Bayesian sampling for practical design of multilayer microperforated panel absorbers

Abstract

Bayesian sampling is applied to produce practical designs for microperforated panel acoustic absorbers. Microperforated panels have the capability to produce acoustic absorbers with very high absorption coefficients, without the use of porous materials. However, the absorption produced by a single panel is limited to a narrow frequency range, particularly at high absorption coefficient values. To provide broadband absorption, multiple microperforated panel layers may be combined into a multilayer absorber. To design such an absorber, the necessary number of layers must be determined and four design parameters must be specified for each layer. Using Bayesian model selection and parameter estimation, this work presents a practical method for designing multilayer microperforated panel absorbers. Particular attention is paid to aspects of the underlying sampling method that enable automatic handling of design constraints such as limitations of the manufacturing process and availability of raw materials.