Risk characteristics analysis of road segments: Considering multiple scales and temporal stages

Different from research that statistically models discrete conflicts in space and time, this study focuses more on the dynamic process of conflicts and proposes a continuous multi-scale method for analyzing the risk characteristics. Firstly, using conflicts as a reference point, three scales of traffic entities-vehicle pair, vehicle cluster, and vehicle group-are defined based on the interaction range. Corresponding risk expression models are constructed for each scale. Subsequently, considering the temporal process of conflict formation, maintenance, and dissipation, the dynamic sequential structure is established. Next, for risk level at different scales, Spearman correlation analysis and Friedman test are employed to investigate the traffic features and their stage differences. Finally, road segment risk level is differentiated into four temporal patterns, and an unordered multinomial Logistic regression analysis is adopted to explore the occurrence conditions for each pattern. The findings indicate that: (1) Risk levels do not strictly follow a monotonic increase or decrease, instead showing dynamic variations; (2) Traffic entities at different spatial scales (such as vehicle pairs, vehicle clusters, and vehicle groups) exhibit significant differences in risk-related characteristics during the stages of conflict formation, maintenance, and dissipation; (3) Unimodal low-risk patterns and unimodal high-risk patterns are the dominant risk evolution modes, with mean speed identified as the most critical precursor variable influencing these patterns. This study provides an analysis of the conflict development process across multiple spatial scales and temporal stages. It reveals notable differences in risk characteristics and their spatiotemporal evolution among different traffic entities. This multi-dimensional approach offers a perspective for more thoroughly describing and analyzing the evolution of traffic risk and holds implications for improving road traffic safety management.