Safety risk assessment for intersection rear-end conflicts based on collision risk force indicator

This study introduces a novel method for assessing the risk of rear-end conflicts based on collision risk force indicator. Drawing on the theory of road user safety fields, we construct a collision risk force indicator tailored for rear-end conflict scenarios. By analyzing the impact of driver attributes and vehicle features on risk force, as well as key road environmental factors influencing rear-end accidents, we propose a specific calibration method for the collision risk force indicator. This is further refined using historical accident data for parameter calibration. We then employ the Peak-Over-Threshold (POT) model and utilize the collision risk force values as input indicator to predict the annual average accident frequency. As a case study, we select intersections in Guilin City and extract vehicle trajectory information. The calibrated collision risk force method is applied to obtain accident prediction values for the intersections. The results demonstrate that the predicted annual frequency of rear-end accidents, based on the collision risk force indicator, aligns with the 95% Poisson confidence interval of actual accidents, validating the accuracy of our research method. The calibrated collision risk force indicator serves as a non-accident indicator for evaluating the safety risks of rear-end conflicts at intersections.