Full Bayesian conflict-based models for real time safety evaluation of signalized intersections. (August 2019)

• Record Type: Journal Article
• Authors: Essa, Mohamed; Sayed, Tarek
• Is Part Of: Accident analysis and prevention. Volume 129(2019)
• Page Start: 367
• Page End: 381
• Publication Date: 2019-08
• Subjects: Automated conflict analysis -- Effect of signal design on safety -- Safety performance functions -- Real time safety evaluation -- Full Bayesian analysis; Accidents -- Prevention -- Periodicals; Accident Prevention -- Periodicals; Accidents -- Prévention -- Périodiques; 363.106
• DOI: 10.1016/j.aap.2018.09.017
• Languages: English
• ISSNs: 0001-4575

Abstract

Highlights: The study develops conflict-based safety performance functions for signalized intersections at the cycle level. Multiple traffic conflict indicators are investigated including the TTC, MTTC and DRAC. The independent variables include: traffic volume, queue length, shock wave speed and area, and the platoon ratio. The temporal distributions of conflict frequency and severity along the signal cycle are also investigated. The models can be used in real-time safety optimization of signalized intersection in the connected vehicles environment. Abstract: Existing advanced traffic management and emerging connected vehicles (CVs) technology can generate considerable amount of data on vehicle positions and trajectories. This data can be used for real-time safety optimization of intersections. To achieve this, it is essential to first understand how changes in signal control affect safety in real-time. This paper develops conflict-based safety performance functions (SPFs) of signalized intersections at the cycle level using multiple traffic conflict indicators. The developed SPFs relate various dynamic traffic parameters to the number of rear-end conflicts at the signal cycle. The traffic parameters included: queue length, shock wave speed and area, and the platoon ratio. The Time-to-Collision, the Modified-Time-to-Collision, and the Deceleration Rate to Avoid the Crash were used as traffic conflict indicators. Traffic video-data collected from six signalized intersections was used in the analysis. The SPFs were developed using the Full Bayesian approach to address the unobserved heterogeneity and the variation among different sites. Overall, the results showed that all the developed SPFs have good fit with all explanatory variables being statistically significant. Also, the highest conflict frequency was noticed at the beginning of the green time, while the highest conflict severity was noticed at the beginning of the red time. Lastly, the results can be used most beneficially in real-time safety optimization of signalized intersection.