Multi-scale Temporal Analysis of Connected Vehicle Data for Safety Analysis

Abstract: Abstract
Highway safety analysis and crash modeling present significant challenges due to the complex variability of roadway, vehicle, and driver characteristics. Relying solely on traditional crash data can often overlook essential risk factors, including near-misses and minor collisions that fall below typical reporting thresholds. To address these gaps, this study leverages Connected Vehicle (CV) data collected across Oklahoma to assess traffic parameters, such as hard braking events, vehicle speeds, and traffic volume, in order to identify potential safety risk zones. The study analyzed three months of CV data from January, May, and August 2021, aggregating it at both hourly and daily intervals to capture seasonal variations and their effects on traffic behavior. Additionally, data from May 2020 were incorporated to assess the impacts of the COVID-19 pandemic on traffic streams. Through spatial and temporal analysis, this research revealed distinct temporal patterns by examining different hours (rush versus non-rush), days (weekdays versus weekends), and seasons. By correlating these patterns with crash data from the Oklahoma highway collision database, spatial and statistical analyses identified associations between CV data and crash frequency. The research proposes potential surrogate safety measures for identifying crash hotspots and underscores the critical factors influencing highway safety.

Keywords: Highway Safety, Connected Vehicle (CV) data, Traffic Patterns, Statistical Analysis