Skip to content
DraconiticResearch
Search Sources API
226,742 studies31 sourcesUpdated 35 min ago
Peer-reviewedCC BYJASA Express Letters

Domain-decoupled physics-informed neural networks for inverse material property estimation from acoustic fields

  1. 1Fachgebiet Systemzuverlässigkeit, Adaptronik und Maschinenakustik, Technische Universität Darmstadt, Darmstadt, Hessen 64287, Germany

Abstract

Estimating material properties like Young's modulus and internal damping from radiated acoustic fields is challenging under sparse, noisy measurement conditions. This work employs sequential physics-informed neural networks with Fourier features to estimate these material properties for a vibrating plate. The inverse problem is solved by decoupling fluid and structural domains to effectively filter boundary differentiation and data noise. Multi-frequency statistical ensembles are introduced to quantify the uncertainty coming from neural network initialization robustness and learning biases. Following a validation of the method in a proof of concept on vast, low-noise data conditions, its limits under sparse, noisy data are demonstrated.

Related

  1. Parameter Estimation in Machining Dynamics with Regenerative Delay and Nonsmooth Friction using Physics-Informed Neural Networks
    Meiyazhagan Jaganathan, Vikram Pakrashi, Aasifa Rounak
    arXivPreprintCC BY-NC-ND 4.0Full textStatistical & nonlinear physicsarXiv
  2. How Sparse and How Noisy? Systematic Benchmarking of Inverse Physics-Informed Neural Networks for Manning Friction Estimation in Shallow Water Equations
    Soheil Radfar
    arXivPreprintCC BY 4.0Full textGeophysics & seismologyComputational physicsarXiv
  3. Fabrication framework for three-dimensional colloidal particles with decoupled geometry and material composition
    Hamed Almohammadi, Gurminder K. Paink, Vasumathi Venkat +6
    arXivPreprintarXiv licenceFull textCondensed matterarXiv
  4. Parameter-Decoupled Quantum Superresolution without Multiparameter Estimation
    X. -F. Qian
    arXivPreprintCC BY-NC-SA 4.0Full textOptics & photonicsQuantum physics & quantum informationarXiv
  5. Parameter-Decoupled Superresolution with Airy-Disk-Informed Learning: Application to NGC 4486B
    James Qu, Isabella L. Lee, Gang Zhang +3
    arXivPreprintCC BY-NC-SA 4.0Full textGalaxiesOptics & photonicsarXiv
BibTeX
@article{brand2026domaindecoupled,
  title = {{Domain-decoupled physics-informed neural networks for inverse material property estimation from acoustic fields}},
  author = {Paul Brand and Robert Feldmann and Tobias Melz},
  journal = {JASA Express Letters},
  year = {2026},
  month = nov,
  doi = {10.1121/10.0047523},
  url = {https://doi.org/10.1121/10.0047523},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - Domain-decoupled physics-informed neural networks for inverse material property estimation from acoustic fields
AU  - Paul Brand
AU  - Robert Feldmann
AU  - Tobias Melz
PY  - 2026
DA  - 2026/11/16
JO  - JASA Express Letters
DO  - 10.1121/10.0047523
UR  - https://doi.org/10.1121/10.0047523
AB  - Estimating material properties like Young's modulus and internal damping from radiated acoustic fields is challenging under sparse, noisy measurement conditions. This work employs sequential physics-informed neural networks with Fourier features to estimate these material properties for a vibrating plate. The inverse problem is solved by decoupling fluid and structural domains to effectively filter boundary differentiation and data noise. Multi-frequency statistical ensembles are introduced to quantify the uncertainty coming from neural network initialization robustness and learning biases. Following a validation of the method in a proof of concept on vast, low-noise data conditions, its limits under sparse, noisy data are demonstrated.
SN  - 2691-1191
ER  -