Underwater Acoustic Research Trends with Machine Learning: Ocean Parameter Inversion Applications

Abstract

Underwater acoustics, which is the study of the phenomena related to sound waves in water, has been applied mainly in research on the use of sound navigation and range (SONAR) systems for communication, target detection, investigation of marine resources and environments, and noise measurement and analysis. Underwater acoustics is mainly applied in the field of remote sensing, wherein information on a target object is acquired indirectly from acoustic data. Presently, machine learning, which has recently been applied successfully in a variety of research fields, is being utilized extensively in remote sensing to obtain and extract information. In the earlier parts of this work, we examined the research trends involving the machine learning techniques and theories that are mainly used in underwater acoustics, as well as their applications in active/passive SONAR systems (Yang et al., 2020a; Yang et al., 2020b; Yang et al., 2020c). As a follow-up, this paper reviews machine learning applications for the inversion of ocean parameters such as sound speed profiles and sediment geoacoustic parameters.

Enhancing Few-Shot Prediction of Ocean Sound Speed Profiles through Hierarchical Long Short-Term Memory Transfer Learning

“…With the continuous development of the global ocean observation network and the maturation of theories and practices in machine learning and deep learning, an increasing number of scholars are applying machine learning and deep learning methods to address relevant issues in the field of oceanography [33,34]. For example, Jain et al, aiming to demonstrate the feasibility of estimating SSPs using satellite-derived sea surface parameters, successfully inverted SSPs using artificial neural network (ANN) methods [35].…”

GIT: A Transformer-Based Deep Learning Model for Geoacoustic Inversion

“…PE (pos,2i) = sin pos/10000 2i/dim PE (pos,2i+1) = cos pos/10000 2i/dim , (11) where pos means the order of the frequency or sensor, i ∈ [1, dim/2], and dim is the embedding dimension. For broad signals with F frequency points and L sonar receivers, the token embedding is attached with an (F, dim) frequency positional embedding in the first dimension, and an (L, dim) sensor positional embedding in the second dimension.…”

“…Motivated by the development of deep learning, its excellent techniques have been increasingly applied in marine acoustics, leading to remarkable achievements and superior performance compared to traditional approaches [10,11]. Various neural networks have been adopted for geoacoustic inversion.…”

Inversion of shallow seabed structure and geoacoustic parameters with waveguide characteristic impedance based on Bayesian approach

“…The geoacoustic parameters in each seabed layer, such as sound speeds, density and attenuations, all have significant effects on the underwater acoustic field prediction, the sonar performance prediction, the underwater acoustic detection and the underwater acoustic communication (Zhang et al, 2022;Zhao et al, 2022). Therefore, the acquisition of geoacoustic parameters is particularly important in shallow sea (Zhang et al, 2019;Yang et al, 2020). Although, the geoacoustic parameters can be obtained by direct core sample measurements, it is very time consuming and costly to obtain enough samples for a large area.…”

“…In these studies, the geoacoustic parameters could be inverted by matching the propagation characteristics of the acoustic waves with replicates from the acoustic computational model. As a results, the geoacoustic parameters inversion method was proposed (Yang et al, 2020). Hermand (Hermand, 1999) investigated an inversion method to rapidly estimate the distribution of seabed acoustic features using a controlled source and a single hydrophone.…”