摘要：AbstractShort-term traffic flow forecasting for junction of isomerism road network which consists of freeway network and urban road network, is the key technology of traffic management and intelligent control. Real-time and reliability of short-term traffic flow forecasting is directly related to traffic management and collaborative control. There are various methods that have been established to forecast traffic flow, but most of the forecasting models are constructed according to analysis of the historical and current traffic flow series in selected sections or crossings, without considering the dynamic information of related traffic network. According to the traffic flow characters of junction of isomerism road network system, the paper analyzes the relationship between the traffic flow of a certain section with other section's flow, and calculates the dynamic correlation coefficient. Following, the paper selects input variables, and establishes the Radial Basic Function (RBF) neural network model for prediction on the basis of dynamic correlation coefficient. Finally, the paper takes the G2 freeway and Beijing road network that consists of the junction of isomerism road network as an example to forecast the short-term traffic flow. The result of forecasting is very accurate, and the relative errors are within the 15%. It indicates that the model could be used to forecast the short-term traffic flow for junction of isomerism road network system.
关键词：Freeway network;Urban Road network;Junction of Isomerism Road Network;Dynamic Correlation coefficient;Radial Basic Function Neural Network
