Pedestrian’s Accidents Prediction in Suburban Roads Using Artificial Neural Network (Case study of Amol city)

Abstract

Walking is considered as the most common type of transportation, because part of the route in all human journeys is necessarily done in the form of walking. In comparison to other transportation systems, walking has unique advantages, such as that it does not make environmental and noise pollutions, and neither does it require fuel consumption or cost pedestrians. Studies show that the most severe level of injuries especially in suburban roads is dedicated to pedestrians in traffic accidents due to high speed of vehicles. This issue asserts the necessity of analysis and more attention to the methods of reducing road accidents in the disaster- prone suburban roads. In this paper, Artificial Neural Network is used for the prediction of the traffic accidents in suburban roads of Amol city by focusing on pedestrians. This model proceeds to train the network by considering the parameters of road's length and width, the presence or absence of median, surface quality, surrounding land uses fitness, police presence, route's brightness quality and presence or absence of gradient in the route of twelve suburban roads, and then proceeds to test it with the parameters of Amol-MahmoodAbad road. The result shows an accuracy of 85 percent in the pedestrians' accidents prediction process. Then, for further validation, the model was tested with the parameters of Amol-Babol' new road and the result showed the accuracy of 83 percent in the prediction process.