Stochastic multiscale simulation of porous sound absorbing material based on adaptive bayesian quadrature

• Published in: The Journal of the Acoustical Society of America 2022-10, Vol.152 (4), p.A186-A186
• Main Authors: Komatsu, Yosuke, Yamamoto, Takashi
• Format: Article
• Language: English
• ISSN: 0001-4966; 1520-8524
• DOI: 10.1121/10.0015977

This presentation proposes a method to analyze porous sound absorbing materials by multiscale simulation with the adaptive Bayesian quadrature method. The homogenization method calculates equivalent physical acoustic properties from the microstructure of a porous material. This method is powerful in designing new porous materials because it can predict acoustic properties from arbitrary microstructures. On the other hand, it is difficult to consider the randomness of actual materials because of the assumption of periodic microstructures. We propose a method for quantifying uncertainty in acoustic properties by incorporating a Bayesian perspective into Gaussian process regression. Assuming the microscopic structure of porous material as random variables, the response obtained by the homogenization method is approximated by Gaussian process regression. Bayesian quadrature is used to calculate the statistical moments of the integral of the characteristic functions.An additional integration point that minimizes the variance of the integral is adaptively selected, and further responses are obtained using the homogenization method. Repeating this process shows that the integral of characteristic functions can be calculated accurately and efficiently, and their uncertainties can be quantitatively evaluated.