基于多元混合模型的煤矿矿井突水水源分析

Application of Multivariate Mixed Models in Analysis of Coal Mine Water Invasion Source

为准确判定煤矿矿井突水水源,基于多元混合模型理论,在水化学分析的基础上选取含水层和突水水样化学离子,建立矿井突水的多元混合模型。分析计算3个不同煤矿的矿井突水水样,并与BP神经网络方法和模糊综合评判法这2种传统突水水源分析方法相对比。结果表明:多元混合模型理论、BP神经网络方法和模糊综合评判方法均能够结合水化学数据对矿井突水水源进行有效判别,其中多元混合模型理论不仅简洁易懂,而且准确率高于传统的BP神经网络方法和模糊综合评判法。

In order to determine exactly the source of bursting water in coal mine, multivariate mixed models were built on basis of multivariate mixed models theory and hydro-chemical analysis. The multivari- ate mixed models were applied to three selected coal mines in China. The results show that multivariate mixed models theory, BP neural network method and fuzzy comprehensive effectively to analyze mine water inrush, for three methods, the first one is in accuracy. evaluation method can be used simplest logically and the best