Traffic incident impact analysis with random matrix theory and cluster analysis

Management of traffic incident is an essential part of emergency management. Impact analysis of incident becomes more and more important in the operation of Intelligent Traffic System. This paper uses flow data collected from loop detectors to analyze the impact of incident in urban traffic system. A complex network model using correlation coefficient as the edge weight is constructed. Then, the random matrix theory is used to evaluate the impact level of incident on network-wide traffic status. At last, cluster analysis is also applied to obtain the largest cluster in the network which indicates the scale and location of critical congestion caused by the incident.