Research On Expressway Traffic Operation State Based On Network Toll Data

With the continuous construction of the expressway,the road network is gradually improved,and the expressway network has become the economic lifeline of our country.As an important part of the expressway construction,the toll system records a large number of vehicle toll data.With the development of the new generation of artificial intelligence and computer technology,data mining is also applied to the field of transportation,with the development of intelligent transportation year by year.How to make full use of the data in the expressway toll system,mining valuable information,providing reliable information for the traffic travelers and road managers,and providing theoretical support for the development of intelligent transportation become the direction of the further development of expressway.This paper takes this as the starting point,through the study of highway toll data,to explore the road network traffic operation state,and apply machine learning method to the highway traffic operation state research.It provides data support for travelers and traffic managers to evaluate road condition.1.High quality data can improve the accuracy of data mining,reasonable data processing methods can get more effective information,so data mining needs to be preprocessed before.First,through data cleaning,integration,extraction and elimination of obvious error data,the preliminary effective data is generated,and the data is screened by quartile method and statistical principle.The experimental results show that the data within the upper limit and lower limit of quartile method can contain most of the effective data.Effective data screening lays the foundation for further research.2.In order to explore the reliability of vehicle travel time,the estimation of travel time distribution is the basis of calculation reliability.The traditional travel time reliability distribution model includes: Normal distribution,Log-normal distribution,Weibull distribution,Gamma distribution,etc.In addition to summarizing the traditional model,this paper also uses the mixture gauss distribution model to fit the travel time distribution,EM algorithm to estimate the parameters and test the goodness of fit.Finally,the travel time reliability is obtained by integrating the probability density function of the Gaussian mixture model.The results show that the travel time distribution of small vehicles meets the Gaussian mixture distribution model,and has a good fitting effect.3.In the existing papers,density,speed and headway are often used as the characteristic parameters of traffic flow.The data collection costs a lot of cost and has a large error.Highway toll data accurately records the time of vehicles entering and leaving toll station,and can accurately calculate the travel time of vehicles in the road section,with small error.However,this part of data can not accurately determine the traffic status of vehicles in a certain period of time in the driving section.This paper discusses this issue,combining the "reliability" with the "operation status",using the unblocked reliability,the basic unblocked reliability and the light congestion reliability as the characteristic parameters of the road section to distinguish the traffic status of the road section.4.In this paper,the Fuzzy C-Means clustering algorithm is used to distinguish and analyze the traffic operation state.The traditional Fuzzy C-Means clustering algorithm is generated randomly when the initial clustering center is selected,which results in the decrease of the efficiency.In this paper,K-means algorithm is used to select the initial clustering center,and the clustering center of K-means algorithm is used as the initial clustering center of Fuzzy C-Means clustering,so as to improve the operation efficiency.The improved clustering algorithm is used to distinguish the traffic state.The research shows that the improved algorithm can effectively improve the efficiency of operation.At the same time,the improved clustering algorithm is used to evaluate the traffic state of a road section by selecting the data of a certain day in the working day and holiday respectively.It is found that the congestion degree of the road section in the holiday is greater than that in the working day,and the congestion degree of the road section in the afternoon is higher than that in the morning.This paper combines "reliability" and "traffic operation status",adopts unblocked reliability,basic unblocked reliability,and light congestion reliability as the characteristic parameters of the road segment to distinguish the traffic state of the road segment.By improving the Fuzzy C-Means clustering algorithm,the operation efficiency is improved.It provides reference value for the traffic management department to analyze the road network status from another angle.