Analysis on Sideswipe Collision Precursors Considering the Spatial-Temporal Characters of Freeway Traffic

摘要: The study attempts to investigate the precursors sideswipe collisions through analyzing correlation between and road geometric, environmental, real-time traffic factors. To avoid ‘bias’ caused by high percentage of rear-end in generic models, only observed on I-5 central Puget Sound area are used analyze matched case-control logistic regression model. Considering spatial-temporal character flow short-term variation collision occurrence, a comprehensive analysis occurrence its relationship with freeway across lanes detector locations is conducted. results imply that more likely occur at straight level segments multi-lane off-peak hours. On one hand, average occupancy low speed variance upstream location tend increase probability congested scenarios. other coefficient speed, occupancy, absolute difference adjacent downstream may decrease likelihood uncongested conditions. Furthermore, Average have an increasing effect collision, standard deviation occupancies speeds exactly opposite. findings this suggest variables, such as ratio vehicle lane changing lanes, further supplemented identify effectively differential statistically significant factors collisions.