Estimation And Prediction Of Expressway Section Flow Based On The Spatiotemporal Characteristics Of Traffic Flow

The section flow is an important parameter for describing the traffic state of the expressway.Effective estimation and prediction of it will not only help the traffic management department to reasonably configure the traffic facilities and traffic guidance,but also help the public choose the best travel route.At present,most of the researches on section flow are based on the sections where traffic detection equipment is located.Due to the sparse distribution of the existing traffic detection equipment,the scope of research sections is limited;and studies that use charge data to estimate section flow are mainly based on statistical analysis of historical data,so they are lack of timeliness.Therefore,the reasonable use of existing data sources to make more comprehensive and real-time accurate estimation of the section flow of the expressway is of great significance for improving the quality of expressway management.In this article,according to the different installation conditions of road section traffic detection equipment,the flow transfer coefficient is used to reflect the spatiotemporal relationship of the flow between the toll stations of the highway network,and the section flow estimation methods based on the single charge data and the combination of the vehicle detector and the toll data are proposed.Meanwhile,the data from the vehicle detector is used to correct the estimated results.Finally,the section flow prediction model is established by analyzing the time-space characteristics of the section flow time series.The main contents of the article are as follows:(1)Section flow estimation method based on charge data.In view of the situation where there is no section detection equipment such as a car detector in the road section,considering the factors such as the split and merge of vehicles in the highway interweaving area and the mutual influence of the vehicles between the ODs,etc.,will affect the driving state of the vehicle,and the downstream charging data when performing more real-time section flow estimation is different.In this article,the OD information of the charge data is used to estimate the travel time of the road section,and the flow transfer coefficient between the toll stations is calculated.Based on these,a cross-sectional flow estimation method based on toll data is proposed,and part of the expressway network in Chongqing is used.The charging data has experimented with the estimation method to verify the effectiveness of the method.(2)Section flow estimation method based on the vehicle detector data and charge data.According to the situation that the vehicle inspection device is installed on the road,the vehicle inspection device and the charge data are used to establish a section flow estimation model based on data fusion using a weighted average method.In order to further reduce the estimation error,an RBF neural network model is proposed.The experimental results show that the estimation model combining the vehicle inspection device and the toll data is better than the estimation model based on the single toll data.The RBFNN model has the correction effect on both estimation models and has better applicability.(3)Section flow prediction model considering the spatiotemporal characteristics of traffic flow.Since the current methods of section flow prediction mainly consider the regularity of time and lack the measurement of spatial characteristics,the Euclidean distance and gray correlation degree are used to select the suitable spatiotemporal characteristics that have obvious influence on the section flow.Then,according to the installation of highway vehicle detectors,based on the previous section flow estimation results,the LSTM neural network is used to establish a section flow prediction model that considers the spatiotemporal characteristics of traffic flow.Finally,the effectiveness and applicability of the section flow prediction model are verified.