Physics-informed neural networks for shunted piezoelectric systems
Acta Mechanica · 2026-06 · 237 (6) · 2745-2767
DOI: 10.1007/s00707-025-04619-9 ↗Kurz gefasst
Fachartikel in der Zeitschrift Acta Mechanica, online erschienen am 05.02.2026. Verfasst von Aliki D. Mouratidou und 7 weiteren Autor:innen. Die Arbeit wurde bisher 4-mal zitiert (Crossref) und nennt 41 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Greek-French Scientific Cooperation Hubert Curien, Technical University of Crete.
Abstract (Originalsprache)
Shunted piezoelectric systems are often used for vibration suppression in smart structures. Multiphysics modeling using finite elements leads to coupled partial differential equations reflecting the strong electromechanical interaction. In the present contribution, a multi-physics-informed neural network (multi-PINN) is proposed, to solve for the first time the direct electromechanical vibration suppression problem with the shunted circuit. The mathematical framework to convert the electromechanical problem to the appropriate PINN formulation is analytically provided. Static and dynamic analysis using PINN models for a cantilever piezoelectric beam are presented. The problem of vibration reduction by means of a resonant shunt is also studied. To facilitate the solution of the dynamic problem, Galerkin’s projections are applied to discretize the dynamic model with respect to the spatial variable. A time partitioning approach is also proposed to decrease the computational cost of the dynamic problem. The obtained system of ordinary differential equations (ODE) is solved by means of the multi-PINN to accurately predict the dynamic response to a principal mode resonance excitation. The proposed method is validated using finite element analysis and MATLAB solutions. Numerical results show the efficiency and robustness of the proposed techniques.
Förderung
- Greek-French Scientific Cooperation Hubert Curien
- Technical University of Crete
Zitieren
APA
Mouratidou, A. D., Daraki, M., Drosopoulos, G. A., Foutsitzi, G., Larbi, W., Deü, J., Ohayon, R., & Stavroulakis, G. E. (2026). Physics-informed neural networks for shunted piezoelectric systems. Acta Mechanica, 237(6), 2745-2767. https://doi.org/10.1007/s00707-025-04619-9
Vancouver
Mouratidou AD, Daraki M, Drosopoulos GA, Foutsitzi G, Larbi W, Deü J, et al. Physics-informed neural networks for shunted piezoelectric systems. Acta Mechanica. 2026;237(6):2745-2767. doi:10.1007/s00707-025-04619-9
BibTeX
@article{Mouratidou2026,
author = {Mouratidou, Aliki D. and Daraki, Maria-Styliani and Drosopoulos, Georgios A. and Foutsitzi, Georgia and Larbi, Walid and Deü, Jean-Francois and Ohayon, Roger and Stavroulakis, Georgios E.},
title = {Physics-informed neural networks for shunted piezoelectric systems},
journal = {Acta Mechanica},
year = {2026},
volume = {237},
number = {6},
pages = {2745--2767},
doi = {10.1007/s00707-025-04619-9}
}
EUVIMED