Travel Time Reliability Analysis And Application Based On The Highway Toll Data

Travel time is one of important basic data in ITS. Dynamic and random traffic flow,infrequent events et al, determines the travel time of vehicle on the highway networkuncertainty. The unreliability of travel time will affect travelers’ choice of departure time,route and ways of travel. Therefore, increasing attentions have been put into the travel timereliability study. Travel time reliability studies must be support by long-term data. At present,travel time research oversea mainly based on vehicle detector data, but overall installationdetectors require a lot of investment and the data is not stability, it is difficult to collectcomprehensive and complete data. Closed highway toll management is used in China’shighway, and the charge management system records almost all vehicle information. Itprovides the possibility to study travel time reliability for the highway. In the age of big data,travel time reliability study based on highway toll data shows important significance.However, there are still some key issues must be resolved for the research based on toll data,including nose of data and data cannot track the vehicle’s trajectory. Aimed to these problems,specific studies as follow:（1）For the abnormal data, proposing a two-step data preprocessing methods based onprobabilistic screened and the highway speed limit. Experimental result proved that thismethod can effectively remove the effects of abnormal data for research. On the basis of datapreprocessing, fitting to distribution of the average travel time of vehicle starting fromdifferent time intervals and the same section of the day during three months. The distributionis proved following the lognormal distribution.（2）Because of using the original time record of highway toll system data couldn’t estimatetravel time accurately, which would directly affects the accuracy of traffic flow estimation;this paper proposes a new travel time estimation algorithm to calculate the space-timecoordinate displacement of vehicles according to the trajectory method. Furthermore, a speedcorrection algorithm is developed to adjust the travel speed. In addition, application with thedata in Guangzhou-Shenzhen freeway toll collection system, it has been proved that theestimated travel time after correction is more close to the actual travel time. Furthermore,using travel time estimation to estimate section traffic flow accurately. （3）The buffer index is selected as the evaluation of travel time reliability. By the means ofregression analysis, buffer index as the dependent variable, heavy vehicles ration, V/C andsection length as independent variables, establishing the empirical equation between traveltime reliability. Besides, the impact of various factors has been analyzed.（4）According to the research result, proposing the application scenarios of the traffic flowestimation and comparing with other detective methods which based on vehicle detector.Demonstrate its advantages over other methods in terms of cost-effectiveness. Besides,describing the application of travel time reliability in advanced traffic management system(ATMS).Moreover, there are abundant numerical example analyses in this paper, to prove thefeasibility and accuracy of the above methods. All of these meaningful conclusions fromnumerical example analysis not only enrich the harvest in the research results, but alsoproviding thinking and research results into practical application. The application of researchresults will make important contribution to improve the level of highway service and reducetransportation costs.