AUTHOR=Du Libin , Wang Zhengkai , Lv Zhichao , Wang Lei , Han Dongyue 

TITLE=Research on underwater acoustic field prediction method based on physics-informed neural network

JOURNAL=Frontiers in Marine Science

VOLUME=Volume 10 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2023.1302077

DOI=10.3389/fmars.2023.1302077

ISSN=2296-7745

ABSTRACT=In the field of underwater acoustics, numerical simulation methods and machine learning techniques are commonly employed for forecasting underwater acoustic fields. However, the numerical simulation method requires grid division while the machine learning method often lacks the ability to analyze the physical significance of the model. To address these problems, this paper proposes an underwater acoustic field prediction method based on physics-informed neural network (UAFP-PINN).Firstly, a loss function incorporating physical constraints is introduced, incorporating the Helmholtz equation that describes the characteristics of the underwater acoustic field. This loss function serves as a foundation for establishing the underwater acoustic field prediction model using a physics-informed neural network. The model takes the coordinate information of the acoust field point as input and employs a fully connected deep neural network to output the predicted value of the coordinates. The predicted value is then refined using the loss function with physical information, ensuring that the trained model possesses clear physical significance.Finally, the proposed prediction model is analyzed and validated in two dimensions: the two-dimensional sound field and the threedimensional sound field. The results demonstrate the effectiveness of the model in predicting the distribution of the two-dimensional underwater acoustic field. By appropriately increasing the complexity of the prediction network, the model also exhibits excellent predictive performance for three-dimensional acoustic field prediction.