Affiliation:
1. Department of Electronics, Information and Bioengineering, Politecnico di Milano , Milan, Italy
Abstract
The acoustical properties of wood are primarily a function of its elastic properties. Numerical and analytical methods for wood material characterization are available, although they are either computationally demanding or not always valid. Therefore, an affordable and practical method with sufficient accuracy is missing. In this article, we present a neural network-based method to estimate the elastic properties of spruce thin plates. The method works by encoding information of both the eigenfrequencies and eigenmodes of the system and using a neural network to find the best possible material parameters that reproduce the frequency response function. Our results show that data-driven techniques can speed up classic finite element model updating by several orders of magnitude and work as a proof of concept for a general neural network-based tool for the workshop.
Publisher
Acoustical Society of America (ASA)
Subject
Acoustics and Ultrasonics,Arts and Humanities (miscellaneous)
Reference45 articles.
1. Feed-forward neural networks for failure mechanics problems;Appl. Sci.,2021
2. Characterization of material parameters of composite plate specimens using optimization and experimental vibrational data;Compos. B Eng.,1996
3. On the prediction of the frequency response of a wooden plate from its mechanical parameters,2022
4. Violin makers can't pick out good wood;Nature,2007
5. Materials selection for musical instruments;Proceedings of the Institute of Acoustics UK,1997