Research On Short-term Traffic Flow Prediction Based On GRU-BP Combined Model

As motor vehicles become more and more popular,China’s car ownership is also a straight line,China’s highway mileage growth rate is far from catching up with the growth rate of car ownership.Traffic congestion has become a difficult problem to solve,but also brought environmental pollution,noise pollution and other problems.With the emergence of various intelligent traffic auxiliary devices and the explosive growth of traffic data,real-time traffic condition prediction becomes particularly important.Traffic flow prediction in intelligent transportation system is the basis of reasonable traffic guidance and control.It can provide a solid foundation and data support for traffic management system.In this paper,a traffic flow prediction model based on the combination of GRU neural network and BP neural network is proposed,and the feasibility of the model is verified by using the traffic flow data of Twin Cities in America.The experimental results show that the combined model has high prediction accuracy and can capture the fluctuating state of the traffic flow in the rush hour.Meanwhile,the model has strong robustness.The main work contents of this paper are as follows:1.Research on short-term traffic flow prediction based on GRU modelBecause GRU neural network has certain advantages in time series data prediction,we can build a prediction model of GRU neural network for traffic flow prediction.By comparing the evaluation indexes of the prediction model,the GRU model was repeatedly trained and the relevant parameters of the neural network were modified to minimize the loss function of the model.2.Research on short-term traffic flow prediction based on BP modelBP neural network has a good ability of nonlinear mapping and generalization,which can improve the application range of data to a certain extent.In this paper,BP neural network is selected to predict traffic flow,and the neural network model is established.By comparing the evaluation indexes of the prediction model,the relevant parameters of the neural network are optimized to minimize the loss function of the prediction model.3.Research based on GRU-BP combined traffic flow prediction modelIn order to further improve the accuracy of traffic flow prediction,based on the study of the above two traffic flow prediction models,the GRU-BP combined traffic flow prediction model was proposed.Compared with the single prediction model,the combined model can better extract the long-term dependency characteristics from the data.By taking advantage of the processing advantage of GRU neural network for time series data and the ability of BP neural network to mine complex data information,the combined model can reflect the long-term dependence of data into the network for further training,and finally achieve the purpose of improving the accuracy of short-term traffic flow prediction.