Output-only substructural identification for local damage detection

Abstract

To ensure safety and to reduce maintenance cost of infrastructures, the development of structural health monitoring methods for damage identification has received considerable attention recently. For large and complex structural systems such as long-span bridges and high-rise buildings, not only a large number of sensors are needed but also formidable computational challenges are faced in order to achieve reliable results for damage identification. Furthermore, damage detection becomes even more difficult if the input force measurement is not possible. To this end, an output-only substructural identification is proposed to detect and monitor some critical locations within a large structure based on acceleration response only. Structural damage is detected by identifying changes in stiffness parameters in the substructure of concern. A soft computing approach of genetic algorithm is adopted to identify the stiffness and damping parameters of the substructure of concern. A key advantage of the proposed strategy is that it can identify structural parameters as well as input excitation force within the substructure, based on acceleration measurement only. This is done ingeniously by adopting a predictor-corrector algorithm to correct the output responses of internal acceleration, velocity, and displacement that are predicted using numerical integration. Numerical and experimental studies are presented for a continuous truss bridge and a laboratory frame model, respectively. Numerical and experimental results demonstrate the good performance of the output-only substructural strategy for local damage detection using acceleration measurements only. © 2010 Taylor & Francis Group, London.