基于层次聚类算法的结构损伤识别研究  

STRUCTURAL DAMAGE IDENTIFICATION BASED ON HIERARCHICAL CLUSTERING ALGORITHM

作　　者：何欣 蔡金标[1] 刘鸾翔 HE Xin;CAI Jinbiao;LIU Luanxiang(Zhejiang University,College of Civil Engineering and Architecture,Hangzhou 310058,China;Zhejiang Provincial Hongtu Transportation Construction Co.,Ltd.,Hangzhou 311308,China)

机构地区：[1]浙江大学建筑工程学院,杭州310058 [2]浙江交工宏途交通建设有限公司,杭州311308

出　　处：《低温建筑技术》2025年第3期20-25,共6页Low Temperature Architecture Technology

摘　　要：为探究层次聚类算法在结构健康检测中的应用,文中基于压电阻抗(EMI)技术,对一维钢梁和二维铝板损伤过程的阻抗变化进行监测。引入层次聚类算法进行结构损伤识别的定量研究,通过与均方根差RMSD指标法的比较,结果表明两种方法均能比较直观地判断是否存在损伤。层次聚类算法将结构的损伤分为无损伤、小损伤、中损伤、大损伤四大类的方法更为合理。层次聚类算法的欧氏距离定位方法,在概念上更加清晰,为结构健康检测提供新途径。To explore the application of hierarchical clustering algorithms in structural health monitoring,this study utilizes the electro-mechanical impedance(EMI)technique to monitor impedance variations during the damage process of a one-dimensional steel beam and a two-dimensional aluminum plate.A hierarchical clustering algorithm is introduced for quantitative structural damage identification.By comparing it with the root mean square deviation(RMSD)method,the results indicate that both methods effectively determine the presence of damage.However,the hierarchical clustering method,which categorizes structural damage into four levelsno damage,minor damage,moderate damage,and severe damageis more reasonable.Additionally,the Euclidean distance-based localization approach in hierarchical clustering provides a clearer conceptual framework,offering a novel approach for structural health monitoring.

关 键 词：EMI技术 RMSD指标法 层次聚类算法 损伤识别 阻抗谱 

分 类 号：TU311[建筑科学—结构工程]