Metamodeling elastic wave propagation using a mixed factorized Fourier encoder–decoder for online laser-ultrasound testing in additive manufacturing

Laser-ultrasound (LU) testing has emerged as a promising technique for characterizing the polycrystalline microstructure of metal components produced by wire-laser additive manufacturing (WLAM), with potential for real-time online application. Numerical models simulating elastic waves propagation pr...

Celý popis

Uloženo v:

Podrobná bibliografie
Vydáno v:	Engineering applications of artificial intelligence Ročník 160; číslo Part C; s. 111893
Hlavní autoři:	Allaire, Frédéric, Gatti, Filippo, Miorelli, Roberto, Dorval, Vincent, Laurent, Jérôme
Médium:	Journal Article
Jazyk:	angličtina
Vydáno:	Elsevier Ltd 27.11.2025 Elsevier
Edice:	Engineering Applications of Artificial Intelligence
Témata:	Acoustics Additive manufacturing Artificial Intelligence Computer Science Deep learning Laser-ultrasound Machine Learning Mechanics Modeling and Simulation Optics Physics Polycrystalline media Surface acoustic waves Surrogate modeling Deep learning Polycrystalline media Additive manufacturing Surrogate modeling Laser-ultrasound Surface acoustic waves modeling simulation anisotropy elastic waves neural network non-destructive testing Laser guided waves
ISSN:	0952-1976
On-line přístup:	Získat plný text
Tagy:	Přidat tag Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!

Popis
Shrnutí:	Laser-ultrasound (LU) testing has emerged as a promising technique for characterizing the polycrystalline microstructure of metal components produced by wire-laser additive manufacturing (WLAM), with potential for real-time online application. Numerical models simulating elastic waves propagation provide valuable insights into the relationship between microstructural properties and laser-induced displacements, but their computational cost renders them impractical for automated high-throughput characterization. To overcome this limitation, we build a metamodel that maps a wide variety of two-dimensional anisotropic polycrystalline microstructures — simplified but representative of features commonly observed in WLAM — to simulated surface displacements. Addressing this challenging high-dimensional regression problem, several neural network surrogates relying on a novel combination of layers are investigated. Their architectures include usual convolutional encoder–decoder elements and spectral layers inspired from the Fourier neural operator (FNO) framework. These layers are adapted and some variants are proposed to provide versatility in network design. All metamodels can run both a forward and backward pass at least 100 times faster than a single forward call of the original model. The best architecture implies a trade-off between computational cost and accuracy. Notably, the architecture involving the channel-wise factorized variant of the spectral layers, which is characterized by a relatively small number of parameters, achieved the lowest approximation error. The metamodel successfully captures the primary effects of anisotropy on wave propagation, even for low-anisotropy inputs not included in the training data. These findings represent a promising initial step towards addressing inversion problems and facilitating the development of online LU testing protocols in additive manufacturing. •Metamodeling of simulated laser-induced 2D elastic wave propagation in polycrystals.•Including variability of input microstructure in wire-laser additive manufacturing.•Deep neural networks for a high-dimensional regression problem.•Hybrid design with spectral layers (cf. Fourier neural operators) and convolutions.•Fast and accurate approximation of anisotropy and scattering effects.
ISSN:	0952-1976
DOI:	10.1016/j.engappai.2025.111893