摘要
为准确预测煤层底板的突水状况,收集18例矿井的实测数据作为学习样本,4例作为预测样本,首先运用主成分分析法对选取的含水层水压(x1)、隔水层厚度(x2)、煤层倾角(x3)、断层落差(x4)、距断层距离(x5)5个自变量进行降维处理以消除变量间的共线性,然后计算每例样本的主成分得分值并以此为中间变量建立煤层底板突水的Fisher判别模型。对18例训练样本进行回代判别,模型的综合准确率为83.3%,在对3例测试样本的判别中,误判率为0,表明该模型对于类似地质条件下底板突水危险性的判别具有一定的参考价值。
In order to predict the water inrush condition of coal seam floor,the measured data of 18 coal mines were collected as learning samples and 4 samples as prediction samples.First,using the principal component analysis method,the dimensionality reduction of 5 independent variables,such as aquifer water pressure(x 1),aquifuge thickness(x 2),coal seam dip(x 3),fault throw(x 4)and distance from fault(x 5)was carried out,so as to eliminate collinearity between variables,then,the Fisher discriminant model of water inrush from coal seam floor was established by calculating the principal component scores of each sample.The regression analysis of 18 training samples shows that the model has a certain reference value for the discrimination of water inrush risk under similar geological conditions.
作者
张兆威
滑怀田
ZHANG Zhao-wei;HUA Huai-tian(Shigetai Coal Mine,Shenhua Shendong Coal Group Co.,Ltd.,Yulin 719000,China;Shanxi Institute of Technology,Yangquan 045000,China)
出处
《陕西煤炭》
2018年第2期9-12,32,共5页
Shaanxi Coal
关键词
FISHER判别
主成分分析
底板突水
预测
Fisher discriminant
principal component analysis
water-inrush from floor
prediction
