基于组态分析与机器学习的煤矿运输事故致因研究  

Research on the causes of coal mine transportation accidents based on configurational analysis and machine learning

作　　者：钱敏[1] 王哲 成连华 QIAN Min;WANG Zhe;CHENG Lianhua(College of Management,Xi’an University of Science and Technology,Xi’an 710054,China;College of Safety Science and Engineering,Xi’an University of Architecture and Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学管理学院,陕西西安710054 [2]西安科技大学安全科学与工程学院,陕西西安710054

出　　处：《西安科技大学学报》2025年第2期286-295,共10页Journal of Xi’an University of Science and Technology

基　　金：国家自然科学基金项目(51974238);陕西省自然科学基础研究计划项目(2021JM-397)。

摘　　要：为应对煤矿运输事故频发的严峻形势,防止事故进一步恶化,以68起煤矿运输事故为研究对象,基于改进的HFACS理论构建煤矿运输事故致因模型与属性表,并在此基础上运用清晰集定性比较分析方法与随机森林方法对事故致因进行综合分析,构建了一个结合组态分析和机器学习的煤矿运输事故综合分析框架。利用沙普利值(SHAP)揭示了阻断事故发生的关键因素,并验证了组态分析的稳健性。结果表明:组态分析识别出12种煤矿运输事故组态解,聚类归纳出技术缺陷、监管不力和安全教育缺失3组高阶构型;随机森林模型总体准确率达到92.9%,尤其在预测事故发生(类别1)时显示出高精确度和召回率;模型的SHAP值散点图显示,技术环境、监管不力、组织氛围是导致事故发生的核心条件,进一步验证了组态分析的稳健性。根据研究结果,针对诱发煤矿运输事故的高阶构型路径提出预防与应对措施。To address the frequent occurrence of coal mine transportation accidents and prevent further deterioration,this paper studied 68 coal mine transportation accidents.It constructed a causation model and attribute table for these accidents based on an improved HFACS theory.Using crisp-set qualitative comparative analysis and random forest methods,the paper performed a comprehensive analysis of the causes of the accidents,establishing an integrated analysis framework combining configurational analysis and machine learning.Finally,the key factors preventing accident occurrence were revealed using SHAP values,and the robustness of the configurational analysis was validated.The results indicate that:The configurational analysis identifies 12 configurations of coal mine transportation accidents,clustering into such three high-level configurations as technical defects,inadequate supervision,and lack of safety education.The random forest model achieves an overall accuracy of 92.9%,with particularly high precision and recall rate in predicting accident occurrence(category 1).The SHAP value scatter plot of the model shows that technical environment,inadequate supervision,and organizational atmosphere are core conditions leading to accidents,further validating the robustness of the configurational analysis.Based on the study results,preventive and responsive measures for the high-level configuration paths inducing coal mine transportation accidents are proposed.

关 键 词：清晰集定性比较分析 组态致因模型 机器学习 煤矿运输事故 

分 类 号：TD91[矿业工程—选矿] X928[环境科学与工程—安全科学]