Research on Traffic Delay Predicting at Signalized Intersection Based on Neural Network

Available traffic delays prediction models for signalized intersection tend to predict the traffic delays under certain conditions and they are weak in adapt to different situation. In the paper, based on the theories of BP neural network, a network model, having a strong ability to adapt to different conditions, for traffic delay in average hours at a signalized intersection is established. It is trained and tested utilizing the data of traffic delay in average hours at a certain entrance of a signalized intersection. The predicted results and the actual data are compared with each other and the results prove the reliability and effectiveness of BP neural network in predicting traffic delays.