基于主成分分析与Fisher判别分析法的矿井突水水源识别方法

为有效地预防矿井突水事故,及早识别突水水源是关键工作之一。根据矿井各含水层水化学成分的差异性,选取7种水化学成分指标作为突水水源识别的样本变量。在此基础上,采用主成分分析(PCA)与Fisher判别分析相结合的方法建立突水水源判别模型。以新庄孜煤矿不同水层的水化学特征资料中的33个为学习样本,12个为预测样本,对该模型进行检验和应用,并与传统Fisher判别分析模型的结果进行比较。研究结果表明:利用PCA与Fisher突水水源判别模型能够有效地消除样本变量指标间的相互影响,使突水水源判别结果更加准确。

Recognizing of Mine Water Inrush Sources Based on Principal Components Analysis and Fisher Discrimination Analysis Method

It was held here that the early recognition of water bursting source was the key to water inrush prevention.Mass concentrations of seven water chemical components(C2+a,Mg2+,K++N+a,HCO-3,SO2-4,Cl-and the total hardness) were selected as the sample variables in water bursting source recognition,A prediction model of water inrush source was built by combining PCA with Fisher discriminant analysis.The model was tested and applicated in the different water layers of Xinzhuangzi coal mine with 33 training samples and 12 forecasting samples,and compared with the traditional Fisher discrimination model.The results show that prediction made by this model is more accurate than that by the traditional one.