基于ELM原理的砂土液化判别模型及应用  

Sand liquefaction discriminant model and application based on ELM principle

作　　者：叶童 李治广 Ye Tong;Li Zhiguang(Hebei GEO University,Shijiazhuang 050031,China)

机构地区：[1]河北地质大学,河北石家庄050031

出　　处：《山西建筑》2022年第16期7-10,73,共5页Shanxi Architecture

基　　金：国家自然科学项目(41807231);河北地质大学教学改革研究与实践项目(2020J36);河北省级研究生示范课程(HDDYKCX2021004);河北地质大学第十七届学生科技基金科研项目(KAZ202106)。

摘　　要：为合理判断砂土的液化状态,借助极限学习机(Extreme Learning Machine,ELM)对砂土液化进行判别。共选取25个砂土液化案例为样本,其中18个案例作为学习样本,8个影响因素作为评价指标,建立了ELM砂土液化判别模型,模型回判结果全部正确;用另外7个案例开展模型验证工作,并将验证结果与规范法、反向传播法(Back Propagation,BP)、距离判别分析(Distance Discriminant Analysis,DDA)法和主成分分析(Principle Component Analysis,PCA)与距离判别分析结合法进行比较,结果表明:ELM砂土液化模型预测结果准确率高达100%,并且具有建模过程简单,分类效果好的优势。将该模型应用于工程实例,砂土液化判别结果与实际情况一致,表明该模型是一种可行的砂土液化判别方法,可在实际工程中进一步应用。In order to reasonably judge the liquefaction state of sand,the Extreme Learning Machine(ELM)was used to judge the liquefaction of sand.A total of 25 sand liquefaction cases were selected as samples,18 of which were used as learning samples,and 8 influencing factors were used as evaluation indicators.The ELM sand liquefaction discrimination model was established.Validate the work,and compare the validation results with the normative method,Back Propagation(BP),Distance Discriminant Analysis(DDA)method and Principle Component Analysis(PCA)combined with distance discriminant analysis,the results show that the prediction accuracy of ELM sand liquefaction model is as high as 100%,and it has the advantages of simple modeling process and good classification effect.The model is applied to an engineering example,and the results of sand liquefaction discrimination are consistent with the actual situation,indicating that the model is a feasible sand liquefaction discrimination method and can be further applied in practical engineering.

关 键 词：砂土液化 预测模型 极限学习机(ELM) 判别方法 

分 类 号：TU441.4[建筑科学—岩土工程]