基于DTW距离聚类的交叉口拥堵检测

Study of Detecting Intersection Congestion based on DTW Distance Clustering

交叉口是城市交通的关键节点,常发生交通拥堵,在分析交叉口采集的地点平均车速、流量和时间占有率时间序列特征的基础上,提出了交叉口入口交通状态动态时间弯曲(dynamic time warping,DTW)相似度量和K均值聚类的交叉口交通拥堵检测方法,并以某城市70个交叉口实测交通数据为例进行了拥堵检测,结果表明了该方法的可行性.

Aiming at intersection of urban transport where traffic congestion often occurs, the intersection trafficcongestion identification method of dynamic time warping （dynamic time warping , DTW） similar measureof the entrance intersection traffic state and K-means clustering is proposes based on the analyzing time seriescharacteristics of detecting average speed and traffic volume. With true data, 70 intersection traffic congestionsare detected by the method in a city. The results show the feasibility of the approach.