The highest frequency to be identified imposes a lower bound on the sampling rate, and, given a sampling rate, the lowest frequency that can be identified is limited by noise and finite precision. When the bandwidth to be identified is large these two requirements imply that identification of all modes of interest in a single analysis is not feasible and some type of zooming becomes necessary. The traditional approach to realize time domain zooming is decimation. This paper reviews another alternative, presented within the framework of the Eigensystem Realization Algorithm (ERA), where zooming is realized without filtering by appropriate shifting of the two data matrices used in the algorithm. The method is known as ERA/S, where the S indicates shifting and skipping, and the skipping is due to the arrangement of the Markov Parameters (MP) in the data matrices. Since ERA/S does not filter the data, the modes outside the (shifted) Nyquist band appear aliased. The aliased modes, however, can be easily identified and removed. ©2010 Society for Experimental Mechanics Inc.
CITATION STYLE
Abramo, D., & Bernal, D. (2011). Minimizing distortions from time domain zooming. In Conference Proceedings of the Society for Experimental Mechanics Series (Vol. 3, pp. 367–374). Springer New York LLC. https://doi.org/10.1007/978-1-4419-9834-7_35
Mendeley helps you to discover research relevant for your work.