Highway Level of Service Evaluation Using the Reliability Model

The highway network toll data are used for extracting the travel time distribution. Three indexes are put forward to compare the fitting results, including error sum squares (SSE), the determination coefficient (R²), and cumulative distribution of the error sum squares (SSE of CDF). The results show that it is the most appropriate to model the travel time with the extreme value distribution, compared with log-normal, normal, and Weibull distribution models. On the basis of annual average daily traffic data, the level of service (LOS) is quantified by the evaluation index of the service reliability. The evaluation method of highway LOS is established. Finally, an example of verification is carried out on the highway sections in the Shaan’xi province. Evaluation results show that reliability can reflect the quality of highway sections and the fluctuation objectively. It is more suitable than the traditional LOS on describing the actual conditions in highway networks. The proposed method is effective as the calculated result is consistent with the actual observation.