膨胀土胀缩等级分类中的距离判别分析法

Distance discriminant analysis to the classification of the grade of shrink and expansion for the expansive soils

膨胀土胀缩等级的分类是治理膨胀土工作中的首要问题。本文将距离判别分析方法应用于膨胀土胀缩等级的分类问题中,建立了膨胀土胀缩等级分类的距离判别分析模型,选用黏粒含量、粉粒含量、液限、塑限和塑性指数5项指标作为距离判别分析模型的判别因子,以膨胀土实测数据作为学习样本进行训练,建立相应线性判别函数对待判样本进行分类。研究结果表明:距离判别分析模型分类性能良好,预测精度高,交叉确认估计的误判率很低,是膨胀土胀缩等级分类的一种有效方法,可以在实际工程中进行推广。

The classification of the grade of shrink and expansion for the expansive soils was the initial and essential work for engineering construction in expansive soil area. Based on the principle of Mahalanobis distance discriminant analysis, a classification model of expansive was established in this paper, including five indexes reflecting the shrink and expansion of expansive soil： content of clay grain, content of silt grain, liquid limit, plastic limit and plasticity index,. Linear discriminant functions were obtained through training a large set of expansive samples. It was shown that the classification model of distance discriminant analysis performed excellently with high prediction accuracy and could be used in practical engineering.