【作者】 程健维；

【导师】 杨胜强；

【作者基本信息】 中国矿业大学 ， 安全技术及工程， 2012， 博士

【摘要】 世界范围内的主要煤炭生产国家，均经历过由于煤矿开采而导致的安全事故频发，引起严重后果的阶段。当前，我国正处在煤矿生产事故多发，人员伤亡，财产损失严重这一阶段，统计表明，尽管2003年以后，我国年度煤矿事故死亡人数呈逐年下降趋势，但事故死亡人数却仍然占全世界煤矿死亡总人数的70%左右。在组成煤矿生产的各个单元系统中，矿井通风系统无疑占有极其重要的位置。这一典型的具有动态、随机性的系统必须能够保持长时间、稳定的运行状态，这对于矿井抗灾能力的提升，优化矿井综合安全生产能力，进而促进煤炭企业经济效益提高，乃至确保我国能源安全均具有重要而现实的意义。本文以煤矿通风系统生产的全过程安全为切入点，从通风系统各个不同服务阶段入手，着重从技术角度展开以下四个方面的研究：（一）针对矿井通风系统设计过程中的系统可靠性分配进行了优化研究可靠性分配则是可靠性设计中最初步的也是最为关键的一步。本文着重对可靠性分配方法在通风系统设计过程中的应用做了研究。在保证矿井通风系统总体可靠度的前提下，通过考虑技术水平、复杂程度、单元重要程度、工作环境等因素，对通风系统各单元进行人力、物力、财力进行了综合、统一、合理的决策。采用与模糊数学相结合的技术手段确定了矿井通风系统各个单元的可靠度的分配并进行了相应的优化，最后编制了相关可靠性分配软件。（二）采用“动态”随机模拟手段发展了矿井通风系统可靠性评估方法相比用传统的解析方法进行可靠性评估而言，概率性评估方法不仅能有效解决建模中存在的“维数灾”问题，更能充分考虑所要评估系统的负荷、元件故障等随机特性。本文从任务可靠性的逻辑关系角度规划了矿井通风可靠性模型，采用统计建模分析方法，归纳并拟合所要研究对象的概率分布特征及相关参数。利用Monte Carlo随机模拟方法，对矿井通风风网系统的可靠性进行评估。通过大量的重复随机试验，完成对系统进行状态抽样的过程，最后对试验结果进行了整理分析，完成可靠性评估。（三）矿井通风系统的预警机制研究构建矿井通风系统的预警机制，其目的就是在于在通风系统发生灾变之前能够预先发出警报，提示矿井管理人员及时地调整系统或实行相关对应的维护措施来减少灾害发生的风险，从而防范和制止通风系统失效发生的一种工作机制。论文利用数据挖掘理论中相关技术方法建立了矿井通风系统预警机制，主要分为指标选取与状态识别两个模块。首先应用数据挖掘理论中的粗集理论方法，利用其强大的属性简约和模式识别功能来对纷繁芜杂的、能够反映矿井通风系统运行状态的相关指标进行了分析与优选。然后采用建立于统计学习理论的支持向量机方法构建矿井通风系统运行状态判定模块。最后依据来源于现场的统计资料进行了预警预测验证，结果表明了该预警机制的可行性和实用性，这为准确昭示通风系统未来可能的风险状态及对应相关措施的采取，提供了理论支持与技术保证。（四）基于系统动力学的矿井通风系统建模及其干涉实验分析矿井通风系统的服务时间几乎等同于矿井的整个服务时间，而矿井通风系统又包含非常多且复杂的子系统。本文深入研究了这些子系统间以及子系统内子因素之间的因果影响关系，建立了基于系统动力学的矿井通风系统动力学流图模型并进行了实证研究。然后从未来系统运行的预测角度出发，设计了相关系统动力学干涉实验。从经济学角度探究了资金投入与系统安全总水平提高的关系，最后依据这些实验的结果，从不同的角度及层面提出了矿井通风系统事故防范的长效机制及建议，来保证系统长期稳定的安全运行，也为煤矿企业管理者提供了科学的决策参考依据。更多还原

【Abstract】 There are lots of historical mine accidents happened and fatality experiences inthe worldwide main coal production counties due to coal mining activities. Currently,China has been undergoing the critical stage with high frequency of mining accidents,high casualties and property damages. According to the statistics, Although thenumber of coal mine fatalities has been kept decline since the year of2003, it stilltakes up to70%of all fatalities all over the world.Among all systems to consist of the coal mine, the mine ventilation system isdefinitely an important subpart. It is a typical, dynamic system which can keep a longtime and stable running state, and is also significant for improving the anti-disasterability, optimizing the comprehensive safety production ability, promoting the miningcompany’s profit and guaranteeing our country’s energy security s strategy.Making the whole producing process safety as a starting point, based on thedifferent service sections of the mine ventilation system, the dissertation hasperformed the research works in the following four subjects:Research on the systemic reliability allocation during the mine ventilationsystem design stageThe reliability allocation is the first and also the key step in a reliability designwork. The dissertation did the research work of the reliability allocation applicationin mine ventilation system. Under the condition of ensuring the whole reliability ofthe system, the reliability allocation for each subsystem can be comprehensively,unified and reasonably determined based on the considerations including technicallevel, complicated grade, importance and working environment, etc. The fuzzymathematics technology is used in this allocation application and its optimization. Thecorresponding software is also developed in this research effort.The―dynamic‖random simulation method for the reliability assessment ofmine ventilation systemComparing with the traditional analytical assessment method, probabilityassessment method not only can solve the―Curse of dimensionality‖problem inbuilding model, but also consider the load, random failure property of the assessmentobject. Starting from the point of the logical relationship of mission, the mineventilation model is reprogrammed. With using the statistical modeling method tosummary and fit the probability properties and related parameters of the research objects, Monte Carlo random simulation method is used to assess the mine ventilationsystem. The sampling process for the system status can be done based on a largerepeated and random experiments. Then, the reliability evaluation could be finishedvia experimental results analysis.Research on the early warning mechanism of mine ventilation systemThe aim of designing the mine ventilation of early warning mechanism is toidentify the potential hazards before the failure of mine ventilation system and toremind the mine operators to take proper measures to avoid these risks. Thetechniques used in the proposed the early warning mechanism of mine ventilationsystem consisted of two models which are the index selection model and recognizingstatus model are from the data mining theory. Firstly, By using the attribute reductionalgorithm in the Rough Set, the best indexes which can reflect the behavior of mineventilation system can be selected. Then, Support Vector Machine method from thestatistical study theory is used to set up the model of status determination. Finally, theproposed mechanism has been tested by the reality cases, and the results shows thatthe predicted results agrees with the real results. This early warning can provide thetheoretical supports and technology insure with identifying the future running status.Modeling mine ventilation system by system dynamic and correspondingpolicy interferential experimentsThe life-span of mine ventilation system is almost equal to the whole servicetime of a coal mine. The mine ventilation system also includes lots of subsystems.The dissertation in-depth analyzes these subsystems and causalities between them andsetup a system dynamic model of mine ventilation system and finishes demonstrationresearches. Then, starting from the future point, a series of interferential experimentsis designed to explore the relationship between the capital investment and theimprovement of system safety level. Based on these experimental results, suggestionsto insure the long-term mine ventilation safety are proposed from different angles andlevels. In addition, the experimental results also can be used as scientific referenceevidences for mine operators.更多还原