摘要
为分析煤矿事故报告中的危险致因因素,统计分析了2018—2022年全国煤矿事故报告数据,采用Apriori关联规则算法,并利用Gephi进行关联规则可视化,探究各个致因之间的复杂关系。首先对数据进行预处理,计算词频-逆向文件频率(Term Frequency-Inverse Document Frequency, TF-IDF),提取了78个煤矿事故致因因素,其中人因层包括31个因素,设备层包括9个因素,管理层包括31个因素,环境层包括7个因素;然后,经过关联规则挖掘算法,得到了585条关联规则,绘制了其支持度、置信度和提升度的散点图;最后,根据Gephi生成的事故致因复杂网络图,分别分析了高支持度、高置信度和高提升度关联规则致因因素。结果表明:基于Apriori算法的煤矿事故致因分析,得到了人因层、管理层、设备层和环境层4个方面的关键致因因素;对煤矿关键致因因素进行直观、多视图的展现,有助于提高煤矿安全管理水平。
This paper conducts a statistical analysis of the national coal mine accident report data from 2018 to 2022,and uses the Apriori association rule algorithm to find all the rules that satisfy the minimum support and confidence by generating candidate itemsets and pruning based on the downward closure property of frequent itemsets.The paper also uses Gephi to visualize the association rules and further analyzes the high-support,high-confidence,and high-lift rules to reveal the complex relationships between various causes.The data show that from 2018 to 2022,there were 424 coal mine accidents and 914 deaths in China,and the types of accidents with a large proportion of occurrences were roof accidents,transportation accidents,and other accidents.Through data preprocessing and calculating Term Frequency-Inverse Document Frequency(TF-IDF)values,this paper extracts 80 factors causing coal mine accidents,including 33 factors in the human layer,9 factors in the equipment layer,31 factors in the management layer,and 7 factors in the environmental layer.According to the complex network graph of accident causes generated by Gephi,after preliminary analysis,the key factors for accident occurrence are weak safety awareness,on-site workers,inadequate safety education and training,transportation accidents,roof accidents,poor self-protection and mutual protection awareness,illegal operations,inadequate safety supervision and inspection,inadequate hidden danger investigation and management,etc.The paper then analyzes the factors causing high support,high confidence,and high-lift association rules separately.In terms of the human layer:the more important frequent itemsets are on-site workers and weak safety awareness,employees and weak safety awareness,no temporary support and weak safety awareness;in terms of the management layer:the more important frequent itemsets are roof accidents and poor safety management,three violations and transportation accidents,inadequate local outburst prevention measures and regional outburst prevention measures not eliminating media outburst hazards in coal uncovering areas;in terms of the equipment layer:the more important frequent itemsets are wire rope and transportation accidents,anchor cable and roof accidents,high-pressure filter quality defects and improper operation handling;in terms of the environmental layer:the more important frequent itemsets are fault and roof accidents,over-deep over-boundary and water accumulation in goaf areas,project department and other accidents.The paper reveals the key points of accident prevention and the causal relationships before and after the accident and points out the direction for the controllability prevention of accidents.
作者
景国勋
秦洪利
蒋方
JING Guoxun;QIN Hongli;JIANG Fang(School of Safety Science and Engineering,Henan Polytechnic University,Jiaozuo 454000,Henan,China;Anyang Institute of Technology,Anyang 455000,Henan,China)
出处
《安全与环境学报》
CAS
CSCD
北大核心
2024年第6期2313-2320,共8页
Journal of Safety and Environment
基金
国家自然科学基金重点项目(U1904210)
国家自然科学基金项目(51774120)
河南省科技攻关项目(232102320239)。
