Scholarlink

Tunnels are bottlenecks in highway traffic, characterized by a closed driving environment. Due to the unique nature of the traffic environment within tunnels, traffic safety accidents occur frequently. However, there are still certain limitations in the current research on tunnel safety evaluation. This paper utilizes the advanced object detection technology YOLOv8 to extract traffic state parameters of vehicles from tunnel surveillance videos, and employs an extreme value theory model based on Time to Collision (TTC) to assess the collision risks of different lanes within the tunnel. Experimental results indicate that the collision risk is lowest in the slow lane. Although the conflict rates for the fast lane and the middle lane are similar, the collision risk in the fast lane is greater than that in the middle lane.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Traffic Conflicts-Based Crash Risk Assessment at Tunnels Using Extreme Value Theory Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Chuanyun Fu,
  • Han Yan,
  • Zhaoyou Lu,
  • Jiaming Liu,
  • Huahua Liu,
  • Wei Bai

摘要

Tunnels are bottlenecks in highway traffic, characterized by a closed driving environment. Due to the unique nature of the traffic environment within tunnels, traffic safety accidents occur frequently. However, there are still certain limitations in the current research on tunnel safety evaluation. This paper utilizes the advanced object detection technology YOLOv8 to extract traffic state parameters of vehicles from tunnel surveillance videos, and employs an extreme value theory model based on Time to Collision (TTC) to assess the collision risks of different lanes within the tunnel. Experimental results indicate that the collision risk is lowest in the slow lane. Although the conflict rates for the fast lane and the middle lane are similar, the collision risk in the fast lane is greater than that in the middle lane.