Road traffic accident prediction for mixed traffic flow using artificial neural network

Abstract

Transport scenariosin developingcountries arefundamentally different fromthose indevelopedcountries. The latterconsists primarilyof passenger cars and can beadequately describedas“homogeneous” traffic,butthe formerconsistsof vehicle types withdifferentstatic and dynamiccharacteristics thatoccupy the same right ofway. Vehiclemovementis asynchronous. Fewstudies have attempted tounderstand the characteristics of mixed traffic. Thisarticleexplores the sharing attributes and influencing causes of traffic accidents inamixed trafficarea. A predictability model isemployed todescribethe connection between highwaydisastersand appropriate constraints such astraffic capacity, road provisions,and atmosphereissues. In this paper,the comparison has been done between the Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) predictive models. The study has been conducted at Pimpri Chinchwad Muncipal Corporation (PCMC) region of Pune, Maharashtra, India. For this work, nine years data has been used ranging from the year 2011 to 2019. Results revels that, maximum numbers of accidents were occurred in clear weather condition. Distinctive accidents were occurred due to overloaded vehicle conditions. Also it has been found that the less number of female drivers are responsible for accident. Forecasting model using ANN presents outstanding precision. In this study, additional prominence has been givento the real constraints which are accountable for accident cause in heterogeneous traffic flow.