The Canadian railway industry has improved safety performance in the last decade as measured by freight loss incidents per billion gross ton-miles. Further improvements in safety performance require a deeper analysis of the leading causes to identify weaknesses in implementing safety systems. In this paper, we classify the causes of railway loss incidents using a Safety Management System (SMS) framework to identify system weaknesses. The role of human factors is further analyzed through the Human Factors Analysis and Classification System (HFACS) approach. For this, we utilized data from 42 main track derailments and collisions involving the transport of dangerous goods in Canada between 2007 and 2018, which have been investigated by the Transportation Safety Board of Canada in detail. Associations between adjacent sub-categories of the HFACS framework are analyzed to identify any interdependency that exists between active and latent errors using a Chi-square test and Kruskal's lambda analysis. Furthermore, we implement the Decision-Making Trial and Evaluation Laboratory (DEMATEL) method and the Analytical Network Process (ANP) to identify causal relationships between different sub-categories of the HFACS framework and calculate the weighted influence of each sub-category on main track derailments and collisions. Finally, a comparison is made between this work and others', which have analyzed human factors in the railway industry. There is good agreement between the results of these studies that highlight the importance of supervisory and organizational factors in the prevention of railway loss incidents. Based on these findings, we make recommendations to reduce railway loss incidents.

以每十亿总吨英里的货运损失事件来衡量，加拿大铁路业在过去十年中提高了安全绩效。安全绩效的进一步改进需要对主要原因进行更深入的分析，以确定实施安全系统的弱点。在本文中，我们使用安全管理系统 (SMS) 框架对铁路损失事故的原因进行分类，以识别系统弱点。通过人因分析和分类系统 (HFACS) 方法进一步分析了人为因素的作用。为此，我们利用了 2007 年至 2018 年加拿大运输安全委员会详细调查的 42 起涉及危险货物运输的主要轨道脱轨和碰撞的数据。分析 HFACS 框架的相邻子类别之间的关联，以使用卡方检验和 Kruskal 的 lambda 分析识别主动错误和潜在错误之间存在的任何相互依赖性。此外，我们实施决策试验和评估实验室（DEMATEL）方法和分析网络过程（ANP）来识别HFACS框架不同子类别之间的因果关系，并计算每个子类别对主轨道的加权影响脱轨和碰撞。最后，将这项工作与其他人的工作进行了比较，这些工作分析了铁路行业的人为因素。这些研究的结果非常一致，突出了监督和组织因素在预防铁路损失事故中的重要性。