Physics-informed neural networks for inverse problems in structural dynamics

Rafael de O. Teloli; Mael Bigot; Lucas Coelho; Emmanuel Ramasso; Roberta Tittarelli; Patrice Le Moal; Morvan Ouisse

doi:10.1117/12.3010918

9 May 2024 Physics-informed neural networks for inverse problems in structural dynamics

Rafael de O. Teloli, Mael Bigot, Lucas Coelho, Emmanuel Ramasso, Roberta Tittarelli, Patrice Le Moal, Morvan Ouisse

Proceedings Volume 12950, Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, Civil Infrastructure, and Transportation XVIII; 129500E (2024) https://doi.org/10.1117/12.3010918
Event: SPIE Smart Structures + Nondestructive Evaluation, 2024, Long Beach, California, United States

Abstract

This study introduces an innovative approach that employs Physics-Informed Neural Networks (PINNs) to address inverse problems in structural analysis. Specifically, we apply this technique to the 4-th order PDE of Euler-Bernoulli formulation to approximate beam displacement and identify structural parameters, including damping and elastic modulus. Our methodology incorporates partial differential equations (PDEs) into the neural network’s loss function during training, ensuring it adheres to physics-based constraints. This approach simplifies complex structural analysis, even when specific boundary conditions are unavailable. Importantly, our model reliably captures structural behavior without resorting to synthetic noise in data. This study represents a pioneering effort in utilizing PINNs for inverse problems in structural analysis, offering potential inspiration for other fields. The reliable characterization of damping, a typically challenging task, underscores the versatility of methodology. The strategy was initially assessed through numerical simulations utilizing data from a finite element solver and subsequently applied to experimental datasets. The presented methodology successfully identifies structural parameters using experimental data and validates its accuracy against reference data. This work opens new possibilities in engineering problem-solving, positioning Physics-Informed Neural Networks as valuable tools in addressing practical challenges in structural analysis.

Conference Presentation

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Rafael de O. Teloli, Mael Bigot, Lucas Coelho, Emmanuel Ramasso, Roberta Tittarelli, Patrice Le Moal, and Morvan Ouisse "Physics-informed neural networks for inverse problems in structural dynamics", Proc. SPIE 12950, Nondestructive Characterization and Monitoring of Advanced Materials, Aerospace, Civil Infrastructure, and Transportation XVIII, 129500E (9 May 2024); https://doi.org/10.1117/12.3010918

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available