Research And Program Of Incident Detection Algorithm Based On Probe Vehicle Data

With the rapid increase of the population of motor vehicles, urban traffic is facing more and more challenge in congestion mitigation. In this context, the nonrecurring congestions induced by traffic incidents will have significant impacts on urban traffic efficiency. An efficient algorithm for automatic incident detection can identify the traffic incidents accurately in time and then decrease the effect caused by incidents. This thesis proposed a new algorithm of automatic incident detection (AID) based on probe vehicle data in Beijing. After a comprehensive review of domestic and foreign research literatures, this thesis summarized AID algorithms and their evaluation indexes according to their development progress. The advantages and disadvantages of these algorithms were comparisond for the purpose of providing reference for this research. Then, on the basis of understanding the theory of CUSUM , and the characteristic of probe vehicle data and traffic flow in Beijing, the AID algorithm model was developed. Finally, this thesis used simulative and practical data to test and testify this algorithm and then completed a program to detect traffic incidents. In conclusion, the CUSUM AID algorithm was found of a high efficient algorithm based on probe vehicle data.