Scholarlink

<p>A critical gap persists in quantitatively understanding the distinct causal pathways that lead to different types of mechanical injury accidents, hindering the development of targeted prevention strategies. Bridging this gap, this study presents an integrated framework that combines a modified Human Factors Analysis and Classification System (HFACS), Association Rule Mining (ARM), and Bayesian Networks (BN) to quantify how Man–Machine–Management–Environment factors interact to shape accident pathways. We analyzed 340 mechanical injury accidents investigation reports collected predominantly from Chinese manufacturing plants, with additional cases from other industrial sectors, and coded 34 3M1E factors across five HFACS levels. Frequent co-occurrence patterns extracted via the Apriori algorithm provided data-driven priors for BN structure learning, while expectation-maximization estimated conditional probabilities. Sensitivity and influence analyses uncovered injury-type-specific causal chains. The results allow safety engineers to prioritize the redesign of human–machine interfaces, targeted cognitive training, and supervisory policies according to predicted risk reductions. By fusing HFACS with probabilistic modelling, the framework demonstrates a reproducible methodology for analyzing how human, technical, and organizational elements jointly contribute to accident risks, thereby supporting evidence-based safety interventions and improved system design across industrial sectors.</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

Quantifying causal pathways of mechanical injury accidents through an integrated HFACS, ARM and bayesian network framework

  • 期刊论文

  • 发表日期：

    2025年9月12日

查看全文
  • Xin Li,
  • Leyao Wang,
  • Zongshuai Wei,
  • Ruipeng Tong

摘要

A critical gap persists in quantitatively understanding the distinct causal pathways that lead to different types of mechanical injury accidents, hindering the development of targeted prevention strategies. Bridging this gap, this study presents an integrated framework that combines a modified Human Factors Analysis and Classification System (HFACS), Association Rule Mining (ARM), and Bayesian Networks (BN) to quantify how Man–Machine–Management–Environment factors interact to shape accident pathways. We analyzed 340 mechanical injury accidents investigation reports collected predominantly from Chinese manufacturing plants, with additional cases from other industrial sectors, and coded 34 3M1E factors across five HFACS levels. Frequent co-occurrence patterns extracted via the Apriori algorithm provided data-driven priors for BN structure learning, while expectation-maximization estimated conditional probabilities. Sensitivity and influence analyses uncovered injury-type-specific causal chains. The results allow safety engineers to prioritize the redesign of human–machine interfaces, targeted cognitive training, and supervisory policies according to predicted risk reductions. By fusing HFACS with probabilistic modelling, the framework demonstrates a reproducible methodology for analyzing how human, technical, and organizational elements jointly contribute to accident risks, thereby supporting evidence-based safety interventions and improved system design across industrial sectors.