Vectorized Matrix Formulation for Sound Power Determination Through Inverse Problems and Regularization Techniques

Abstract

Sound power estimation is a widely investigated research topic. Some of its state-of-the-art measurement solutions include taking velocity values on the source boundary, discrete pressure assessment on a virtual outer surface or evaluations with a sound intensity probe. The increased availability of microphone arrays proposes a new tool to obtain these results. The inverse boundary element method, the equivalent source method, the Helmholtz equation least-squares, and an approach based on the radiation resistance matrix provide vectorized formulations for estimating sound power. In this study, the results for different regularization techniques on the sound power deviation from a discrete line source are presented based on a backward reconstruction scheme from simulated sound pressure field points on four microphone arrays. Additionally, in this study, time improvement between single-frequency calculations and multi-frequency execution on multiple graphics processing units is presented. The outcome shows good accuracy and reduced computation time in graphics card implementations.