Prediction the number of run-off-road accidents in the horizontal curves of rural two-way roads based on characteristic of horizontal curves, case study: ILAM-MEHRAN road

Regression models prediction are one of the best tools to identify the factors affecting the occurrence of road accidents. Horizontal curves are one of the segments of the road network that have high risk of occurrence road accidents, Run-Off-Road(ROR)accidents are the most common accidents in these parts. Therefore, regression prediction models can be used to identify the contributories factors occurrence of ROR accidents. In this study, by using data related to ROR accidents (according to policedatabase 2010-2017), one of the busiest two-lane roads in ILAM province(ILAM-MEHRAN) and types regression prediction models, the factors affecting the occurrence of such accidents were investigated. Traffic volume and characteristics of horizontal curve (including radius, central angle, tangent length, curve length, etc.) were used from Data andmaps of the road organization. The results showed that exponential regression and power regression are the best type of predictive model to identify the parameters affecting the occurrence of this type of accidents. The exponential regression prediction model showed that the radius of the curve, the length of the circular curve and the total length of the curve are the factors influencing the occurrence of ROR. But the power regression prediction model proved that the radius of the curve, the lengthof the circular curve and the length of the tangent are the factors influencing the occurrence of ROR. It is hoped that the results of this study will be used for prioritization in the correction of accidental horizontal curves.

Keywords:

Run-off-road accidents , regression models prediction , horizontal curves , rural two-way roads and geometric design.