Modal Property Extraction Based on Frequency Domain Stochastic Subspace Identification

Abstract

Structural integrity can be investigated through observing modal properties (e.g., natural frequencies, mode shapes, and damping ratios) determined by system identification methods. To understand the dynamic behavior of a structure over time, automated modal tracking techniques are developed to reduce human-computer interaction. Moreover, modal tracking using earthquake responses is essential since the magnitudes of ambient vibrations are less effective for structural responses as well as for identification. Therefore, this study is focused on performance evaluation of the automated frequency-domain stochastic subspace identification (SSI) under both ambient and seismic excitations. The method first divides the measurements into sequential portions and then employs the refined frequency domain decomposition (rFDD). A peak-picking method is subsequently applied to extracting modal candidates with respect to natural frequencies in accordance with the rFDD results. The sequential portions of measurements are also used to construct the accumulated frequency-domain Hankel matrix, and the Hankel matrix is utilized for the frequency-domain SSI. Finally, the modal properties of the stable structural modes are selected from quick stabilization diagrams. In the numerical example, the proposed method is applied to a constant-stiffness structure with 500 sets of ambient excitation events in order to examine the stability and accuracy. Moreover, the proposed method is investigated for a continuously degraded structure using 500 sets of ambient excitation events of which the structure encounters an aging problem. Additionally, seismic excitation events are also examined using the proposed method. As a result, the proposed method is capable of extracting modal properties of structures under different types of excitations with high accuracy and shows consistency in modal tracking.