堤防管涌发生可能性识别的网格搜索-支持向量机方法

来源期刊:中南大学学报(自然科学版)2015年第4期

论文作者:翟越 刘浪 于澍

文章页码:1497 - 1504

关键词:堤防工程;管涌;支持向量机;网格搜索法;预测

Key words:embankment engineering; piping; support vector machine; grid-search method; forecast

摘    要:针对堤防管涌的评价涉及多种变量且各变量之间存在着高度的非线性关系,应用统计学习理论并结合工程实际,提出基于支持向量机(SVM)理论的堤防管涌发生可能性识别方法。将影响管涌演化、发生和导致破坏的因素进行归纳,选取坝高H、坝前水深Hp、下游边坡m、土的有效凝聚力c、有效内摩擦角ψ、饱和单位容重γ、渗透系数K、最大有效粒径疏db和下游滤层倾角δ共9种代表性的优势参数作为模型的输入,将堤坝管涌发生的可能性因子λ作为模型的输出,以16个堤防管涌工程实例作为学习样本进行训练,采用RBF核函数,建立堤防管涌发生可能性识别的支持向量机分类模型。为提高预测模型的泛化能力和预测精度,利用网格搜索寻优方法对支持向量机模型的参数进行了优化,并对7组待判实例进行判别。研究结果表明:建立的网格搜索优化支持向量机分类模型对堤防管涌发生可能性识别结果与实际结果吻合,正确率达95%,可考虑在实际工程中进行推广。

Abstract: As the piping evaluation involved a high degree of non-linear relationship between a variety of variables, a new method of the support vector machine (SVM) to predict the piping occurring probability in embankment was proposed on the basis of the statistical learning theory and the actual characteristics of the project. Comprehensive consideration of the evolution, occurrence and the piping factors lead to failure, nine major representative parameters of piping evaluation, i.e. the dam height H, the front of the dam water depth Hp, downstream slope m, the soil cohesion c, the effective internal friction angle ψ, saturated unit weight γ, the permeability coefficient K, the maximum effective diameter of sparse db and downstream of the filter layer angle δ are taken into account input variables for the proposed model, and the likelihood of piping in embankment factor λ is selected as output value for the proposed model. 16 typical cases of piping in embankment are used for training data by introducing radial basis function (RBF) kernel function. To enhance the generalization performance and prediction accuracy, grid-search method (GSM) was used to search for suitable values of parameters of the predicting model in the current study, thus the piping occurring probability in embankment prediction of GSM-SVM classification model was established, and 7 other group cases were sentenced to distinguish samples for further study of the effectiveness and practicality of the proposed model. The results show that the establishment of SVM classification model prediction of the piping occurring probability in embankment can achieve a high accuracy, and are coincided with the actual results, the correct rate is 95%, which provides a new approach to evaluation of the piping occurring probability in embankment and can be applied in practical engineering.

相关论文

  • 暂无!

相关知识点

  • 暂无!

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号