Increasing the Detection Performance of Genuine Separation Minima Infringements with XGBOOST

The Automatic Safety Monitoring Tool (ASMT) is a key component of the Safety Management System of an air navigation service provider. Its main function is to monitor and record separation minima infringements and runway incursions. The recording of these events helps monitoring key safety performance indicators. The ASMT provides candidate events which are suitable to be considered as genuine infringements. These candidate events are often non-genuine due to different circumstances, such as noise in the vertical profile of the aircraft trajectories. The process of filtering genuine events from the initial population of candidate is an important step for accurately reporting on the safety performance indicators. This paper presents the workflow followed for deploying XGBoost for discriminating between genuine and non-genuine events within PERSEO, which works as the Automatic Safety Monitoring Tool for ENAIRE. Results show a significant increase on the detection performance of the separation minima infringements, which can facilitate larger analysis due to an increment of the detection precision while maintaining its sensitivity.