东南大学学报（自然科学版）2025，Vol.55Issue(1)：51-58,8.DOI:10.3969/j.issn.1001-0505.2025.01.006
基于高斯过程的高空间分辨率振型识别
Gaussian process-based high-resolution structural mode shape identification
摘要
Abstract
To address the challenge of identifying full-field mode shapes of a structure with limited sensors,a Bayesian identification approach based on the Gaussian process(GP)prior is proposed.Based on the principle of Bayesian inference,it effectively fuses finite element analysis(FEA)and modal identification data.Firstly,considering physical constraints,a GP prior of full-field mode shapes is constructed through probabilis-tic FEA or engineering experience.Subsequently,a posterior distribution of full-field mode shapes is estab-lished by introducing the measured vibration data and integrating the results of Bayesian modal identification on the basis of the GP prior.Finally,the efficiency of the proposed approach is validated by identifying full-field mode shapes of an experimental lab shear model and the benchmark model of Canton Tower.The re-sults demonstrate that the proposed approach can effectively infer high-spatial-resolution mode shapes of the structure with limited sensors,and quantify the associated identification uncertainty,which can be effectively applied to structural health monitoring.关键词
结构健康监测/高斯过程/振型扩展/不确定性量化/物理约束Key words
structural health monitoring/Gaussian process/mode shape expansion/uncertainty quantifica-tion/physical constraints分类
建筑与水利引用本文复制引用
李宾宾,叶挺,兰春光..基于高斯过程的高空间分辨率振型识别[J].东南大学学报（自然科学版）,2025,55(1):51-58,8.基金项目
国家自然科学基金资助项目(51908494). （51908494）
