Real time and accurate predictions of intersections help set scientific traffic signal programs, expand road capacity and improve traffic conditions. The paper establishes a short-term forecasting model of intersection channel imports according to the Levenberg-Marquardt (LM) neural network Algorithm, which is based on the analysis of intersection traffic volume time and spatial correlation, combining LM neural network distributed processing, self-organizing, adaptive, self-learning, and other good characteristics. Using MATLAB to forecast short-term traffic volume of intersection imports with the prediction model and some specific examples, the empirical results show that the prediction model has preferable prediction accuracy.
展开▼
