Real-time crash risk prediction models using loop detector data for dynamic safety management system applications.

Abstract

There is a growing trend in development and application of real-time crash risk prediction models within dynamic safety management systems. These real-time crash risk prediction models are constructed by associating crash data with the real-time traffic surveillance data, collected by double loop detectors. The objective of this paper is to develop a real-time prediction model that will potentially be utilized within safety management systems. This model aims to predict the traffic safety condition of a motorway where potential prediction variables are confined to traffic related characteristics. Given that the dependent variable (i.e. traffic safety condition) is considered dichotomous (i.e. “no-crash” or “crash”), the binary logistic regression technique is selected for model development. The crash and traffic data used in this study were collected between June 2009 and December 2011 on a part of the European route E313 in Belgium between Geel-East and Antwerp-East exits, on the direction towards Antwerp. The results of analysis show that several traffic flow characteristics such as standard deviation of speed and occupancy at the upstream loop detector, and the difference in average speed on upstream and downstream loop detectors are significantly contributing to the crash occurrence prediction. The final chosen model is able to predict more than 60% of crash occasions while it predicts more than 90% of no-crash instances correctly. The findings of this study can be used to predict the likelihood of crashes on motorways within dynamic safety management systems.