边坡位移非线性时间序列采用支持向量机算法的智能建模与预测研究  

Research on non-linear time sequence intelligent model construction and prediction of slope displacement by using support vector machine algorithm

作　　者：刘开云[1] 乔春生[1] 滕文彦[2] 

机构地区：[1]北京交通大学土木建筑工程学院,北京100044 [2]石家庄铁路工程职业技术学院土木工程系,河北石家庄050041

出　　处：《岩土工程学报》2004年第1期57-61,共5页Chinese Journal of Geotechnical Engineering

基　　金：国家自然科学基金资助项目(50078002)

摘　　要：介绍了人工智能领域最新的基于结构风险最小化原理的数据挖掘算法———支持向量机算法,运用Matlab语言编写了程序,采用不同的核函数对具体的边坡工程实例作了计算,并将人工神经元网络计算结果与之对比,可见无论是在学习或预测精度方面,支持向量机算法较基于经验风险最小化原理的人工神经元网络算法都有很大的优越性,可以运用于实际工程。Based on the Structural Risk Minimization principle,the latest data mining method in artificial intelligence field—support vector machine algorithm was introduced in this paper.A program was worked out in language Matlab for a slope engineering project by using different kernel function.Compared with the result obtained by using the Artificial Neural Network algorithm based on the Empirical Risk Minimization principle,the SVM algorithm is obviously superior to the ANN algorithm whatever on machine learning or prediction accuracy and it can be used to practical engineering.

关 键 词：边坡 位移 非线性 时间序列 支持向量机 回归算法 位移预测 

分 类 号：TU457[建筑科学—岩土工程]