Automatic Detection of Flight Maneuvers with the Use of Density-based Clustering Algorithm

Ever-changing situation in the aviation demands a change in flight training programs that would reflect present needs and threats in contrary to the traditional training that didn't changed much for decades. Therefore, new alternative training concepts have been developed that cover these needs. However, these concepts do not apply to initial training which seem to be a crucial phase of a pilot training. Thus, the aim was to create a software solution that would identify individual flight maneuvers and evaluate them so that the overall evaluation would be done by considering objective evaluation and flight instructors' subjective expertise. A study was done with strictly given flight schedules. For the purpose of automatic maneuver detection, density-based spatial clustering of applications with noise - DBSCAN clustering algorithm was used, which could determine maneuvers and thus exclude the noise from clusters of maneuvers. The results indicate that the proposed solution was able to identify the prescribed maneuvers with high sensitivity. The solution could be extended in the future to identify all flight maneuvers considering as many parameters from flight data recorder as possible and thus carry out complete objectively based pilot performance evaluation.