Anomaly detection in onboard-recorded flight data using cluster analysis

Abstract

▓ Developed a method to detect anomalous flights using cluster analysis • Advantages ▓ Minimum prior knowledge of data required S Identify multiple nominal data patterns • Limitation ▓ Current transformation method is only applicable to flight phases start or end with a specific event ▓ Applied the proposed method on a dataset of B777 flights for takeoff phase and approach phase ▓ Initial evaluation indicates that cluster analysis is a promising approach for anomaly detection in FDR/QAR data