Equivalent source method-based sound field reconstruction using sparse Bayesian learning within a compressive sensing framework

This paper addresses the issue of sound field reconstruction with a limited number of measurement points by proposing an equivalent source method based on sparse Bayesian learning within a compressive sensing framework. The approach establishes a reconstruction model that integrates the equivalent source method with sparse Bayesian learning. Hyperparameters are inferred using the MacKay iteration, enabling the determination of the optimal sparse solution for equivalent source strengths and thereby achieving efficient sound field reconstruction. Numerical simulations demonstrate that, compared to the conventional equivalent source method, the proposed approach achieves higher reconstruction accuracy across the entire frequency range and exhibits favorable robustness.