Node Importance Assessment of Traffic Complex Network Based on C-Means Clustering

Abstract

Hub nodes of urban traffic complex network are very important for region traffic signal control. Traditionally, region traffic signal control system like SCOOT/SCATS use traffic flow, vehicle queue and distance between junctions as reference in sub-control-area selection practice, in which process engineers’ experience should play importance roles. In this paper, node degree, node betweeness and high peak hour traffic flow are selected as indexes for traffic network node importance assessment. Moreover, C-Means clustering is applied to analysis which junction could be act as a hub node for regional traffic control. To test the effectiveness of this method, urban network around Chang’an Street in Beijing including almost fifty nodes, China is used as trail field. Data result shows Changchunjie, FuyoujieNankou and Hepingmen junction have high clustering characteristics when clustering number are 3, 4 and 5. And the clustering center shows very similar prosperities with real hub node in practice. In conclusion, the multi-index and clustering analysis could provide theoretical support for urban traffic complex network hub node assessment.