Application of DBSCAN to Anomaly Detection in Airport Terminals

Early identification of abnormal flight conditions is extremely critical in terms of safety precautions and preventing possible incidents. Because there is a massive amount of raw data streaming daily via the ADS-B to Flightradar24, which often approaches hundreds of millions, dealing with such unorganized details and identifying abnormal cases is extremely difficult. In this paper, the DBSCAN algorithm proposes to analyze and identify unknown events within routine flight data without previous knowledge of historical flight data. DBSCAN does not need a complex configuration or consumes a significant amount of time to implement, but rather relies on the appropriate selection of features and also on the particular threshold (ε), The algorithm analyzed three dataset types for Najaf International Airport consisting of 41 B738 flights (DXB), 39 B738 flights (IST), and 29 A310 flights (IKA). The percentage of anomalies in these airlines was as follows: Fly Dubai 1.32 percent, Istanbul Airlines 0.32 percent, while Iranian Airlines recorded the highest anomalies at 1.66 percent. The results of these analyzes must be made available to airspace experts and pilots to assess the value of safety risks.