Bayesian finite-element model updating of passively controlled building structures
Yang Pengchao1,2 Xue Songtao2,3 Xie Liyu2
1. Tongji University, Shanghai 200092, China; 2. China Construction Eight Engineering Division Co., Ltd., Shanghai 200112, China;
3. Tohoku Institute of Technology, Sendai 982-8577, Japan
Abstract：The structural measurements of actual passively controlled structures are often incomplete. The process of updating the model of such structures is typically ill-conditioned and can be computationally challenging. This study investigated the process of updating the model of a building equipped with oil dampers. To address the ill-conditioned issue, an effective system-level strategy for model correction is presented, and a Bayesian inference approach is proposed to update the model. First, a simple, shear-type supplementary model is superimposed onto the initial finite element (FE) model of the building. This supplementary model belongs to the group of system-level model correction. Subsequently, modal sensitivity analysis is carried out to quantify the importance of the updating parameters of this supplementary model. These parameters are grouped and reduced into a few important parameters, which are eventually used in the model update. Then, a Bayesian inference approach is employed to estimate the posterior distribution of the updating parameters using earthquake measurements, and the identifiability of the parameters and the effectiveness of the proposed correction strategy are verified. Finally, a set of the earthquake measurements of the investigated building is selected to demonstrate the effectiveness of the proposed approach for updating the model.
