Professor Wang Chen's latest research findings published in the top journal of traffic safety

Recently, Professor Wang Chen's team at the School of Transportation of Southeast University, and Professor Xie Yuanchang's team from the University of Massachusetts Lowell, have made a significant breakthrough in the risk modeling of single-vehicle road departures on curved sections of highways (SVROR). To more realistically represent the dynamic risk of vehicles when turning on highways, this study, for the first time, introduced the diquark-antidiquark particle structure from high-energy physics and proposed a method for measuring the driving risk on highway single-vehicle curved sections (SVROR-CRM). This research has filled the gaps in the field of crash risk measure (CRM) for single-vehicle accidents, and the method can effectively quantify the risk of SVROR accidents, identify the driving risk trajectories, high-risk locations, and risk periods on horizontal curves of highways.

In highway curves, single-vehicle run-off-road (SVROR) accidents are influenced by multi-dimensional factors such as road alignment, road configuration characteristics, vehicle motion, driver operation, and traffic environment, which often lead to vehicle rollovers and severe casualties. However, the existing methods for measuring accident risks (such as traffic conflict technology, safety field theory) mainly focus on the collision risks between traffic participants, such as vehicle-to-vehicle and vehicle-to-pedestrian conflicts, and there is no risk measurement method specifically for modeling the risk of SVROR accidents on horizontal curves. In light of the limitations of current research and the development trend of connected vehicle (CV) technology, this study, relying on the high-granularity characteristics of CV data, has pioneered a method for measuring the risk of SVROR accidents. The main contributions are:

The diquark-antidiquark particle structure from high-energy physics was introduced to model the interaction between a single vehicle and the road on horizontal curves. A revised position deviation risk force and a revised attitude deviation risk moment were proposed to explain the spatial and motion state relationship between the vehicle and the driving lane, fully expressing the potential risk of single-vehicle run-off-road (SVROR) accidents;

By establishing the intrinsic relationship expression between model coefficients, a self-estimation of coefficients driven by connected vehicle (CV) data was realized, avoiding the over-reliance on accident data of existing methods, and these coefficients potentially consider the characteristics of driver attributes, road conditions, and the heterogeneity of the human-vehicle combination.

A method for the objective selection of risk thresholds based on the mean absolute error function was developed, enabling the automatic selection of risk thresholds from the initial threshold range;

Unlike the existing consistency tests at the same location but different times, this study designed a site risk ranking consistency test index (RCTI), which can effectively verify the adaptability of accident risk measurement methods for risk estimation at different locations.