Specific acoustic impedance estimation; Bayesian-network-based inverse finite difference frequency domain method

Abstract

Design for acoustic performance for an interior fluid domain requires accurate description of boundary materials' specific acoustic impedance. The standard approach for the estimation of this material characteristic is the two-microphone, impedance-tube method. Modifications to the processing of the sampled acoustic field have been proposed to allow for more general test geometries. While analytical methods may be applied to a small class of ideal geometries, numerical methods provide greater geometric flexibility. In general, solutions to the wave equation forward problem are found from boundary element, finite element, or finite difference methods. The inverse problem of parameter estimation is solved by evaluating accuracy of prediction of the acoustic field for given distributions of the specific acoustic impedance parameter against observed data. In this presentation a Bayesian-network sampling approach is used to estimate specific acoustic impedance of a micro-perforated panel in an impedance tube test geometry. The choice of geometry and material allow for direct comparison to the two-microphone, impedance-tube method within the appropriate frequency range, and a theoretical model for the material beyond that frequency range. The potential to extend the frequency range of operation of the impedance tube is explored. Sensitivity of the method to nuisance parameters is discussed.