Research on Travel Time Distribution Characteristics of Expressways in Shanghai

The distribution characteristics of travel time, which present the uncertainty of travel time fluctuation, are valuable for the study on traffic assignment, traffic guidance, network design, and so on. In this paper, we choose four routes of expressways in Shanghai as the research objects. The travel time of the study routes is derived from floating car data and the distribution characteristics of travel time are analysed in four typical periods in a day. Based on the statistical analysis of travel time, we conduct a comparative study of the statistical distribution of travel time using Kolmogorov-Smimov test and Anderson-Darling test. The hypotheses of the fitting tests include Beta distribution, Normal distribution, Lognormal distribution and other four kinds of distribution. It is concluded that the travel time fluctuation have features of pike apex and thick trail on the routes with obvious peak hours. Concretely, the travel time distribution cannot be well fitted by Normal distribution in most cases.Beta distribution is the most capable of describing the distribution pattern of the routes with obvious crest in peak periods. However, for these routes in common periods or other routes in most of periods, Lognormal distribution has the best goodness of fit.