Korea.