Identification of high-risk expressway segments using connected vehicle data: an empirical analysis

Area of Science:

• Transportation Engineering
• Traffic Safety Analysis
• Data Science

Background:

• Traditional road safety analysis relies on historical crash data, facing limitations like data scarcity and underreporting.
• Connected vehicle (CV) technology offers real-time driving behavior data, enabling proactive safety assessments.
• CVs generate critical warnings such as headway monitoring warnings (HMWs) and forward collision warnings (FCWs).

Purpose of the Study:

• To proactively identify high-risk expressway segments using CV warning data.
• To develop an integrated framework combining spatial hotspot identification and statistical regression modeling for road safety.
• To assess the relationship between CV warning frequencies and collision occurrences.

Main Methods:

• Utilized CV data from nine Shanghai expressways.
• Applied spatial hotspot identification techniques (Moran's I, Getis-Ord Gi*) to locate clusters of HMWs and FCWs.
• Employed Poisson, Negative Binomial, and instrumental variable Poisson models to analyze the association between warning data and collisions.

Main Results:

• HMW and FCW frequencies are significantly positively associated with collision occurrences.
• Accounting for endogeneity between traffic conflicts and collisions enhances estimation robustness.
• Segments identified as CV warning hotspots, even without prior collision data, show higher collision rates, indicating early risk detection.

Conclusions:

• CV warning data provides a valuable tool for the timely and proactive identification of road safety risks.
• The developed framework effectively integrates spatial analysis and statistical modeling for enhanced traffic safety assessment.
• This approach advances the application of CV data for proactive risk management on expressways.