距离判别法在矿山防治水中应用

以淮南潘二矿区、山西河池矿区和河南焦作矿区水样作为突水水源数据，采用距离判别分析理论，对突水水源进行判别分析。选用六大常规离子作为判别因子，分别建立矿井突水水源的距离判别分析模型。经回判检验表明，潘二水样和焦作水样的距离判别模型回判准确率超过90％，而河池矿区水样距离判别模型的回判准确率仅50％。为此，增加总硬度、碱度、PH值和矿化度作为判别因子，重新建立河池矿区水样的距离判别分析模型，回判准确率提高至90％，证明适当增加特征判别因子对改善距离判别分析模型的判别准确率有利。最后对三个矿区的未知样本进行了距离判别分析，并与Bayes逐步判别法和模糊综合评判法判别结果对比，结果表明距离判别法稳定性较好，判别准确率与Bayes逐步判别基本相同，比模糊综合评判要好。因此，在判别因子选择合适的情况下，距离判别法是目前矿井突水水源判别的有效方法。

Application of Distance-based Discriminant Method on Prevention of Water-inrush of Mine

Based on Mahalanobis Distance Discriminant Analysis theory , the study establishes a distance -based discriminant analysis model, according to which location of water -inrush sources in Huainan Pan＆#39;er Digging, Shanxi Hechi Digging and Henan Jiaozuo Digging are identified by analyzing data of sample water collected from the three diggings .Results show that when discrimi-nant factors of the model are the six regular ions the accuracy of discriminating results drew by water sample of Pan ＆#39;er Digging and Jiaozuo Digging is over 90 percent, while that of discriminating result drew by water sample of Hechi Digging is as low as 50 per-cent.Then, the study establishes another model by adding total hardness , alkalinity, pH and salinity to the list of discriminant fac-tors.The accuracy of the new model is proved to be 90 percent.The increase indicates that appropriate discriminant factors make contribute to a high accuracy of discriminant analysis model .Then, the comparison between discriminating results generated by dis-tance-based discriminant analysis model and fuzzy comprehensive evaluation model is taken .The result shows that two models share the same accuracy while the former one is more stable than the latter one .Accordingly , taking appropriate index as discrimi-nant factors, distance-based discriminant method is a sound and effective method to identify sources of water -inrush.