Frequency-domain damage detection of seismically-excited buildings
Journal
9th European Workshop on Structural Health Monitoring, EWSHM 2018
Date Issued
2018
Author(s)
Abstract
Earthquakes strike or devastate urban cities around the world every year. Structural health monitoring serves as a gateway that enables information of a structure to be acquired through sensing technology, and is of need to early diagnose structural integrity. Many studies have successfully exploited time series analysis to extract features from measured signals and then to identify damage of a structure. However, the time-variant nature may result in difficulties to interpret the state of a structure. In this study, an innovative damage diagnosis method is proposed by utilizing the orthogonal projection of frequency-domain Hankel matrices. For an instrumented building, the measured responses are converted into frequency-domain Hankel matrices via the short-time Fourier transform. Then, an instant subspace is obtained by the orthogonal projection of one time-variant Hankel matrix to two Hankel matrices in the beginning and end. Similarly, another type of subspaces can be calculated over time by projecting a time-variant Hankel matrix onto two Hankel matrices at the delayed and advanced steps. These two subspaces are applied with the singular value decomposition in order to obtain both right and left singular vectors, which represents the spectral magnitude envelopes of the delayed and advanced projections, respectively. By comparing these two singular vectors, the damage occurrence and locations of buildings can be determined. Moreover, the deviation in this comparison also implies the damage intensity. The proposed method is validated by a seismically-excited laboratory-scale twin-tower building. The results exhibit that the proposed method effectively identifies the damage occurrence and locations. ? 2018 NDT.net. All rights reserved.
Subjects
Damage detection; Frequency domain analysis; Singular value decomposition; Technology transfer; Time series analysis; Vectors; Damage diagnosis; Frequency domains; Instrumented buildings; Measured signals; Orthogonal projection; Sensing technology; Short time Fourier transforms; Spectral magnitudes; Structural health monitoring
SDGs
Type
conference paper