基于HDBSCAN算法的终端区航迹聚类  

Aircraft Trajectory Clustering in Terminal Area Based on HDBSCAN Algorithm

作　　者：王志伟 刘继新[1] 杨宋瑞雪 田文[1] WANG Zhi-wei;LIU Ji-xin;YANG Song-rui-xue;TIAN Wen(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing Jiangsu 211106,China)

机构地区：[1]南京航空航天大学民航学院,江苏南京211106

出　　处：《计算机仿真》2024年第1期44-47,348,共5页Computer Simulation

基　　金：国家重点研发计划项目资助(2021YFB1600500);国家自然科学基金项目(52002178);国家自然科学基金项目(71971112)。

摘　　要：针对机场终端区运行环境复杂,空中交通流密度分布不均的问题,提出了一种基于快速动态时间规整(FastDTW)和层次密度聚类(HDBSCAN)算法识别具有密度差异的交通流,以挖掘终端空域的交通流模式。结合航迹的多维特征,对原始航迹数据进行重采样;利用Fast DTW方法构建航迹相似度矩阵,应用HDBSCAN方法对输入矩阵进行聚类。运用终端区真实航迹数据进行仿真验证,仿真结果表明,上述方法可以实现航迹精细化聚类,有效识别不同密度的空中交通流。Aiming at the problem of complex operating environment and uneven distribution of air traffic flow density in the airport terminal area,a Fast Dynamic Time Warping(FastDTW)combined with Hierarchical Density-Based Spatial Clustering of Applications with Noise(HDBSCAN)algorithm is proposed to identify traffic flows with density differences in order to explore the traffic flow patterns in the terminal airspace.Combining the multidimensional characteristics of the trajectory,the original trajectory data are resampled;the trajectory similarity matrix is constructed using the FastDTW method,and the HDBSCAN method is applied to cluster the input matrix.The simulation results show that the above method can achieve fine clustering of trajectories and effectively identify air traffic flows of different densities by using real trajectory data in the terminal area for simulation verification.

关 键 词：航空运输 航迹聚类 动态时间规整 层次密度聚类 

分 类 号：V355[航空宇航科学与技术—人机与环境工程]