Multi-level damage identification of a bridge structure: a combined numerical and experimental investigation

Abstract

Damage identification of a large and complex bridge structure often requires an enormous computational effort to solve an ill-posed inverse problem with a large number of unknowns. Moreover, the predefined sensors usually cannot sufficiently cover all the potential damage regions for a large bridge structure, which will introduce additional difficulties in damage identification. To deal with these challenges, a multi-level damage identification method with response reconstruction that uses a divide-and-conquer approach is proposed. An entire bridge structure is firstly decomposed into several manageable substructures and condensed as super elements using component mode synthesis (CMS); damage identification is then carried out at the substructure level to locate potentially damaged (target) substructures. The second level is at the element (member) level to further localize and quantify damage for the target substructures. To this end, a Kalman filter-based response reconstruction is performed on the target substructure for more accurate damage quantification. To examine the feasibility and effectiveness of the proposed method, a combined numerical and experimental investigation is performed on a laboratory testbed model of the Tsing Ma suspension bridge (TMB). Numerical studies are firstly conducted to inform optimal sensor placement for response reconstruction and multi-level damage identification. The sensor system is then installed on the TMB testbed model, and the proposed multi-level damage identification method is validated through comparison with experimental results. The numerical and experimental results demonstrate that the proposed multi-level damage identification method is capable of identifying damage in a large bridge structure.