Research On Abnormal Trajectory Identification Of Aircraft In Terminal Area Based On Multidimensional Feature

In the 21 st century,the rapid development of China's air transport industry has brought about the increase of flight traffic,which not only increases the workload of controllers,but also threatens to endanger flight safety.Therefore,it is necessary to study the massive trajectory data of aircraft in the terminal area.By mining and analyzing the massive trajectory data in the terminal area,the hidden trajectory information can be found,which provides a basis for the development of an efficient controller-assisted decision-making system.In this article,the clustering analysis in data mining technology is used to study the automatic classification of aircraft terminal area based on multi-dimensional features.Based on the trajectory classification,the identification and detection of aircraft anomaly trajectory are realized based on two methods: position and motion.The methods are mutually verified.At the same time,the research on aircraft anomalies mostly focuses on the abnormal position of the trajectory,and proposes anomaly identification detection of the vertical direction of the aircraft--climbing and descending rate.In this article,the range of the terminal area is determined firstly.The preprocessing of the ADS-B trajectory data is completed by Lambert projection,linear interpolation and velocity correction.The definition of the anomaly trajectory of the aircraft in the terminal area is given in combination with the definition of the anomaly trajectory.For the Euclidean distance trajectory similarity model,the Euclidean distance similarity model is improved,and the multi-dimensional feature trajectory similarity model based on positional features(longtitude,latitude,altitude),heading and velocity is constructed,and spectral clustering is used to complete the pair.Research on automatic classification of aircraft trajectories.Based on the automatic classification of the trajectory,the concept of the trajectory angle is proposed.The trajectory is determined by the trajectory angle.The feature trajectory is fitted by the multi-linear linear fitting method combined with the aircraft trajectory position feature(3D)to obtain the feature trajectory position expression.Whether the feature track point to the position expression distance is greater than the range of setting the 95% confidence interval completes the identification detection of the abnormal track.Considering the majority of abnormal trajectory recognition research,the motion attribute features of trajectories are not fully considered.Combined with the definition of abnormal trajectory,the relationship between data is analyzed,and the aircraft abnormal trajectory recognition detection method based on multi-dimensional motion features such as curvature,heading and altitude is proposed.The identification and detection of abnormal trajectories are completed using principal component analysis and random forest model.At present,the research on trajectory anomalies mostly focuses on the anomaly of the trajectory position and lacks the consideration of the anomaly in the vertical direction of the trajectory.This paper proposes an anomaly identification method for the climb rate of the aircraft,and defines the abnormality of the climb rate by the moment estimation.The analysis of the factors affecting the rate of climb and decline,using random forest classifiers to achieve anomaly detection of the rate of climb and decline.