Scholarlink

<p>The urban expressway network is a critical infrastructure for modern transportation systems, with the weaving area being a key component. However, the complex lane-changing behaviors in these areas, particularly overtaking lane-changing (OLC), have become a primary bottleneck affecting traffic safety and efficiency. Existing studies predominantly focus on merging/diverging zones or descriptive analyses, lacking causal insights into driver decision-making under dynamic conditions. This gap limits proactive safety measures. Our study addresses this by employing causal inference and counterfactual estimation to quantify how driver perceptions, traffic density, and inter-vehicle distance influence OLC acceptance. Integrating machine learning, we propose actionable strategies for real-time OLC control. Through a decision-making experiment, we collected the decision data on OLC behavior. Average treatment effect and conditional average treatment effect are calculated to identify key indicators influencing driver decisions during OLC maneuvers. Combined with machine learning techniques, we also developed a counterfactual estimation model to estimate overtaking acceptance under specific interventions. The estimation results can be used to support active control of OLC behavior. The findings from this research provide a deeper understanding of the intricate relationships and mechanisms underlying OLC behavior. The predictive model offers a significant advancement, enabling more accurate evaluation of current systems and informing the development of more effective transportation solutions. The study’s implications are crucial for optimizing traffic flow and safety within urban expressway interchange weaving areas.</p>

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

Causal inference-based research on decision characteristics and active control of overtaking lane-changing behavior in expressway interchange weaving areas

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Zhen Zhou,
  • Yi Zhao,
  • Renteng Yuan,
  • Minghan Yang,
  • Ziyi Shen

摘要

The urban expressway network is a critical infrastructure for modern transportation systems, with the weaving area being a key component. However, the complex lane-changing behaviors in these areas, particularly overtaking lane-changing (OLC), have become a primary bottleneck affecting traffic safety and efficiency. Existing studies predominantly focus on merging/diverging zones or descriptive analyses, lacking causal insights into driver decision-making under dynamic conditions. This gap limits proactive safety measures. Our study addresses this by employing causal inference and counterfactual estimation to quantify how driver perceptions, traffic density, and inter-vehicle distance influence OLC acceptance. Integrating machine learning, we propose actionable strategies for real-time OLC control. Through a decision-making experiment, we collected the decision data on OLC behavior. Average treatment effect and conditional average treatment effect are calculated to identify key indicators influencing driver decisions during OLC maneuvers. Combined with machine learning techniques, we also developed a counterfactual estimation model to estimate overtaking acceptance under specific interventions. The estimation results can be used to support active control of OLC behavior. The findings from this research provide a deeper understanding of the intricate relationships and mechanisms underlying OLC behavior. The predictive model offers a significant advancement, enabling more accurate evaluation of current systems and informing the development of more effective transportation solutions. The study’s implications are crucial for optimizing traffic flow and safety within urban expressway interchange weaving areas.