The Research Of Urban Road Traffic Flow Short-term Prediction

In recent years, with the rapidly development of intelligent transportation system, traffic control and traffic flow guidance as an important component of intelligent transportation system, traffic management department has become the effective way to ease city traffic. Traffic control and traffic flow guidance has become the intelligent transportation system (ITS) popular research object. Accurate and rational, scientific forecast of city road traffic is traffic control and traffic flow guidance technology, to be completed on the road intersection and section of future time traffic flow dynamic forecast work, need to make full use of historical traffic data information and real-time traffic flow data information provides more reliable trip basis, therefore study of traffic flow, changes in the law, and the future development trend of time traffic volume or real-time, accurate, scientific and reasonable to predict, for traffic planning, traffic, traffic management, traffic control and safety, it is very important. In this thesis, the main research results include:First of all, on road traffic flow prediction research background and significance of the statement, further introduced the domestic and foreign existing traffic flow forecasting methods advantages and disadvantages. After combining the survey the actual traffic flow data, analysis of the traffic flow in the temporal and spatial distribution characteristics. According to the road traffic flow is nonlinear, time-variation and uncertainty characteristics, combining neural network with distributed processing, self-organizing, adaptive, self-learning characteristics of good, is proposed based on neural network traffic flow model of forecasting analysis.Again, with concrete examples, traffic flow forecasting model based on BP, RBF and GRNN neural network using MATLAB platform to forecast road traffic flow forecast. Through the validation of the prediction of actual sampling data, the comparative analysis of simulation results of the three models, come to the findings of the short-term prediction method for road traffic flow:Predicting the accuracy of analysis, using the GRNN neural network model predictions of accuracy performance better than the BP neural network and RBF neural network, three prediction models, RBF forecast accuracy to achieve the best results; forecast time analysis makes own shortcomings of the BP neural network prediction longest, followed by the RBF neural network model, predict the relatively shortest time GRNN neural network model. Comprehensive comparison drawn: RBF neural network is relatively BP neural network and GRNN with suitable for short-term forecasting of urban road traffic flow.Finally, the dissertation summarizes the main work, and then puts forward the traffic flow forecast further problems to be studied in.