Machine learning for data driven sound propagation modelling

声音对水下活动至关重要，包括通信、海洋生物学、防御和导航。有效地使用当前和未来的传感器系统需要对动态海洋环境中的声音传播有深刻的理解。目前的声音传播模型是有限的，缺乏洞察力，所以这个项目寻求创建数据驱动的、物理上真实的模型，包括复杂的现象，如内波、湍流和散射。该项目旨在提高声纳系统在海洋环境中运行的所有实体的有效性。该项目将开发一系列高保真数字双胞胎，能够封装一些影响声波在海洋环境中传播的关键动态现象，包括内波、多尺度结构热变化和时间变化和波动、非光滑界面和边界（如半淹没结构、海床、表面）的散射、洋流、涡流和锋面。与这些现象相关的机制很少被研究，特别是从数学和物理的角度来看，仍然知之甚少；这一领域的大多数研究本质上是随机的，尽管这些模型提供了有用的预测能力，但就其本质而言，它们无法对所涉及的过程提供真正的物理洞察。本项目将通过开发数据驱动和物理上真实的模型来解决这些缺陷，以便加强对声音在海洋中传播有重大影响的复杂和关键动态现象的理解。该项目的重点将放在机器学习模型上，该模型可以成功地表征动态海洋环境中的声音传播，在物理模型中存在计算或数学上难以表示的多尺度过程，以便有效和智能地估计任何声纳部署的声音传播。

Sound is crucial for underwater activities spanning communication, marine biology, defence, and navigation. Effective use of current and future sensor systems requires a deep understanding of sound propagation in dynamic ocean environments. Current sound propagation models are limited and lack insight, so this project seeks to create data-driven, physically realistic models encompassing complex phenomena like internal waves, turbulence, and scattering. The project aims to improve sonar system effectiveness for all entities operating in ocean environments.This project will develop a series of high-fidelity digital twins capable of encapsulating a number of critical dynamic phenomena, which affect the propagation of sound waves through ocean environments, including internal waves, multi-scale structural thermal and temporal variations and fluctuations, scattering by non-smooth interfaces and boundaries (e.g. semi-submerged structures, sea bed, surface), currents, eddies, and fronts. The mechanisms associated with these phenomena are rarely studied and remain poorly understood, particularly from a mathematical and physical perspective; the majority of studies in this area are stochastic in nature and, although these models provide useful predictive capability, by their very nature, they cannot offer real physical insight into the processes involved. The present project will address these deficiencies by developing models that are simultaneously data-driven and physically realistic in order to enhance the understanding of complex and critical dynamic phenomena that have substantial impact on the propagation of sound through our oceans.The focus of the project will be on machine learning models that can successfully characterise sound propagation in dynamic ocean environments, in the presence of multi-scale processes which are computationally or mathematically difficult to represent in physical models, in order to efficiently and intelligently estimate sound propagation for any sonar deployment.