Prediction Method for a Congestion State in an Urban Road Network Bottleneck

A traffic state prediction of congestion areas in traffic networks is an important basis of traffic control and guidance. By knowing the congestion state of traffic networks, typical types of bottlenecks can be determined through statistical analysis of the inflow-outflow rate according to traffic flow directions at each intersection in a congested area. Based on the auto regressive analysis method, a traffic demand prediction model of traffic congestion bottlenecks was developing by using historical and real-time traffic volume as the reference. Furthermore, comparing forecasting traffic volume with outflow ability of each congestion node in a traffic network, a self-correcting discriminate model of real-time state and occurrence time for traffic congestion is proposed. A comparative analysis of predicted results and actual traffic networks was conducted to confirm the validity of the discriminate model, which showed that accumulated volume during traffic congestion can be used to predict the real-time operation state of traffic networks.