Optimization Of Highway Maintenance Construction Organization Scheme Based On ETC Gantry Dynamic Traffic Data

Since 2019,the electronic no-stop toll collection(ETC)system has gradually been popularized at high-speed toll stations nationwide.In addition,national highway construction has become saturated,and the focus of highway work has shifted from "high-speed construction" to "intelligent management".How to combine ETC data with the traffic organization scheme setting of road sections in road maintenance construction areas,and use data-driven traffic state simulation to dynamically evaluate the service level and safety level of construction areas,so as to improve the overall highway traffic efficiency during maintenance and reduce the risk of traffic conflict,road congestion time and construction operation cost.It is the research hotspot of traffic organization optimization in expressway construction area at present.In this paper,the optimization of traffic organization scheme in the maintenance and construction area of two-way four-lane expressway was taken as the research objective.Firstly,preliminary cleaning,outlier detection and noise reduction were carried out on the ETC data of Zhejiang YTW expressway.Secondly,the traffic flow characteristics of normal sections in the upstream of the construction area are analyzed,and the dynamic prediction of traffic flow parameters is carried out based on particle swarm optimization and short and long time memory neural network(PSO-LSTM)model.Then,a dynamic comprehensive evaluation method based on time-series traffic flow data was constructed to select evaluation indexes from two dimensions of service level and traffic safety and establish evaluation criteria.Finally,VISSIM software and dynamic comprehensive evaluation method were used to evaluate the maintenance construction organization schemes of different construction types,different periods of traffic and different work lengths to get the optimal scheme.The research mainly draws the following conclusions:(1)The data noise reduction processing based on travel time is proposed.Wavelet transform was used to denoise the data set.The best wavelet basis function was coif1,the best decomposition scale was 4 layers,and the soft and hard threshold coefficient was 1.0.(2)In the dynamic prediction model of traffic flow parameters,PSO is used to determine the optimal LSTM model hidden layer node number h1 is 22,h2 is 8,iteration times is 200;Compared with the SVR algorithm,the optimized LSTM model improved by 10.312% MAPE and reduced the running time by 17.905 s.(3)In the dynamic comprehensive evaluation of the organization scheme of the construction area,the service level indexes were selected as section saturation,average vehicle speed,traffic density and average trip delay,and the traffic safety level indexes were selected as rear-end collision(TTC)and lane change conflict(PET),which were divided into three evaluation criteria,with thresholds of 0.143 and 0.752,respectively.(4)The optimal maintenance work area length is2000 m and the optimal construction time is 12:00-16:00 or 02:00-04:00 through VISSIM simulation.This paper aims to provide highway managers with real-time traffic operation status and development trend,and do a good job of traffic guidance and traffic control measures in maintenance and construction areas,so as to reduce the adverse impact of maintenance and construction operations on the operation status of highways.