Identification and Analysis of Risky Driving Behaviors at Intersections Based on Vehicle Trajectories

Intersections are bottlenecks that limit traffic efficiency and safety, with risky driving behaviors at intersections being key factors affecting both. By leveraging real-time vehicle trajectory data collected in an intelligent connected environment, we design threshold judgment methods for risky driving behaviors such as rapid acceleration, rapid deceleration, and sharp turns across different vehicle types and speed ranges. On this basis, three evaluation indicators-average time of rapid acceleration, rapid deceleration, and sharp turns—are used. The entropy weight method and TOPSIS are applied to analyze risky driving behaviors at ten intersections. The identification results show that the occurrence frequency of rapid acceleration and deceleration is related to the degree of traffic weaving, while the frequency of sharp turns is related to intersection saturation. The comprehensive analysis of risky driving behaviors indicates a significant variance in risk levels across intersections, with the TOPSIS scores ranging from 0.191 to 0.943.