Freeway Traffic Volume Forecasting Based On Neural Network

Freeway has the large scale, construction cost is high, the operation for a long time, assessment required after the completion of the construction. Assessment can improve the operation management level, improve traffic conditions, evaluate economic benefit effectively. Traffic flow forecasting is an important part of the post-assessment, it determines the evaluation effect.This paper studies on freeway traffic flow forecasting. Analysis of freeway traffic flow forecasting methods at home and abroad. Considering the history of freeway traffic volume can reflect the characteristic of its development trend, using a neural network nonlinear random training advantages of learning, and then put forward the freeway traffic flow forecasting method based on neural network. Method basically has the following two steps. The first step, analyses the influencing factors of freeway traffic volume, to use of analytic hierarchy process to extract the main influencing factors, to be prepared for the neural network prediction model building. The second step, extracted the main affecting factors as the input, the freeway traffic volume as the output, set up the three-layer BP neural network model for freeway traffic flow forecasting.Neural network prediction method and the traditional four-stage method respectively to freeway traffic flow forecasting. The experimental results show that neural network prediction method can save time and decrease the cost of prediction, and can improve the accuracy of prediction. This method is suitable for use in freeway traffic flow forecasting.