Shallow-Water Sound-Source Localization via Surrogate Models Incorporating Spectral-Elements Full-Wave Numerical Simulation

Matched field processing (MFP) is a powerful tool for sound source localization, where the source location parameters are decided by matching the measured signal with its physical model. However, modeling the sound field in shallow water is a difficult task due to the complex environment. Spectral element method (SEM) is able to simulate the propagation of sound waves accurately in such complicated environments. But this scheme is computationally expensive so that it cannot be used for source localization because, in the matching optimization procedure, the numerical model has to be evaluated many times. To address these problems, a novel shallow water sound source localization algorithm is proposed. More specifically, SEM is used to simulate sound propagation in complex shallow water environment for some potential source locations; then, the Kriging method is employed to establish the surrogate model for sound propagation for any source locations in a considered region. Subsequently, the actual sound signal is matched with the output sound field of the Kriging model based on MFP principles, so as to realize rapid and accurate source localization in the complex shallow water environment. An application example is introduced to demonstrate that the proposed method has the advantages of high localization accuracy and computational efficiency.