【作者】 李润求；

【导师】 施式亮；

【作者基本信息】 中南大学 ， 安全管理工程， 2013， 博士

【摘要】 摘要：煤炭开采是中国最危险行业,尤以瓦斯爆炸灾害以其破坏性强、经济损失大、人员伤亡多等显著特点对煤矿安全生产造成严重危害。论文针对瓦斯爆炸灾害预控中存在的薄弱环节,运用安全科学、系统科学、风险管理科学、信息科学、计算机科学与工程以及矿山安全等相关理论,采用理论研究、数值分析、现场观测和计算机应用相结合的方法,研究瓦斯爆炸灾害风险识别与预警体系,对实现煤矿安全生产系统风险“早期识别”和事故“事先预防”、提高煤矿本质安全化水平、促进煤矿安全生产及煤炭工业的安全可持续发展等具有重要的意义。(1)针对现有煤矿瓦斯爆炸灾害预控管理理论与方法体系尚不完善等问题,提出了从风险辨识、监控、分级、评估、预测、预警、对策等一体化视角建立瓦斯爆炸灾害风险识别与预警对策的灾害预控体系。(2)对大量瓦斯灾害事故案例进行了统计分析,获得了事故特征与统计规律及时序分形特性；针对瓦斯爆炸灾害中瓦斯涌出与积聚和引爆火源产生等核心问题,对瓦斯涌出与积聚、引爆火源、引爆地点等典型特征进行了分类研究,通过大量瓦斯爆炸事故案例分析获得了瓦斯爆炸灾害典型特征规律与耦合规律,并提出了有效减少瓦斯爆炸事故发生的有关建议和措施。(3)系统探讨了风险管理理论、技术与方法,将风险管理理论引入到煤矿瓦斯爆炸灾害事故预控中,建立了瓦斯爆炸事故演化的因果链和风险分析模型,构建了瓦斯爆炸灾害风险识别与预警的体系框架,并对风险识别与预警体系构建的内容和程序进行了系统论述。(4)针对煤矿瓦斯爆炸灾害风险源的多样性与复杂性以及监测信源与信息的多样性和复杂性,对风险信息源进行了系统研究和分类；建立了信息采集、多源信息耦合、风险评估与预警、预控对策与响应等的风险识别与预警信息平台架构。(5)针对煤矿安全生产系统具有的非线性动力学特性,对矿井瓦斯爆炸灾害风险因素进行了系统辨识并划分了逻辑单元,建立了完善的层次结构风险指标体系和风险分级体系；提出了区间线性化方法对指标信息无量纲化方法的改进；阐述了风险识别与预警关键指标“一票否决制”的独立指标变权评估与预警机制。(6)将模式识别理论、技术与方法引入到煤矿瓦斯爆炸灾害风险管理中,系统研究了瓦斯爆炸灾害风险模式识别的基本原理、内容、技术与方法,构建了瓦斯爆炸灾害风险模式识别系统；提出了单项指标分段功效函数对EMC进行改进,建立了基于IAHP-ECM的瓦斯爆炸灾害风险模式识别模型,同时还应用DTW模板匹配技术、小样本学习SVM理论和PSO算法等构建基于DTW和基于PSO-SVM的瓦斯爆炸灾害风险模式识别模型,并对上述模型进行了性能检验与应用。(7)系统研究了瓦斯爆炸灾害风险预测的基本原理和过程、预测的内容和预测技术与方法,建立了煤矿瓦斯爆炸灾害单项风险预测和综合风险预测的基础数学模型；构建了瓦斯涌出量预测的GM(1,1)模型、残差GM(1,1)模型、等维信息GM(1,1)模型、AP(p)模型、以及与SVM结合的组合预测模型等,并对上述模型分别进行了性能检验与应用。(8)系统研究了瓦斯爆炸灾害风险预警的程序与内容和风险预警等级与预警阀值设置,提出了风险因素观测值和风险水平的时间梯度是风险预警的重要组成部分；研究了瓦斯浓度超限报警机制与技术和瓦斯浓度梯度报警机制与技术,建立了基于Elman的瓦斯浓度预警模型和基于PNN的瓦斯爆炸灾害风险综合预警模型,并分别进行了模型性能检验与应用。论文针对煤矿瓦斯爆炸灾害风险预控中存在的薄弱环节,在国家自然科学基金资助项目(51274100)等资助下,运用多学科交叉理论进行了瓦斯爆炸灾害风险识别与预警体系的构建与应用研究,有较高的学术研究价值和推广应用前景。更多还原

