Bayesian model selection for multilayer microperforated panel sound absorber design

Abstract

Microperforated panel (MPP) sound absorbers are capable of providing high sound absorption coefficients without the use of fibrous materials; however, they typically function in narrow frequency ranges. By combining multiple MPPs into a multilayer absorber, the frequency bandwidth may be increased while maintaining a high absorption coefficient. Modeling the acoustic properties of an MPP absorber requires four physical parameters per MPP layer. Since each additional MPP layer in a multilayer absorber increases the complexity of the acoustic model, Bayesian model selection is well-suited to the task of designing a multilayer MPP absorber. In such a design, minimizing the number of layers used while still satisfying the design goals is desirable, in order to optimize material usage, cost, and space required by the absorber. In a full Bayesian design framework, model selection determines the number of MPP layers required, while parameter estimation determines the (physical) design parameters for each layer. In this work, an example design scheme is specified to satisfy a practical need for acoustic absorption. The Bayesian framework produces a three-layer MPP design which meets the target requirements. This absorber design is constructed, and impedance tube measurements are obtained to validate the acoustic absorption properties.