【作者】 王鹏；

【导师】 伍永平；

【作者基本信息】 西安科技大学 ， 安全技术及工程， 2010， 硕士

【摘要】 我国是一个资源大国，但也是地质灾害最严重的国家，煤与瓦斯突出就是其中的一种较普遍的矿井生产自然灾害。我国煤与瓦斯突出矿井数量多，分布范围广且突出事故频繁。尤其是随着矿井开采深度的增大，开采地质条件日益复杂，瓦斯压力不断增大，大大提高了突出的危险性。因此，致力于煤与瓦斯突出预测预报和预处理的方法研究对提高矿井的安全生产有着非常重大的现实意义。本文采用非线性理论对煤与瓦斯突出机理及危险性预测方法进行了较深入研究，通过分析煤与瓦斯突出非线性机理，以表征地应力大小的声发射指标和表征瓦斯状况的瓦斯浓度指标为影响瓦斯突出的敏感指标。并以此分别对声发射和瓦斯单项指标进行了研究，编制相应的BP神经网络程序，建立了ANNS声发射预测模型和ANNC瓦斯浓度指标预测模型。在总结和分析煤与瓦斯突出机理及影响因素的数字表征基础上，运用滚动预测法，确定了1小时声发射频数值和10分钟瓦斯浓度平均值为样本的输入参数，并探讨了网络模型结构的设计，包括输入输出层单元指标的选取、隐层单元、训练样本的确定以及输入输出数据的处理。在此基础上运用层次分析法得出了声发射1小时和8小时6个指标的权重，从而计算出声发射的综合指标值，最后结合瓦斯浓度指标得出了预测煤与瓦斯突出的综合指标CI，并给出了危险临界值。通过对现场数据进行预测分析，对比常规指标，验证了综合指标在预测工作面前方后续时间段内突出危险状况是准确且可行的。更多还原

【Abstract】 China is a country with extensive resources, but also is one of the most affected countriesof geological disasters, coal and gas outburst is one of the common natural disasters in mineproduction. Large numbers, wide distribution and the frequent accidents are the characteristicsof coal and gas outburst mine in our country, especially with the increase of mining depth, thegrowing complex geological conditions and the gas pressure increasing constantly, magnitudeof coal and gas outburst is greatly enhanced. Therefore, the research of prediction andpretreatment method of coal and gas outburst contributed to improve the safety of mine.In this paper, based on the nonlinear theory mechanism and prediction method of coaland gas outburst were further studied. By analyzing the theory, the acoustic emission indexwas used to display the ground stress value and the gas concentration index was used tocharacterize the status of gas, both of the two indices were made as the sensitive indicatorsaffecting the coal and gas outburst. And on this basis, acoustic emission and gas index werestudied, the corresponding BP neural network was programmed, ANNS AE model and ANNCprediction model of gas concentration were established.Based on the summary and analysis of gas outburst mechanism and the number land ofinfluencing factors, using rolling forecast method, determined the sample input parameterswith 1 hour acoustic emission frequency value and 10 minutes gas concentration average, anddiscussed the design of network model structure, including the selection of input or outputlayer unit index, the determination of hidden layer unit and training sample, and the process ofinput or output data. Based on these above, the weighting of six indexes including 1 hour and8 hours and so on, were obtained by the analytical hierarchy process (AHP) as well as theacoustic emission comprehensive index. Finally, comprehensive index CI ,used to predict coaland gas outburst, and the critical value of risk were got by synthetically analyzing both the acoustic emission comprehensive index and gas density index.Through analysis of field data to predict the contrast of conventional indicators, we havebeen validated for the front side indicators in forecasting the follow-up time period outburstsituation is accurate and feasible.更多还原