Abstract: To ensure the flight safety in the terminal area and reduce the workload of the controller, factors relating to aircraft trajectory were considered such as the position, speed and altitude, a definition and identification process of the aircraft abnormal trajectory in the terminal area were proposed, a speed correction coefficient based model was built for trajectory similarity,by which trajectories can be automatically classified by the spectral clustering method, the characteristics of each type of trajectory data were calculated, two central trajectories were selected to characterize the position features of each type of trajectory, and the similarity distance and flight distance between the trajectories were extracted as anomaly feature factors for identifying aircraft anomaly trajectories. The results show that the trajectory similarity model based on velocity correction coefficient can be used to realize the automatic classification of aircraft trajectory, and the classification result is more reasonable and accurate, and that both improving the abnormal feature factor and setting two central trajectories are helpful in improving the abnormal detection effect.

Key words: aviation safety, abnormal trajectory detection, trajectory similarity, spectral clustering, terminal area