School of Geosciences and Engineering, Shandong University of Science and Technology, Qingdao , China

针对传统的矿井突水水源判别存在准确率较低的问题,选取K++Na+、Ca2+、Mg2+、Cl-、SO42-及HCO3-6种水化学成分指标作为矿井突水水源判别依据,基于SPSS因子分析建模,确定21组突水水样训练样本中与第一主因子密切相关的水质指标,依据距离判别模型,将得到的水质指标作为样本数据输入Matlab平台,采用SQRT、MAHAL函数确定突水水样训练样本总体间的马氏距离矩阵。结果表明:构建二者相结合的矿井突水水源判别模型,其回判准确率高达99%,对8个未知的测试样本进行突水水源的识别,实例验证在判别指标选择合适的情况下,因子分析及距离判别相耦合的突水水源判别方式能有效地消除判别指标间的相互影响,提高判别率,为矿井水害防治提供理论基础。

According to the traditional water source discrimination of the mine water inrush existed with a low accuracy problem, the paper selected six kinds of the water chemical component index with K++Na+, Ca2+, Mg2+, Cl-, width=35,height=17,dpi=110 and width=46,height=17,dpi=110 as the water source discriminated basis. Based on the modeling with the SPSS factor analysis, by the training samples of the 21 group inrush water, the first principal factor was closely related to the water quality index. According to the distance distinguished analysis algorithm, the water quality index obtained as the sample data was inputted to the Matlab platform and the SQRT and MAHAL functions were applied to determine the Mahalanobis distance matrix between the training sample general of the water inrush water samples. The results showed that the water source discrimination model of the mine inrush water was established combined with two factors. The discrimination accuracy of the model was up to 99 %. Thus the discrimination of the mine inrush water sources was conducted on eight unknown test samples. The cases showed that under the suitable selection of the discrimination index, the discrimination method of the inrush water sources coupled with the factor analysis and distance could effectively eliminate the interactions between the discrimination indexes and could improve the discrimination rate in order to provide theoretical base to prevent and control of the mine water disaster.

国家自然科学基金资助项目(41572244)；