Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks

A method of solving the 2D acoustic wave equation using Fourier Neural Operator (FNO) networks is presented. Various scenarios including wave superposition are considered, including the modelling of multiple simultaneous sound sources, reflections from domain boundaries and diffraction from randomly...

Full description

Saved in:
Bibliographic Details
Main Authors: Middleton Michael, Murphy Damian T., Savioja Lauri
Format: Article
Language:English
Published: EDP Sciences 2025-01-01
Series:Acta Acustica
Subjects:
Online Access:https://acta-acustica.edpsciences.org/articles/aacus/full_html/2025/01/aacus240111/aacus240111.html
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849388032952631296
author Middleton Michael
Murphy Damian T.
Savioja Lauri
author_facet Middleton Michael
Murphy Damian T.
Savioja Lauri
author_sort Middleton Michael
collection DOAJ
description A method of solving the 2D acoustic wave equation using Fourier Neural Operator (FNO) networks is presented. Various scenarios including wave superposition are considered, including the modelling of multiple simultaneous sound sources, reflections from domain boundaries and diffraction from randomly-positioned and sized rectangular objects. Training, testing and ground-truth data is produced using the acoustic Finite-Difference Time-Domain (FDTD) method. FNO is selected as the neural architecture as the network architecture requires relatively little memory compared to some other operator network designs. The number of training epochs and the size of training datasets were chosen to be small to test the convergence properties of FNO in challenging learning conditions. FNO networks are shown to be time-efficient means of simulating wave propagation in a 2D domain compared to FDTD, operating 25 × faster in some cases. Furthermore, the FNO network is demonstrated as an effective means of data compression, storing a 24.4 GB training dataset as a 15.5 MB set of network weights.
format Article
id doaj-art-e40c08fd96c54145b45acb3883af5fbd
institution Kabale University
issn 2681-4617
language English
publishDate 2025-01-01
publisher EDP Sciences
record_format Article
series Acta Acustica
spelling doaj-art-e40c08fd96c54145b45acb3883af5fbd2025-08-20T03:42:25ZengEDP SciencesActa Acustica2681-46172025-01-0192010.1051/aacus/2024078aacus240111Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networksMiddleton Michael0https://orcid.org/0000-0002-0862-7524Murphy Damian T.1https://orcid.org/0000-0002-6676-9459Savioja Lauri2https://orcid.org/0000-0002-8261-4596AudioLab, School of Physics, Engineering and Technology, University of YorkAudioLab, School of Physics, Engineering and Technology, University of YorkDepartment of Computer Science, Acoustics Lab, Aalto UniversityA method of solving the 2D acoustic wave equation using Fourier Neural Operator (FNO) networks is presented. Various scenarios including wave superposition are considered, including the modelling of multiple simultaneous sound sources, reflections from domain boundaries and diffraction from randomly-positioned and sized rectangular objects. Training, testing and ground-truth data is produced using the acoustic Finite-Difference Time-Domain (FDTD) method. FNO is selected as the neural architecture as the network architecture requires relatively little memory compared to some other operator network designs. The number of training epochs and the size of training datasets were chosen to be small to test the convergence properties of FNO in challenging learning conditions. FNO networks are shown to be time-efficient means of simulating wave propagation in a 2D domain compared to FDTD, operating 25 × faster in some cases. Furthermore, the FNO network is demonstrated as an effective means of data compression, storing a 24.4 GB training dataset as a 15.5 MB set of network weights.https://acta-acustica.edpsciences.org/articles/aacus/full_html/2025/01/aacus240111/aacus240111.htmlmachine learningneural networkslinear acousticsreflectionsgeneralisation
spellingShingle Middleton Michael
Murphy Damian T.
Savioja Lauri
Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
Acta Acustica
machine learning
neural networks
linear acoustics
reflections
generalisation
title Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
title_full Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
title_fullStr Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
title_full_unstemmed Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
title_short Modelling of superposition in 2D linear acoustic wave problems using Fourier neural operator networks
title_sort modelling of superposition in 2d linear acoustic wave problems using fourier neural operator networks
topic machine learning
neural networks
linear acoustics
reflections
generalisation
url https://acta-acustica.edpsciences.org/articles/aacus/full_html/2025/01/aacus240111/aacus240111.html
work_keys_str_mv AT middletonmichael modellingofsuperpositionin2dlinearacousticwaveproblemsusingfourierneuraloperatornetworks
AT murphydamiant modellingofsuperpositionin2dlinearacousticwaveproblemsusingfourierneuraloperatornetworks
AT saviojalauri modellingofsuperpositionin2dlinearacousticwaveproblemsusingfourierneuraloperatornetworks