基于数据驱动的浅海波导声辐射预测方法  

A Data-Driven Method for Predicting Shallow Sea Waveguide Acoustic Radiation

作　　者：翟景娟 符宁 商林源 ZHAI Jingjuan;FU Ning;SHANG Linyuan(College of Aerospace Engineering,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学航空宇航学院,辽宁沈阳110136

出　　处：《郑州航空工业管理学院学报》2024年第6期95-102,共8页Journal of Zhengzhou University of Aeronautics

基　　金：国家自然科学基金(12102270)。

摘　　要：结合计算力学和深度学习理论,提出一种基于数据驱动的浅海波导声辐射预测方法。通过集成有限元法—虚质量法—基于虚源法的边界元法的计算框架,获取浅海波导声辐射响应,为深度学习提供大量声压数据;进而,通过训练ConvNeXt卷积神经网络提取采样点声场特征;最终实现浅海水下结构辐射声压级的快速预测,并通过算例表明预测方法的有效性。This article combines computational mechanics and deep learning theory to propose a data-driven shallow water waveguide acoustic radiation prediction method.Integrating finite element method,virtual mass method,and boundary element method based on virtual source method,a computational framework for vibration sound radiation of structures in shallow seawater has been developed to obtain waveguide sound radiation response and provide a large amount of sound pressure data for deep learning.Furthermore,by training ConvNeXt convolutional neural network,the sound field features of sampling points are extracted.Ultimately,the rapid prediction of radiation sound pressure lev-els for structures in shallow seawater is achieved.The numerical results demonstrate the effectiveness of the proposed shallow water waveguide acoustic radiation prediction method.

关 键 词：波导声辐射 基于虚源法的边界元法 ConvNeXt网络 深度学习 

分 类 号：U674.70[交通运输工程—船舶及航道工程]