基于最小二乘支持向量机的膨胀土判别与分类  

Distinction and Classification of Expansive Soil Based on Least Square Support Vector Machine

作　　者：李国辉[1] 闵军 陶宗涛 

机构地区：[1]河北建设勘察研究院有限公司,石家庄050031 [2]云南省水利水电科学研究所,昆明650228 [3]河南省水利科学研究院,郑州450003

出　　处：《科学技术与工程》2009年第6期1636-1639,共4页Science Technology and Engineering

摘　　要：膨胀土在我国分布广泛,对工程影响巨大,如何对其进行判别一直是岩土工程中一项重要的工作。现有的膨胀土判别与分类方法,大多仅以简单的双变量分析为依据。所选取的判别指标大多具有相同信息。针对于此,提出两种方法对膨胀土进行分类,第一种是从胀缩机理出发,采用逐步回归分析选取能够表征膨胀土的6项独立指标,再利用最小二乘支持向量机进行分类;第二种是对全部指标进行分类。结果表明最小二乘支持向量机在两种情况下都能对膨胀土进行准确的分类,也证明了最小二乘支持向量机功能的强大性。Expansive soil is widely distributed in China and has great influence on engineering, how to distinguish it is a difficult work. In the distinction and classification of expansive soil, many papers do this work by using the double-variable analysis, so the indexes used have the same character. So, two kinds of method are given to class expansive soil. The first method： based on the swell-shrinking mechanism, six indexes are found by analysis of stepwise regression, after that, least square support vector machine is used to build the classification model by using the six indexes. The second method： least square support vector machine is Used to build the classification model by using all indexes directly. And the result shows that all methods can class expansive soil correctly, which proves the powerful function of LS-LVM.

关 键 词：膨胀土 最小二乘支持向量机(LS-SVM) 分类 逐步回归 

分 类 号：TU443[建筑科学—岩土工程]