Research On Highway Traffic Flow Characteristic Analysis And State Recognition Based On Toll Data

The continuous improvement of economy and unlimited innovation of science and technology support the rapid development of expressway in China.Today,the expressway network toll system has been mature,and the toll data has achieved a comprehensive unity.The accumulated massive toll data can mine rich information of traffic flow operation characteristics.Under the condition of not adding new traffic detection equipment,using toll data to analyze and process traffic state parameters and identify traffic status is of great significance to improve the efficiency of expressway operation and management and solve expressway traffic congestion.This paper takes the analysis of time-space characteristics and traffic state identification of expressway traffic flow as the key research object.Through collecting the toll data recorded during the dynamic operation of the expressway toll collection system,including the location of the entrance and exit of the toll station,the time of passing through the entrance and exit,the vehicle type and so on,using the relevant technologies of data statistics and data mining for reference,the basic framework of data analysis is built under the platform of Oracle and Python,and the information extraction and analysis of the toll data is realized by using various data processing methods.The results of this paper are as follows:(1)Data quality will affect the effect of data mining,but data preprocessing can improve the data quality.A large number of charging data are stored in the database.In view of a series of quality problems,five kinds of abnormal data are summarized,and corresponding processing rules are made for different types of problems.The results before and after data processing are compared objectively,and the feasibility of the processing toll data method is verified.(2)Selecting a toll station of an expressway as the research object,count the flow and total amount of all kinds of vehicles and calculate the proportion of different models in different time periods,analyze the travel characteristics of citizens on working days,weekends and holidays through the time-varying characteristics of traffic flow,analyze the changing rules of holiday and non-holiday flow respectively,and understand the operation characteristics of traffic flow.A specific travel interval is selected and the vehicle travel speed distribution is fitted according to the vehicle type.The travel speed probability density of the six kinds of traffic flow all obeys the Gaussian distribution.(3)Considering the defects of traditional FCM algorithm,this paper takes the three indexes of unit traffic flow,unit travel time and occupancy as characteristic parameters,puts forward the idea of using genetic algorithm,entropy weight method and generalized equilibrium method to optimize them,and constructs the expressway traffic state discrimination decision-making model based on GA-GEFCM algorithm.Taking the toll data of a certain expressway as an example,two algorithms are used to solve the clustering centers of four kinds of traffic states: unimpeded,stable,crowded and congested.Then the membership matrix is used to identify the above traffic states,which proves that the algorithm in this paper is more stable,reliable,effective and feasible.