基于轨迹谱聚类的终端区盛行交通流识别方法

为了改善终端空域扇区和进离场航线对实际空中交通的流量及空间分布的适用性，研究了从大量航空器飞行轨迹中识别主要交通流的方法.在分析飞行轨迹空间特征的基础上，建立了基于3D网格的轨迹间相似性模型.利用谱聚类算法对终端区飞行轨迹样本进行聚类划分，提出了一种基于轨迹聚类核密度估计的盛行交通流和异常轨迹的识别方法，用于从空管雷达记录的飞行轨迹中识别出盛行交通流的实验.实验研究结果表明:该方法将1 476条轨迹划分为5个聚类，识别出5个盛行交通流，且识别结果未受到异常轨迹的影响. 

Identification of Prevalent Air Traffic Flow in Terminal Airspace Based on Trajectory Spectral Clustering

In order to improve the adaptability of terminal airspace and standard arrival/departure routes to real air traffic flows and their spatial distribution, a method for detection of main air traffic flows in massive flight trajectories was addressed. After analysis of the spatial characteristics of trajectories, a trajectory similarity model based on 3D grids was proposed. Flight trajectories were partitioned with spectral clustering algorithm, and an identification method for prevalent air traffic flow and outlier trajectories was proposed through kernel density estimation of trajectories in one cluster. Experiments were carried out from trajectories recorded by air traffic control radar to identify prevalent traffic flows. The results show that 1 476 trajectories were divided into 5 clusters, and 5 prevalent traffic flows were identified; in addition, the identification results were not affected by outliers.