摘要
将Bayes判别分析方法应用于矿井突水水源分析问题研究中.选用多项化学指标作为判别因子,建立了适用于不同水质类型矿井的两类和多类水源分析的Bayes判别分析模型.基于Bayes判别分析方法的原理,利用不同矿区突水水源的典型样本,对Bayes判别方法的判别过程和检验方法进行了具体说明.针对两类水源,选用Ca2+、Mg2+、Na+、K+、Cl-、HCO3-、SO42-、NO3-、F-和pH等10项指标作为判别因子.对多类水源,选用Na++K+、Ca2+、Mg2+、Cl-、SO42-、HCO3-等6种离子组合作为判别因子,利用华北某矿区10例典型样本和焦作矿区39例典型样本作为数据源,分别建立相应的判别模型.判别结果完全符合实际情况,并与数量化理论、支持向量机方法进行了比较.研究结果表明,Bayes判别方法的计算过程简单、模型结构稳定,回代估计判别结果以及预测结果的准确性很高,对突水水源的判别具有很强的预测能力,可以在实际工程中进行应用.
Bayes discriminant analysis （BDA） method was used in the study of headstreams of prediction of mine water inrush, and two BDA models-for recognizing two-headstreams and muhi-headstreams were constructed. Based on the principle of BDA theory and the classical headstream samples of different mines, the discriminant process and cross-validation method were introduced. 10 samples from a mine in North China and 39 samples of Jiaozuo Mine were used as data sources. Ca^2＋、Mg^2＋、Na^＋、K^2＋、Cl^-、HCO3^-、SO4^2-、NO3^-、F^- and pH were selected as discriminant genes for two-headstreams BDA model, and Na^＋ ＋K^＋、Ca^2＋、Mg^2＋、Cl^-、SO4^2-、HCO3^- were regarded as discriminant genes for multi-headstreams BDA model. Compared with the results of SQT method, ANN method and SVM method, the results showed that the frame of BDA model was steady and high prediction accuracy. Thus, BDA method can be used in practical mine engineering.
出处
《矿业工程研究》
2009年第3期27-30,共4页
Mineral Engineering Research
关键词
矿井
突水
水源判别
Bayes判别分析
预测
mine
waterburst
headstream prediction
Bayes discriminant
forecasting
