›› 2012, Vol. 14 ›› Issue (6): 775-780.doi: 10.3724/SP.J.1047.2012.00775

• ARTICLES • Previous Articles     Next Articles

Urban Traffic Congestion Detection Based on Clustering Analysis of Real-time Traffic Data

LU Xiaoya1, SONG Zhihao2, XU Zhu2, LI Muzi2, LI Ting2, SUN Weiya2   

  1. 1. Faculty of Computer Science and Technology, Southwest University for Nationalities, Chengdu 610041, China;
    2. Faculty of Geosciences and Environment Engineering, Southwest Jiaotong University, Chengdu 610031, China
  • Received:2012-11-26 Revised:2012-12-03 Online:2012-12-25 Published:2012-12-25

Abstract:

Traffic congestion in urban road network heavily restricts transportation efficiency. Detecting traffic congestions in the spatio-temporal sense and identifying network bottlenecks become an important task in transportation management. Up to now, many traffic congestion detection methods have been proposed, which have focused on the detection of momentary local congestions. Larger-scale, longer-time and regular congestions can't be detected using these methods. That is because congestions have different temporo-spatial scales, and a characteristic is not considered in those methods. This paper proposes a new kind of urban traffic congestion detection method that deals with spatio-temporal extension of congestion. It is based on spatio-temporal clustering analysis of real-time traffic data. By defining a proper spatio-temporal correlation, the classic DBSCAN algorithm is adapted to tackle spatio-temporal clustering. With it we can detect longer time and regular traffic congestion in the spatio-temporal sense. Experiments have been conducted using real traffic condition data of Chengdu to validate the effectiveness of the method. The experiment shows that the proposed method can detect the congestion areas and identify the spatio-temporal extent of congestions accurately. The detected congestion areas were compared with congestion report from local traffic management authority and found to be consistent with the later.

Key words: spatio-temporal correlation, spatio-temporal clustering analysis, DBSCAN algorithm, traffic congestion detection