Modeling and Simulation of Freeway Short-Term Traffic Flow Prediction

Freeway short-term traffic flow prediction can lay a solid foundation for analyzing the road traffic operation status and making real-time traffic surveillance strategies. Firstly, because of the complex modeling process and off-line prediction, differential method together with zero equalization in the data pretreatment and recursive least square method with forgetting factors in the parameter identification are used in establishing improved time-series prediction model. Secondly, as the difficulty in determining the initial value, the statistical method is used to generate initial value in modeling of Kalman filtering prediction model. Thirdly, because of the low speed of convergence in the BP neural network and the local minimum problem existing in the BP algorithm, RBF neural network is used in establishing the traffic volume prediction model. The results indicate that predictability of various models is confirmed, and the characteristic of individual models is also obtained.