ANN based Study to Investigate the Parameters Influencing Collision Type on a Four Lane Divided National Highway

Abstract

Due to the mixed traffic conditions in developing countries like India, there is an exponential growth of road accidents over the decade which leads to the deterioration of road safety. Therefore, road safety has become a major concern for researchers and engineers. It is very important to know the effect of influencing variables on the crash count. Although the data to collect about the influencing variables for the crashes is a challenging task it is very important to know the effect. In this study, a four-lane divided national highway is considered and the analysis is done for the crash records for five years which is from 2013–2018. From the plan and profile drawings of the highway, different geometrical characteristics are extracted. Traffic characteristics are collected from the field studies. Stochastic models were developed to know the effect of selected variables on collision type. Using soft computing tool Artificial Neural Network is developed and the significance of the influencing variables on collision type is known.