Expressway Travel Time Reliability Model Without Incident Based on the Extreme Value Distribution

Massive historical data of the expressway toll was collected and fitted by several distributions, including extreme value, log-normal, normal and Weibull. Travel time reliability model was established based on the extreme value distribution, and a parameter estimation method was provided. To be understood by travelers easily, the threshold was also built considering the difference between cars and trucks. Three indexes were selected as the evaluation index, including the maximum probability travel time, the average delay time and travel time reliability. The proposed model was validated by the case study of Shaanxi. Based on the principles of the less error sum squares (SSE) and the higher determination coefficient ( R ²), the results show that it is rational and acceptable to describe the travel time with the extreme value distribution.