This study developed three algorithms based on vibration signal analysis for identifying and locating earthquake-induced damage in buildings. The first algorithm examines the slopes of unwrapped phases within building seismic response data. The second and third algorithms perform singular spectrum analysis (SSA) on such data and then conduct singular-value decomposition on the resultant data Hankel matrix to generate a singular-value distribution. The second algorithm achieves damage localization by analyzing the differences in the calculated cumulative singular-value distributions of each floor and a reference location; the ground floor is typically used as a reference location. The third algorithm analyzes the differences in the singular-value ratios obtained for different floors by using the first and second pairs of singular values extracted through SSA. These pairs are used because they represent the most crucial principal components in response data. These algorithms can identify damage changes in dynamic structural parameters, such as stiffness, geometric characteristics, and material properties. The performance of the proposed algorithms was evaluated by applying them to shaking table test data from a one-story, two-bay reinforced-concrete frame and a six-story steel frame for damage detection and localization. These algorithms were also applied to seismic response data of two real buildings. In addition, the proposed algorithms were validated against traditional methods based on curvature changes and the frequency response function. Overall, the results revealed that the proposed algorithms achieved high computational efficiency and sensitivity for damage detection and localization.
Loh et al. (Sun,) studied this question.