【Abstract】 Abstract:Coal mining is the most dangerous industries in China, especially, the gas explosion disasters in coal mines, with its significant characteristics of strong destructive, big economic loss and most casualties and others, have caused serious hazard to safety production. For the weak links existing in the gas explosion disaster prevention and control through theoretical research, numerical analysis, field observation, computer application and relevant theories about safety science, system science, risk management science, information science, computer science and engineering and mine safety, gas explosion disaster risk identification and early warning system are researched in this article, which is of great importance to the implementation of early risk recognition and accident prevention in advance in the coal mine safety production system, the improvement of the coal mine intrinsically safe level, and the promotion of the sustainable development in coal mine safety.(1) In view of the imperfection of the theory and methods on coal mine gas explosion disasters pre-control management, the establishment of gas explosion hazard identification and disaster pre-control countermeasure system are put forward from the integration perspective including risk identification, monitoring, classification, evaluation, prediction, early warning, and countermeasures.(2) After statistical analysis on a large number of gas disaster accidents, the gas accident characteristics and statistical rule and fractal characteristics are obtained. As for the core problem of gas emission and accumulation, and the generation of spark origin in gas explosion disasters, the classification of typical characteristics such as gas emission and accumulation, spark origin, accident site are studied. The gas explosion characteristics and coupling rule are obtained and relevant effective suggestions and measures are put forward for decreasing gas explosion accident through a lot of gas explosion accident analysis.(3) The risk management theory, technology and methods are systematically discussed, and the risk management theory is introduced into the accident prevention and control of gas explosion disasters in coal mine. The gas explosion accident causal chain and risk analysis model are established; the gas explosion disaster risk identification and early warning system framework are built, and the content and procedure of this system are systematically discussed as well.(4) In view of the diversity and complexity of gas explosion hazard risk source and the diversity and complexity of information monitoring source in coal mine, the risk information sources are classified. The risk identification and early warning information platform architecture are established, including information gathering, multi-source information coupling, disaster risk assessment and early warning, control counter measures and response based on multi-source information coupling.(5) Based on the characteristics of non-leaner dynamics in coal mine safety production system, the risk factors of gas explosion disaster in coal mines are systematically identified, the production system are divided into some logic units, and a perfect hierarchy risk index system and risk classification system are established for risk assessment and early warning of gas explosion disaster. The interval linearization method is proposed to improve the index information of dimensionless method. The key risk identification and early-warning index "one vote veto" independent index system of the variable weight evaluation and early warning mechanism are also put forward.(6) The pattern recognition theory, method and technology are introduced into the gas explosion disaster risk management in coal mine; the fundamentals, the content, the technology and the methods of gas explosion risk pattern recognition are studied systematically; gas explosion risk pattern recognition are established; the single index piecewise efficacy function is proposed for improving efficacy function method(EMC); the risk pattern recognition model is established for gas explosion disaster based on IAHP-ECM; and at the same time, the other two risk pattern recognition models are established based on the DTW (dynamic time warping) template matching technology, and small sample learning SVM theory and PSO. Besides, all the models mentioned above are tested in application.(7) The fundamentals and process, the content and technology of risk prediction are studied systematically; the basic mathematics models of prediction are established for individual risk prediction and comprehensive risk prediction of gas explosion disaster in coal mines; the gas emission prediction models such as GM(1,1), residual error GM(1,1), equal-dimension metabolism GM(1,1), the AR(p) model and the combination forecast model by combining SVM are established. All these prediction models are anplied and pass performance test. (8) The process, the content, the grade and the threshold setting of early warning system for gas explosion in coal mines are studied systematically. It is also put forward that the observation value of risk factors and the time gradient of risk level are important parts of the early warning system. The transfinite alarm mechanism and technical of gas concentration are researched. The early warning model of gas concentration based on Elman and the comprehensive risk early warning model of gas explosion disaster based on PNN are established, and all the models mentioned above have been tested and applied respectively.In view of weak links of risk pre-control of gas explosion disaster in the coal mine, with funding from the project5127410of National Natural Science Foundation of China, the risk identification and the application of early warning system of gas explosion in coal mines are studied by using multidisciplinary theories in this research, which will be very valuable both in academic development and potential application.更多还原