Structural Damage Detection based on Principal Component Analysis and Hotelling T Control Chart
A structural damage detection approach with multivariate statistical principal component analysis (PCA) and Hotelling T2 control chart on mode shapes is proposed and validated with numerical simulation on a 4-story frame structure. The modal shapes of the frame structure in healthy and damaged status are determined by finite element analysis. The PCA on the first two modal shapes is carried out and results show the first three-order principal components (PC) represent the main feature of the modal shapes, and three score plots are obtained by projecting the PC score to three coordinate planes. The results show the PC scores of damage element nodes deviate from them in healthy status obviously. Then, the Hotelling T2 statistics analysis is carried out. The T2 statistics of damaged element nodes are greater than them in healthy status, and exceed the upper control limits (UCL) of T2 control chart with certain confidence level. Structure damage can be identified by separating the data in the t2 control charts. The performance of the proposed approach for damage identification of the frame structures with different single-damage and multi-damage scenarios are studied. Results show that the proposed method is robust and effective.
damage identification modal shape principal component analysis score chart T2 control chart upper control limits
Rong-rong Jin Bin Xu
School of Civil Engineering, Hunan University, Changsha410082, Hunan, China School of Civil Engineering, Hunan University, Changsha410082, Hunan, China Ministry of Education Ke
国际会议
2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)
上海
英文
257-262
2010-12-25（万方平台首次上网日期，不代表论文的发表时间）