Physics-Informed Neural Networks - A New Field in Mechanical Engineering Education
DOI:
https://doi.org/10.70567/mc.v42.ocsid8481Keywords:
Engineering Education, Physics-Informed Neural Networks (PINNs), Mechanical EngineeringAbstract
This paper presents a short workshop for teaching numerical methods at the School of Engineering
of Universidad de la República (UdelaR) that integrates deep neural networks and, in particular,
PINNs as a complement to traditional approaches. The workshop combines weekly modules with lectures
and interactive notebooks, guiding learners from a basic neural network to PINN applications for
differential-equation problems. Case studies include: a classical physical system with a known analytical
solution, 1D heat transfer, 2D linear solid mechanics, and 2D incompressible fluid dynamics. We
emphasize variable normalization and non-dimensionalization, adaptive loss weighting, and boundarycondition
design. Assessment is a group project that solves a benchmark and contrasts results with numerical
simulation. The workshop links physical modeling and machine learning, strengthens scientific
programming, critical analysis, and collaborative work, and offers transferable guidelines for engineering
education.
References
Barrows H.S. Problem-based learning in medicine and beyond: A brief overview. New directions for teaching and learning, 1996(68):3–12, 1996. http://doi.org/10.1002/tl.37219966804.
Cuomo S., Cola V.S.D., Giampaolo F., Rozza G., Raissi M., y Piccialli F. Scientific machine learning through physics-informed neural networks: Where we are and what’s next. CoRR, abs/2201.05624, 2022. http://doi.org/10.1007/s10915-022-01939-z.
Google. Google colaboratory. https://colab.research.google.com/, 2024. Recuperado el 18 de abril de 2024.
Granger B. y Pérez F. Jupyter: Thinking and storytelling with code and data. En Proceedings of the 26th ACM SIGPLAN conference on programming language design and implementation, páginas 1–7. 2021. http://doi.org/10.1109/MCSE.2021.3059263.
Oommen V. y Srinivasan B. Solving inverse heat transfer problems without surrogate models: A fast, data-sparse, physics informed neural network approach. Journal of Computing and Information Science in Engineering, 22(4):041012, 2022. ISSN 1530-9827. http://doi.org/10.1115/1.4053800.
Prince M. Does active learning work? a review of the research. Journal of engineering education, 93(3):223–231, 2004. http://doi.org/10.1002/j.2168-9830.2004.tb00809.x.
Raissi M., Perdikaris P., y Karniadakis G.E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics, 378:686–707, 2019. http://doi.org/10.1016/j.jcp.2018.10.045.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Argentine Association for Computational Mechanics
This work is licensed under a Creative Commons Attribution 4.0 International License.
This publication is open access diamond, with no cost to authors or readers.
Only those papers that have been accepted for publication and have been presented at the AMCA congress will be published.
