Hossain, Md. Nurtaj and Ghosh, Debraj (2022) A random process based novel training scheme for reduced order models of spatially periodic vibrating systems. In: Journal of Sound and Vibration, 529 . ISSN 0022460X
PDF
jou_sou_vib_529_2022.pdf - Published Version Restricted to Registered users only Download (2MB) | Request a copy |
Abstract
The proper orthogonal decomposition based reduced order model (ROM) is an effective tool for reducing computational cost for analysis of large time-dependent systems such as vibration, fluid dynamics, especially when repeated analyses are sought. Construction of such ROM involves recording few snapshots of the response, performing singular value decomposition of this set, and finally projecting the system to dominant singular vectors. However, the training stage of the ROM can be expensive due to multiple executions of the high-dimensional model solver. To address this issue, a novel random excitation-based training is proposed in this paper. Accordingly, depending upon the parameter range of interest, band-limited random noise excitations are chosen, and the ROM is trained from the corresponding responses. This is applied to linear and nonlinear vibrating systems with spatial periodicity and imperfection. From the numerical studies, it is found that the proposed method reduces the cost of training significantly, and successfully captures the behavior such as localization, alternate pass- and stop-bands of vibration propagation, peak response.
Item Type: | Journal Article |
---|---|
Publication: | Journal of Sound and Vibration |
Publisher: | Academic Press |
Additional Information: | The copyright of this article belongs to the Academic Press. |
Keywords: | Mode localization; Optimal training; Periodic system; Proper orthogonal decomposition; Random process; Reduced order model |
Department/Centre: | Division of Mechanical Sciences > Civil Engineering |
Date Deposited: | 21 May 2022 06:15 |
Last Modified: | 21 May 2022 06:15 |
URI: | https://eprints.iisc.ac.in/id/eprint/72324 |
Actions (login required)
View Item |