一种基于线性声速剖面的射线基神经网络声场预测方法
By optimizing the analytical circular arc propagation model based on the linear sound velocity profile model and training parameters, the problems of large computational load and insufficient interpretability in sound field prediction in marine environments are solved, and efficient and stable multipath sound field reconstruction and extrapolation prediction are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QINGDAO INNOVATION & DEV CENT OF HARBIN ENG UNIV
- Filing Date
- 2026-05-14
- Publication Date
- 2026-07-17
AI Technical Summary
Existing sound field prediction methods are computationally intensive in complex marine environments, and it is difficult to balance speed and accuracy. Furthermore, deep learning models lack interpretability and controllability in terms of propagation path contribution, making it difficult to effectively reconstruct multipath interference sound fields under sparse sampling conditions.
Based on the linear sound velocity profile model and Snell's law, an analytical circular arc propagation model is constructed. By training the logarithmic amplitude parameter, phase offset parameter, and boundary reflection coefficient, a forward model of sound pressure level coherently superimposed along multiple paths is established. The model is then trained and optimized using sparse sampling points to achieve rapid reconstruction and extrapolation prediction of the sound field.
It significantly reduces computational complexity and improves extrapolation efficiency, realizing multipath interferometric acoustic field reconstruction and stable extrapolation prediction under sparse sampling conditions. The path parameters are interpretable and the phase deviation is small.
Smart Images
Figure CN122220790B_ABST