Bayesian-based estimation of acoustic surface impedance: Finite difference frequency domain approach

Area of Science:

• Acoustics
• Materials Science
• Computational Physics

Background:

• Accurate acoustic performance prediction relies on precise surface acoustic impedance data of boundary materials.
• Traditional analytical methods are limited to simple geometries, necessitating flexible numerical approaches.
• Inverse numerical methods are employed for parameter estimation by minimizing discrepancies between predicted and observed acoustic fields.

Purpose of the Study:

• To develop and validate a geometry-agnostic Bayesian-network sampling approach for estimating surface acoustic impedance.
• To mitigate the impact of experimental, model, and numerical noise on parameter estimation accuracy.
• To demonstrate the method's performance using an air-backed micro-perforated panel in an impedance tube.

Main Methods:

• A Bayesian-network sampling approach was implemented for parameter estimation.
• The method was applied to determine the surface acoustic impedance of an air-backed micro-perforated panel.
• Performance was evaluated against established methods like the ISO two-microphone impedance tube method and theoretical models.

Main Results:

• The developed Bayesian method demonstrated robust parameter estimation, even in the presence of noise.
• Results showed good agreement with predictions from the ISO standard method and a theoretical material model.
• Analysis of Bayesian data by-products provided insights into method sensitivity to nuisance parameters.

Conclusions:

• The geometry-agnostic Bayesian sampling method offers a flexible and noise-resilient approach for surface acoustic impedance estimation.
• This technique enhances the accuracy of interior acoustic performance predictions.
• The method's ability to analyze data by-products aids in understanding its reliability and limitations.