Accidents are multi-causal phenomena and could not be controlled without taking all the causes into account. Among the most important factors affecting the classification of the pedestrian-hazardous (accident-causing) walkways are accidents’ environmental factors– which have not yet been considered comprehensively using existing methods. Statistics and data about the current situation are complex and non-linear. Thus, the classification of these passageways cannot be done correctly using simple mathematical models and linear combinations.
In this research, an observational classification method using artificial neural networks is presented, and a new model for pedestrian-oriented classification of hazardous urban walkways has been proposed.
In this model, artificial neural networks and statistics data on pedestrian accidents and walkway risk level are used, while control, traffic, and geometric parameters are considered simultaneously. The factors involved in this model include length and width of the roadways, width of the walkways, parking on the sides of the roadways, median’s width, roadway’s speed limit, roadway’s lights, quality of upper asphalt layer, segregation of roadways and walkways, presence of police, proportionality of velocity and land use, and proportionality of pedestrian pathways and gradient of roadway.
Being observational, this model could significantly increase the accuracy and efficiency of the process of classifying pedestrian-hazardous urban passageways.