Identification of accidental points of suburban roads using fuzzy modeling method(Case study: Qazvin-Lushan axis)

Accidental points are points where the probability of an accident is higher than the other points; Therefore, the identification of such points has a special place in all road safety projects. Due to the high number of traffic accidents in our country, the necessity of this issue is emphasized more than ever.  

In this study, fuzzy inference system of Sogno type was used for modeling of accidents based on influential independent variables. Initially, information from a two-year period was collected from accidents the Qazvin-Lushan in length of 72 km. Also  axis geometry-based data (such as street width, roadside user density, shoulder width, longitudinal slope, etc.) were collected through field survey. Then, fuzzy modeling for everyone was done by  dividing the desired axis into homogeneous sections of horizontal arc and directly. After validating the fuzzy model results, fuzzy model output values were presented for the input variables values.  

The results show the importance of street width, shoulder width, longitudinal slope, number of horizontal arcs and finally the sum of horizontal arc degrees on the dependent variable of the accident index in horizontal arc fragments. In addition, variables such as road user density, street width and shoulder width have been directly influenced by accident index. Finally, for the identification of hazardous parts, components whose their accident index (predicted by the fuzzy model) is above 90% confidence level, were identified as hazardous components.

The results show the proper performance of the model created by fuzzy inference method. In addition, using the fuzzy model has the advantage that in addition to correctly identifying the hazardous points, the effective factors and the severity of their impacts are also identified and helps to take corrective measures to increase safety in these areas.