摘要
提出了基于振动传递率和主元分析相结合的新的损伤识别方法。首先,由结构无损伤时某2个位置的振动传递率样本构成结构的参考总体,以结构损伤状态时相同位置的振动传递率构成待检样本集;其次,把待检样本逐个代入到参考总体中,构成多个原始数据矩阵;然后,对所有原始数据矩阵进行主元分析并构造相应的控制椭圆和T2控制图,以前2阶主元在控制椭圆和T2控制图中的分布来确定结构是否存在损伤;最后,用一钢框架结构试验验证了本文方法的有效性,并引入振动传递率幅值的总体变化来识别结构损伤位置,损伤识别结果显示,此损伤指标能够很好的识别结构单一位置的损伤。
A new damage-identification method based on vibration transmissibility and PCA(Principal component analysis) is presented.Firstly,the vibration transmissibility between two locations of the healthy structure is served as referential samples,and the vibration transmissibility between the same two locations of the damaged structure are constructed as testing samples.Secondly,the testing samples are added into the referential samples separately,and several original data matrixes are constructed.Then,the PCA is used to extract the first two PCs(principal components) of these matrixes,and the corresponding control ellipses and T2control charts are constructed.The distributing of the first-two order PCs in the control ellipses and T2control charts are used to identify whether the structure is damaged.At last,the steel frame experiment is used to verify the validity of this method,and the index of TAC(vibration transmissibility accumulating change ratio) is introduced to identify the damage location.The results show that the index can identify single damage location exactly.
出处
《防灾减灾工程学报》
CSCD
北大核心
2012年第6期700-706,共7页
Journal of Disaster Prevention and Mitigation Engineering
基金
总后勤部专项项目(BY208C09)资助
关键词
钢框架结构
振动传递率
主元分析
控制椭圆
传感器
损伤识别
steel frame
vibration transmissibility
principal component analysis
control ellipse
sensor
damage identification