【作者】 赵建；

【导师】 李贤功；

【作者基本信息】 中国矿业大学 ， 工业工程与管理（专业学位）， 2024， 硕士

【摘要】 煤矿安全问题一直以来都是我国能源安全和经济可持续发展的重要组成部分。作为我国主要的能源支柱产业,煤炭产业对经济社会的贡献是不容否定的。但是,与此同时,由于地下作业环境的特殊性,煤矿安全也面临着重大的隐患和挑战。频发的煤矿事故不仅给产业发展带来巨大损失,也给社会造成深远影响。急需从系统层面对煤矿安全风险进行精准分析,定量评估煤矿安全态势,为事故预防提供依据。本文进行的研究工作如下:(1)使用网络爬虫技术从互联网上检索我国煤矿事故案例,对煤矿事故案例按年份、事故性质和地域进行统计归纳分析。结合事故发生频率与死亡人数,采用系统聚类对事故类型进行风险等级划分。并对事故案例进行文本挖掘,识别出煤矿的潜在风险因素。基于风险因素的辨识,应用关联规则Apriori算法挖掘风险因素之间的关系,并基于关联规则相关性构建复杂网络模型。分析网络的拓扑特征,包括度、网络直径、平均路径长度、聚类系数、接近度中心性和介数中心性等统计指标,确定影响煤矿安全的关键因素,并验证煤矿风险关联网络模型的无标度特性。利用Python编程进行仿真分析网络在随机攻击和蓄意攻击情况下的结构鲁棒性和功能鲁棒性,比较不同攻击方式下的网络鲁棒性效果。采用二分网络研究风险的时间和空间耦合特性,识别高耦合的风险因素组合,并提出对策建议。(2)利用风险关联规则分析和复杂网络演化作为贝叶斯网络结构学习的参考,采用EM算法进行贝叶斯网络参数学习,得到风险节点的条件概率分布。采用交叉验证进行贝叶斯网络有效性检验。结合实际煤矿情况,利用马尔科夫模型计算设备状态的转移概率,构建煤矿风险动态贝叶斯网络模型。通过动态贝叶斯网络模型对煤矿的安全态势进行定量评估,分析煤矿安全风险后验概率,转化为安全评估指标。帮助煤矿管理人员了解当前安全状况,根据风险接受程度做出决策。最后,利用实际煤矿隐患检查记录作为证据,设置动态贝叶斯网络对应证据节点状态,分析评估煤矿安全态势。与实际煤矿安全状况对比,验证本文所提方法的有效性。更多还原

【Abstract】 Coal mine safety has always been an important part of China’s energy security and economic sustainable development.As the main energy pillar industry in our country,the contribution of coal industry to economy and society is undeniable.However,at the same time,due to the particularity of underground operating environment,coal mine safety is also faced with major hidden dangers and challenges.Frequent coal mine accidents not only bring huge losses to the development of the industry,but also have a profound impact on the society.There is an urgent need for accurate analysis of coal mine safety risks from the system level,quantitative assessment of coal mine safety situation,to provide a basis for accident prevention.The research work conducted in this thesis is as follows:(1)This study uses web crawler technology to retrieve China’s coal mine accident cases from the Internet,and makes statistical analysis of coal mine accident cases by year,accident nature and region.Combined with accident frequency and death toll,systematic clustering was used to classify the risk levels of accident types.The text mining of accident cases is carried out to identify the potential risk factors of coal mines.Based on the identification of risk factors,the Apriori algorithm is applied to mine the relationship between risk factors,and the complex network model is built based on the correlation of association rules.The topological characteristics of the network,including degree,network diameter,average path length,clustering coefficient,proximity centrality and intermediate centrality,are analyzed to determine the key factors affecting coal mine safety,and verify the scale-free characteristics of the coal mine risk association network model.Python programming was used to simulate and analyze the functional robustness and structural robustness of the network under random attacks and deliberate attacks,and the network robustness effects under different attack modes were compared.The time and space coupling characteristics of risk are studied by using binary network,the combination of highly coupled risk factors is identified,and countermeasures are put forward.(2)Using risk association rule analysis and complex network evolution as reference for Bayesian network structure learning,EM algorithm is used to learn Bayesian network parameters,and the conditional probability distribution of risk nodes is obtained.The validity of Bayesian network is tested by cross validation.Based on the actual situation of coal mine,Markov model is used to calculate the transfer probability of equipment state,and the dynamic Bayesian network model of coal mine risk is constructed.The dynamic Bayesian network model is used to quantitatively evaluate the safety situation of coal mine,analyze the posterior probability of coal mine safety risk,and transform it into the safety evaluation index.Help mine managers understand the current safety situation and make decisions based on risk acceptance.Finally,using the actual coal mine hidden danger inspection records as evidence,the dynamic Bayesian network is set up to correspond to the evidence node state to analyze and evaluate the coal mine safety situation.The effectiveness of the proposed method is verified by comparing with the actual coal mine safety situation.更多还原