【作者】 李运强；

【导师】 程五一；

【作者基本信息】 中国地质大学（北京） ， 安全技术及工程， 2006， 硕士

【摘要】 煤炭是我国国民经济发展的基础能源。在工业生产领域,煤矿重大灾害危险源最多、安全隐患最大。煤与瓦斯突出是发生在煤矿井下的一种机理极其复杂的动力现象。煤与瓦斯突出是制约煤矿安全生产最为严重的自然灾害之一。鉴于其影响因素众多而突出机理尚未彻底研究清楚、突出灾害却日益加剧的现实状况,本文提出了根据突出机理的“综合假说”,利用人工神经网络模型实施煤层潜在突出危险性评价的指导思想。论文首先阐述了进行煤层潜在突出危险性预测评价必要性和深远意义,分析了其当前的研究现状和存在问题,确立了以突出机理的“综合假说”为理论基础、以人工神经网络模型进行评价的研究方案和技术路线。接着,本文对影响煤与瓦斯突出的瓦斯、地质条件等因素进行全面分析和总结,并抽取相应的定性、定量描述的指标。通过全面分析和总结,建立了以瓦斯、煤体结构特征和地质构造三因素为一级指标的煤层潜在突出危险性评价综合指标体系,并对反映三因素的八个指标进行了分级研究。随后,本文对人工神经网络(ANN)的基本原理、基本结构、处理能力及其应用作了简要的介绍。对煤层潜在突出危险性评价与非线性动力学模型的适应性进行了研究,指出基于人工神经网络的非线性预测评价模型在煤层潜在突出危险性评价中的应用可以较好地解决传统预测方法存在的缺陷,在煤层突出危险性评价中的应用是完全可行的。最后,本文重点研究了在函数逼近和模式识别领域广泛应用的误差反向传播神经网络模型(BPNN),对其结构设计及训练进行了深入的研究。结合沈阳煤业集团红菱煤矿煤层突出危险性预测的课题,运用 MATLAB 神经网络工具箱(ANN)建立了综合考虑煤层瓦斯含量,煤层瓦斯压力,煤的破坏类型,煤的坚固性系数,煤的瓦斯放散初速度,煤层软煤比,褶皱,断层八个指标的煤层潜在突出危险性BP 神经网络评价模型,检验结果表明,该模型能够准确的评价煤层潜在突出危险性,为煤层潜在突出危险性评价提供了一条新的技术途径。更多还原

【Abstract】 Coal is the foundation energy of our national economic development. In theindustry field, coal mines exist the most major hazards and the greatest risks. Coal andgas outburst occurred in coal mines is a dynamic phenomenon whose mechanism isextremely complicate. Coal and gas outburst is one of the most serious naturaldisasters to hamper the safety states during mining. The mechanism of the disaster isnot clear yet, however, the accidents caused by it become much more serious. Thearticle put forward a guideline, which was based on the outburst mechanism’s“comprehensive hypothesis’’ and applied artificial neural network models to evaluatethe potential outburst risk of coal seam.Firstly, this paper discussed the need and significance of evaluating the potentialoutburst risk of coal seam, analyzed the current situation and subsistent problems, andestablished research programs and technical routes, which were based on outburstmechanism’s “comprehensive hypothesis” in theory and applied the artificial neuralnetwork model for the evaluation.Secondly, the paper comprehensively analyzed and reviewed the factors whichhad impact on coal and gas outburst, such as gas, geological conditions and so on inthe round. According to the factors, the qualitative and quantitative indicators weregot. Through the comprehensive analysis and review, the paper establishedcomprehensive indices system of coal seam potential outburst risk which consisted ofthree stair indices: gas, coal body structure features and geological conformation, andclassified the eight indicators to reflect three factors.Thirdly, the basic principles, frameworks, features and capacities of the ArtificialNeural Network (ANN) were briefly introduced. The adaptability combining thepotential outburst risk evaluation of coal seam with the non-linear artificial neuralnetwork technology was discussed and ANN was believed to be a reasonable andpowerful approach to evaluate the potential outburst risk of coal seam. BackPropagation Neural Network (BPNN) was found to be the most feasible networkmodel in the assessment.Lastly, the paper emphatically and deeply studied the structure design and trainingof the BPNN which was widely applied in the function simulation and classreorganization field. On the basis of data collected from the Hongling coal mine inShenyang, Liaoning province, by applying neural network toolbox (NNT) onMATLAB software the author established BP neural network model to evaluate thepotential outburst risk of coal seam. The model considered the effect of the followingfactors: gas content, gas pressure, the destruction type of coal, the solidity modulus ofcoal, gas early speed of diffusion, soft coal ratio, fold, and fault. The test resultsindicated that the model can accurately evaluate the potential outburst risk, thereforethis paper found a new technical mean to evaluate the potential outburst risk of coalseam.更多还原