Short-term Traffic Flow Prediction Based On Deep Learning

With the continuous improvement of urban intelligent transportation system,the city road to install a large number of detector equipment to collect road and vehicle information data,by analyzing the detector to detect a large number of data,from the data found in the traffic laws to support urban traffic Optimization,short-term traffic flow analysis is one of the main research problems.Convolution neural network avoids the problem of extracting features from the data manually.The convolution neural network is used to extract the temporal and spatial characteristics of traffic flow automatically and has good application value.Based on this method,the convolution neural network is used to predict the short-term traffic flow.The main contents are as follows:In this paper,by constructing the convolution neural network model,the traffic flow data of the road network is transformed into traffic congestion level data,and the traffic condition prediction based on such data is less than 5 minutes.The smaller case of the specific type of sample is realized,Using the idea of migration learning to increase the amount of training set of data and enhance the predictive performance of the model.The main work of this paper is as follows:(1)In order to solve the short-term traffic flow forecasting problem,the convolution neural network is used to solve the problem.The influence of the convolution kernel size on the prediction accuracy of the convolution neural network in the scheme is analyzed.Day of the data set,to achieve a 1 minute and 5 minutes of traffic forecast.And to achieve the peak and all-day data forecast.(2)In the case of too few data training samples,the problem of over-fitting of the model is adopted.By using the idea of migration learning,the data samples are increased by the different data sets in different time periods and the method of adding random perturbations in the training set,accuracy.