利用GPS轨迹二次聚类方法进行道路拥堵精细化识别

针对当前在精细识别道路拥堵时空范围方面研究的不足，提出一种利用GPS轨迹的二次聚类方法，通过快速识别大批量在时间、空间上差异较小且速度相近的轨迹段，反映出道路交通状态及时空变化趋势，并根据速度阈值确定拥堵状态及精细时空范围。首先将轨迹按采样间隔划分成若干条子轨迹，针对子轨迹段提出相似队列的概念，并设计了基于密度的空间聚类的相似队列提取方法，通过初次聚类合并相似子轨迹段，再利用改进的欧氏空间相似度度量函数计算相似队列间的时空距离，最后以相似队列为基本单元，基于模糊C均值聚类的方法进行二次聚类，根据聚类的结果进行交通流状态的识别和划分。以广州市主干路真实出租车GPS轨迹数据为例，对该方法进行验证。实验结果表明，该二次聚类方法能够较为精细地反映城市道路的拥堵时空范围，便于管理者精准疏散城市道路拥堵，相比直接聚类方法可以有效提升大批量轨迹数据的计算效率。

Identification of Urban Network Congested Segments Using GPS Trajectories Double-Clustering Method

A double-clustering method is proposed for GPS trajectories in urban networks that efficiently identify trajectories with similar spatio-temporal and speed properties. The Spatio-temporal ranges of congested urban road segments are identified based on the trajectory velocity threshold. This paper defines similar queues and designs a DBSCAN based clustering method. By merging similar sub-trajectories into similar queues, the proposed method reduces computation costs. An improved trajectory similarity measure is used to calculate the spatio-temporal distance between similar queues. A C-means clustering method is applied. An experiment is implemented by clustering the floating car trajectories in Guangzhou, China; the results show that the proposed method can not only reflect the change of traffic congestion in space and time, but also improves computational efficiency when dealing with large-scale trajectory data.