Research On The Toll Lane Configuration Of Expressway Toll Station Based On Congestion Prediction

The traffic volume of expressway toll stations has a time-varying characteristic,and the traffic demand is different in each period.Due to the lack of a reasonable and effective toll lane configuration method,the traffic congestion and idle of toll lane are increasing prominently.Therefore,waste of resources and rising operating costs are becoming more common.In order to improve the traffic efficiency and operation management level of toll stations,the toll lane configuration method based on queuing theory and deep learning is proposed in this paper.The specific research contents are as follows:(1)The classification of toll stations,types of toll lanes and traffic characteristics are analyzed.Aiming at the abnormal data and missing data,the processing rules and analysis flow of abnormal data is proposed.Taking the toll data of Dongshe toll station in Shanxi Province as an example,the feasibility of the data processing method is verified;(2)Based on the analysis of the proportion of multiple charging modes and the distribution of service time,a queuing model of multiple charging modes is proposed.The service time,traffic volume,proportion of multiple charging modes and vehicle type ratio are taken as model input.In addition,the calculation of service time of multiple charging mode is realized and the relevant theory of queuing model is improved;(3)In this paper,the average queue length is proposed as a measure of congestion in toll stations.The time series of the average queue length of the lane is obtained by using the toll data and the queuing model of multiple charging modes.PSO algorithm is used to optimize the number of hidden layer nodes and iterations of LSTM model,and a congestion prediction model of toll station based on PSO-LSTM is proposed.Taking the average queue length,traffic volume and service time as input to predict the average queue length of lane.Taking Dongshe,Changfeng and Linfen toll stations in Shanxi Province as examples,the algorithm is verified and compared with SVR model.The results show that the MAPE of PSO-LSTM model is 2%-3% higher than that of LSTM model and SVR model,which proves that the optimized prediction model improves the prediction accuracy.(4)A toll lane configuration model is proposed based on service level.Considering the relationship between capacity and queue length of toll station,the service level of toll station is divided into four levels.Based on the service level,a toll lane configuration model is proposed.The predicted average lane length of toll lane is taken as model input and the number of toll lane configuration is taken as model output.Based on VISSIM software,a visual model of toll lane configuration is proposed.In addition,the sensitivity of the parameters that affect the model results is analyzed.The results prove that the greater the traffic volume,the greater the number of toll lanes,the larger the proportion of small cars,the lower the number of toll lanes,the higher the mobile QR code usage ratio,the smaller the number of toll lanes and the smaller the effect of swipe payment on the number of toll lanes;(5)An evaluation model based on operating costs is proposed.Taking Dongshe,Changfeng and Linfen toll stations in Shanxi Province as examples,the lane configuration is analyzed.It is proved that the result of the toll lane configuration model is reasonable.