勘探阶段可采煤层顶底板岩体质量分类模型

A Classification Model for Rock Mass Quality of the Roof and Floor of Productive Coal Seams During Coal Geological Exploration

提出了在煤田勘探阶段利用主要物理力学性质指标和岩体质量指标进行可采煤层顶底板岩体质量分类的模型.以新疆伊犁州尼勒克县某煤矿为研究背景,选取了天然单轴抗压强度、饱和单轴抗压强度、天然抗拉强度、饱和抗拉强度、天然抗剪强度、饱和抗剪强度、天然密度、含水率、孔隙率、软化系数和RQD值等11项因子作为分类指标,以该煤矿8个可采煤层顶底板的50组岩样样品作为学习样本,首先采用因子分析和Q型聚类分析方法对样品进行综合评价分类,分类结果理想;然后分别计算每一类别样品11项分类指标的平均值并进行对比分析,结果表明,11项指标选取合理;最后采用Fisher判别分析方法对50组样品进行训练,建立了相应的Fisher判别模型,经检验正确率达到98%;将该判别模型应用到乌鲁木齐市某一煤矿对岩样进行分类,其分类结果与神经网络模型、因子和聚类联合分析的分类结果一致,有3个分类结果与规范分类出现偏差,经分析认为新建的判别模型分类结果更为真实可靠.

This paper proposed a classification model for rock mass quality of the roof and floor productive coal seams during coal geological exploration by using the main physico-mechanical properties and rock mass quality indexes. Taking one coal mine in Nilka County in Xinjiang Yili as the research background,the author selected 11 factors such as the natural uniaxial compressive strength, saturated uniaxial compressive strength, natural tensile strength,saturated tensile strength,natural shear strength,saturated shear strength,natural density,moisture content,porosity,softening coefficient and RQD as analyzing indexes to research the classification of rock mass quality of productive seams roof and floor. The 50 samples of eight productive seams roof and floor were used as study samples. Firstly,using factor analysis and Q clustering analysis method to classify 50 samples,the results showed that the best plan was the five categories. Comparing the classification results of this method and the code,there are 4classification results were inconsistent and the classification results of this method were closer to the actual situation. Then,this paper analyzed each index of all kinds of samples,the results showed the 11 classification indexes were reasonable. Finally,by using Fisher discriminant analysis method to train the 50 samples,the author established the Fisher discriminant model of classification for quality of rock mass,every sample was tested by using resubstitution method according,and the correct rate was equal to 98%. The discriminant model was applied to the geological exploration for a coal mine in Urumqi,and then classified the quality of rock mass,the results showed that the classification results entirely consistent with neural network method and clustering analysis,but there are 3 classification results were inconsistent with the code.By analyzing we could see that the classification results of code can't reflect the actual rock mass quality but the results of this discriminant model were real and reliable. This discriminant model can be widely used in the practical application of quality classification of the roof and floor of productive coal seams during coal geological exploration.