Scholarlink

<p>With increasing automation and operational complexity in recent railway systems, understanding dispatcher human errors has become essential for ensuring traffic safety. This study examined dispatcher human error characteristics and contributing factors in high-speed rail (HSR) and conventional rail (CR) dispatchers using 17,036 proactive safety inspection records. Human error co-occurrence networks for HSR and CR dispatchers were constructed via network analysis within the Human Factor Analysis and Classification System (HFACS) framework. Central error nodes and structural differences were identified through bridge analysis and network comparison test (NCT). The multinomial logit (MNL) model was employed to assess how selected factors associated with the likelihood of different high-frequency error types. Skill-based errors emerged as the most strongly connected nodes within and across HFACS levels in both networks. The NCT results showed significant differences between the HSR and CR networks in global structure, node centrality, and edge weights. In the CR network, inadequate supervision exhibited stronger co-occurrence with skill-based errors, decision errors, and crew resource management. The MNL results indicated that workload (jurisdictional station counts), task type, season, and shift were significantly associated with the likelihood of different error types (<i>χ</i><sup>2</sup> = 6769.01, <i>p</i> &lt; 0.001, Nagelkerke <i>R</i><sup>2</sup> = 0.688). These findings may help inform dispatcher risk identification, task-specific interventions, and systemic safety management.</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

Integrating network analysis and multinomial logistic regression for human error characteristics and influencing factors among train dispatchers

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Huishan Pang,
  • Ying Ge,
  • Zizheng Guo

摘要

With increasing automation and operational complexity in recent railway systems, understanding dispatcher human errors has become essential for ensuring traffic safety. This study examined dispatcher human error characteristics and contributing factors in high-speed rail (HSR) and conventional rail (CR) dispatchers using 17,036 proactive safety inspection records. Human error co-occurrence networks for HSR and CR dispatchers were constructed via network analysis within the Human Factor Analysis and Classification System (HFACS) framework. Central error nodes and structural differences were identified through bridge analysis and network comparison test (NCT). The multinomial logit (MNL) model was employed to assess how selected factors associated with the likelihood of different high-frequency error types. Skill-based errors emerged as the most strongly connected nodes within and across HFACS levels in both networks. The NCT results showed significant differences between the HSR and CR networks in global structure, node centrality, and edge weights. In the CR network, inadequate supervision exhibited stronger co-occurrence with skill-based errors, decision errors, and crew resource management. The MNL results indicated that workload (jurisdictional station counts), task type, season, and shift were significantly associated with the likelihood of different error types (χ2 = 6769.01, p < 0.001, Nagelkerke R2 = 0.688). These findings may help inform dispatcher risk identification, task-specific interventions, and systemic safety management.