Urban traffic flow prediction based on road network model

This paper addresses an issue of short-term traffic flow prediction in urban traffic networks with traffic signals in intersections. An effective spatial prediction approach is proposed based on a macroscopic urban traffic network model. In contrast with other time series based or spatio-temporal correlation methods, this research focuses on the substantial mechanism of vehicles transmission on road segments and the spatial model of the entire urban network. Furthermore, this approach employs a simple speed-density model based on the macroscopic fundamental diagram (MFD) to obtain a more accurate vehicle travel time on the link. Finally, the microscopic traffic simulation software, CORSIM, is adopted to simulate the real urban traffic, and the proposed method is used to predict the traffic flows generated by CORSIM. The simulation results illustrate that our approach performs effective prediction timely in the rush hours, as well as the suddenly changed traffic states.