Research On Traffic Flow Prediction Based On Grey Prediction Model

Short-term traffic flow forecasting is the premise and key of intelligent traffic control and management, identification of the state of traffic flow and real-time traffic flow induced. But so far, its results are unsatisfactory. Previous models of prediction of traffic flow have some disadvantages that the time of operation is too long, a large amount of historical data is required and the precision is low. So the study of short-term traffic flow forecasting has a certain practical significance. Based on comparative analyses of the existing short-term traffic flow forecasting models and research of characteristics of traffic flow, this thesis establishes goals: the models established can remedy the shortcomings of the existing short-term traffic flow forecasting models.In recent years, the grey prediction model is favored by the traffic flow prediction researchers owing to simpler algorithm, less required data and computing time. Based on analyses and summaries of the existing grey prediction model, the thesis presents two grey prediction models of traffic flow on different traffic flow original data. And they are certified effective through the experiments. Major jobs are as follows:1. For only a small number of measured sections of the historical traffic data, the thesis proposes a short-term traffic flow prediction model which improves the background value of GM(1,1) based on in-depth analyses of the factors influencing the accuracy of conventional GM(1,1) model. This model can be applied not only to low-growth sequence, but also can be applied to high-growth sequence. Comparing with other models, improved model is effective. And through simulation experiment of Matlab7.0, the model improved by metabolism predicting is proved more applicable to short-term traffic flow prediction and has a higher accuracy.2. For the data that include not only a little data of the measured section but also the traffic data of the upper and lower sections, based on in-depth study of the grey forecasting model, this thesis establishes MGM (1, n), a short-term traffic flow prediction model, which considers the balance between upstream and downstream traffic flow. And the validity and practicality of the model are verified through an experiment. On the one hand, it extends the range of the MGM (1,n) model, on the other hand, it is proved that traffic flow model considering the upstream and downstream traffic flow is more applicable to short-term traffic flow forecasting and has a higher accuracy than the model only considering historical data.