Method for Estimating Highway Collision Rate that Considers State of Traffic Flow

Among traffic-related variables, vehicle speed is thought to be a major one closely related to collision occurrence. This study aimed to develop a methodology to estimate a collision rate that considers the traffic state in a freeway section and focuses on speed variables. To define the traffic condition of a section, four section-based traffic states related to the speed of upstream and downstream locations were used. The collisions were classified into three types (rear-end, sideswipe, and other), and two variables (speed difference between upstream and downstream and average speed of upstream and downstream) were considered as independent for standard linear regression analysis. The findings showed that both speed difference and average speed had an influencing effect on the collision rate for each type of collision and traffic state. Rear-end and sideswipe collisions showed a different collision rate pattern than the others, and the influence of the speed difference between upstream and downstream changed with the traffic state. Through regression analysis, rear-end and sideswipe collision rates were expressed by speed difference while other types of collisions were described by average speed. This research can provide a methodology for understanding the collision potential of a highway as it relates to the traffic state and can be widely used for roadway traffic safety improvement.