| 引用本文: | 朱雷,孙世钧,阚绍德,周广春.预测砌体墙板开裂模式的支持向量机方法[J].哈尔滨工业大学学报,2013,45(6):22.DOI:10.11918/j.issn.0367-6234.2013.06.004 |
| ZHU Lei,SUN Shijun,KAN Shaode,ZHOU Guangchun.A SVM model for predicting cracking pattern of masonry wall panel[J].Journal of Harbin Institute of Technology,2013,45(6):22.DOI:10.11918/j.issn.0367-6234.2013.06.004 |
|
| 摘要: |
| 为探索一种简单易行、精度良好的砌体墙板开裂模式预测方法,建立了一个预测面外均布荷载下砌体墙板开裂模式的支持向量机方法: 对砌体墙板试验开裂模式进行特征提取,获取开裂模式先验知识; 对提取的开裂模式特征进行数值化定义,得到描述开裂模式的特征值;应用支持向量机学习这些试验墙板开裂模式的特征值; 应用学习/训练后的支持向量机模型,对新墙板进行开裂模式的预测．对该支持向量机方法的验证,分别考察了3种情况:三边支撑墙板、四边支撑墙板、混合两种支撑的墙板．结果表明所建立的支持向量机方法,能够基于试验墙板开裂模式,较精确地预测新墙板开裂模式． |
| 关键词: 支持向量机 砌体墙板 开裂模式 预测 特征值 |
| DOI：10.11918/j.issn.0367-6234.2013.06.004 |
| 分类号: |
| 基金项目:辽宁省(高校)重点实验室开放基金(YT-200903). |
|
| A SVM model for predicting cracking pattern of masonry wall panel |
|
ZHU Lei1, SUN Shijun2, KAN Shaode3, ZHOU Guangchun1
|
|
(1.School of Civil Engineering, Harbin Institute of Technology, 150090 Harbin, China; 2.School of Architecture, Harbin Institute of Technology, 150006 Harbin,China; 3.Guangzhou Metro Design ＆ Research Institute Co., Ltd.,510010 Guangzhou,China)
|
| Abstract: |
| To predict the cracking pattern of masonry under uniformly and laterally load more simply and more accurately, a support vector machine (SVM) method was developed. Firstly, the feature of the cracking patterns of tested wall panels was extracted, and the priority knowledge of these panels was obtained. Secondly, the feature of these panels to form the characteristic data was quantitatively described and the SVM model to study the characteristic data was built. Finally, the cracking patterns of new panels were predicted using the SVM model above. To testify the SVM model, the panels with three simply-supported edges (hereinafter referred to as “the trilateral support”) and a top free edge, the panels with four simply-supported edges(hereinafter referred to as “support on four sides”), and all the panels above were studied. The result indicated that the developed SVM method could predict the cracking patterns of masonry wall panels in a better precision than traditional methods. |
| Key words: support vector machine (SVM) masonry wall panel cracking pattern predict characteristic data
|
