首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于支持向量机的结构损伤分步识别研究
引用本文:刘龙,黄海,孟光.基于支持向量机的结构损伤分步识别研究[J].应用力学学报,2007,24(2):313-317.
作者姓名:刘龙  黄海  孟光
作者单位:上海交通大学,200240,上海
基金项目:国家自然科学基金;航天支撑技术基金
摘    要:支持向量机是一种基于统计学习理论的机器学习算法,能够较好地解决小样本的学习问题。本文介绍了支持向量机分类和回归算法,提出了基于支持向量机的结构损伤分步识别方法:以模态频率作为损伤特征,首先根据支持向量机分类算法的概率估计确定可能的损伤位置,重新构造训练样本,然后利用支持向量机回归算法计算损伤位置;最后估计损伤程度。以梁的损伤识别为例进行了验证,结果表明该方法可以提高损伤识别的精度。

关 键 词:支持向量机  模态频率  损伤识别
文章编号:1000-4939(2007)02-0313-05
收稿时间:2005-05-30
修稿时间:2006-03-20

Multi-step Structure Damage Identification Approach Based on Support Vector Machine
Liu Long,Huang Hai,Meng Guang.Multi-step Structure Damage Identification Approach Based on Support Vector Machine[J].Chinese Journal of Applied Mechanics,2007,24(2):313-317.
Authors:Liu Long  Huang Hai  Meng Guang
Abstract:The support vector classification and regression algorithms are adopted for structure damage identification, a multi-step damage identification approach based on support vector machine is proposed taking the modal frequencies as damage characteristics. The possible damage locations are detected by support vector classification according to the probability distribution. Reconstructing the training set, the accurate damage position and damage severity are identified with support vector regression. An example of rectangular beam damage identification verifies the adaptability.
Keywords:support vector machine (SVM)  modal frequency  damage monitoring
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号