Broadband source localization using near-surface vertical array in deep ocean based on sparse Bayesian learning

This study addresses the limitations of traditional multispectral methods for passive localization of broadband acoustic sources within the deep-sea shadow zone, including incomplete interference structures, insufficient depth resolution, and beamforming grating lobe interference. A high-resolution localization method based on ​sparse Bayesian learning (SBL) is proposed. First, a model of deep-sea shadow zone is established using ray theory, which maps the frequency-angle interference characteristics of the received signals to the depth-range domain. Then, the SBL is introduced into the source localization process to suppress elevation-angle grating lobe interferences while improving angular resolution, thus ensuring the integrity of the interference structures. The method is further extended to estimate the source depth, achieving high-resolution depth-dimensional resolution. Sea trial results demonstrate that the SBL-based approach effectively resolves multi-target localization in passive monitoring of deep-sea broadband acoustic sources.