Application of Data Mining in Mesoscopic Traffic Simulator Modeling

In order to solve the limitation that the classical speed-density model describes the dynamic change characteristics of the traffic flow, more road detected information is utilized in the process of the parameters calibration of the model in the mesoscopic traffic simulator. Firstly, the detector data were preprocessed, and then, the data mining, including locally weighted regression, K-Means clustering and k-nearest neighborhood and agglomerative hierarchical cluster, was used to calibrate vehicle speed, vehicle density as well as densities and flows. The test with field data shows that the proposed algorithms have great performance in the parameters estimation for DTA based simulation.