Real-time highway accident prediction based on grey relation entropy analysis and probabilistic neural network

Be different from the traditional highway traffic accident prediction that focused on historical data analysis, this study attempts to predict the accident by using real-time traffic data. The occurrence of a traffic accident on highway is associated with the short-term turbulence of traffic flow. This study aims to select the main factors that represent the turbulence of traffic flow by using grey relation entropy analysis. Then this study discusses how to identify the traffic accident potential occurrence by using probabilistic neural network. The traffic data are collected from the traffic simulation software VISSSIM. The experimental results show that it is promising for real-time highway traffic accident prediction by using these models.