Identification of accident-prone sections in roadways with incomplete and uncertain inspection-based information: A distributed hazard index based on evidential reasoning approach

Identification of accident-prone road segments based on the results of safety inspection is a widely accepted method for prioritization of safety improvement efforts particularly for low volume traffic roadways. In the existing methods, the items listed in a risk factor checklist are given qualitative or exact scores and then risk items are aggregated without considering the uncertainty in the subjective judgment of the inspector. Other shortcomings include the failure to consider the relative importance of risk factors and the failure to provide a solution to deal with absence or incompleteness of data. This study introduces a method of road safety evaluation and a distributed hazard index (DHI) based on evidential reasoning approach. This approach allows the risk factor to be expressed by several degrees with different values of confidence instead of one definite figure. By adjustment for the extent of user exposure and the approach of road safety authority to improvement, the distributed index is turned into the Hazard Index (HI), which serves as measure for final prioritization of road sections. The proposed method is applied to a case study and the resulting sections ranking show good consistence with the results of empirical Bayesian method based on accident records.