Study On Traffic Flow Forecasting Of Expressway Based On Convolutional Neural Network

With the continuous increase of motor vehicle ownership and frequent occurrence of road congestion,reliable traffic flow prediction technology can provide data support for road traffic management,help motor vehicle drivers to plan their travel routes,and play an inducing and controlling role in road traffic flow.Therefore,it has become the core content of the intelligent transportation system.In recent years,with the application value of deep learning in traffic big data mining becoming more and more powerful,all kinds of artificial neural networks have been widely used in traffic flow prediction research.As an excellent deep learning model,convolutional neural network is stable and accurate,which determines that it may have a good effect in the prediction field.The emergence of many learning platforms and network frameworks,such as Caffe framework,greatly improved the operability of neural network construction.With the neural network as the framework and traffic flow sequence as the research object,the prediction research has certain practical significance.Based on the traffic flow characteristics of the expressway,this paper constructs the expressway traffic flow prediction model based on convolutional neural network.The main research work is as follows:(1)based on the measured traffic flow data of Beijing third ring expressway,the paper discusses and analyzes the traffic flow characteristics and the temporal and spatial correlation of traffic flow data on the urban expressway.The contents include:the original traffic flow data processing,analysis of three-parameter characteristics of expressway traffic flow,discussion of periodicity of expressway traffic flow,correlation between spatial correlation of traffic flow and distance between monitored sections,etc.,laying a theoretical foundation for traffic flow prediction model.(2)according to the characteristics of urban expressway traffic flow and the spatial correlation of traffic flow,the traffic flow data in the space-time dimension are combined to take the form of a two-dimensional matrix as the network input,and the Caffe framework based on deep learning is used to design the prediction model of expressway traffic flow based on convolutional neural network.The model considers the historical traffic flow of the predicted section and the traffic flow of the upstream and downstream sections,and captures the road range with high spatial correlation to limit the model structure.(3)The prediction model of expressway traffic flow based on convolutional neural network is verified and analyzed.The model is trained,tested and predicted based on the traffic flow data of Beijing third ring expressway.The results show that the prediction model of expressway traffic flow based on convolurelatively good prediction results,and the prediction accuracy is about 88%.tional neural network hasThe research in this paper can provide a basic method for the prediction of urban expressway traffic flow,as well as an idea and theoretical basis for the construction of urban road traffic management and control,intelligent traffic flow guidance system based on artificial neural network.The full text includes 31 figures,15 tables,49 references.