Bayesian admittance estimation and model selection for finite-differential method

Abstract

Surface acoustic impedance or admittance at the boundary is essential for estimating and evaluating a room's acoustic performance. In this article, a Bayesian-based model selection method and parameter estimation is presented to predict the admittance beyond the frequency limit of measurement. A frequency-dependent admittance model is chosen to estimate the boundary condition. The Bayesian-network sampling approach presented in this article shows the capability to predict the surface admittance boudary condition dependent on the frequency. This Bayesian method demonstrates a clear estimation of the pole number and parameters of the chosen frequency-dependent admittance model. The analysis results indicate the potential of applying the Bayesian-based method on boundary condition estimation.