Physics-informed Neural Networks for the Resolution of Analysis Problems in Electromagnetics

Authors

  • S. Barmada DESTEC, University of Pisa, Pisa, Italy
  • P. Di Barba Dept. of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy
  • A. Formisano Dept. of Engineering, University of Campania “Luigi Vanvitelli,” Aversa, Italy
  • M. E. Mognaschi Dept. of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy
  • M. Tucci DESTEC, University of Pisa, Pisa, Italy

DOI:

https://doi.org/10.13052/2023.ACES.J.381102

Keywords:

Direct and inverse electromagnetic problems, neural networks, physics informed neural networks

Abstract

Learning from examples is the golden rule in the construction of behavioral models using neural networks (NN). When NN are trained to simulate physical equations, the tight enforcement of such laws is not guaranteed by the training process. In addition, there can be situations in which providing enough examples for a reliable training can be difficult, if not impossible. To alleviate these drawbacks of NN, recently a class of NN incorporating physical behavior has been proposed. Such NN are called “physics-informed neural networks” (PINN). In this contribution, their application to direct electromagnetic (EM) problems will be presented, and a formulation able to minimize an integral error will be introduced.

Downloads

Download data is not yet available.

Author Biographies

S. Barmada, DESTEC, University of Pisa, Pisa, Italy

Sami Barmada received the M.S. and Ph.D. degrees in electrical engineering from the University of Pisa, Italy, in 1995 and 2001, respectively. He is currently a full professor with the Department of Energy and System Engineering, University of Pisa. He is author and co-author of more than 180 papers in international journals and indexed conferences. His research interests include applied electromagnetics, electromagnetic fields calculation, power line communications, wireless power transfer devices, and nondestructive testing.

Prof. Barmada is an Applied Computational Electromagnetics Society (ACES) Fellow, and he served as ACES president from 2015 to 2017. He is chairman of the International Steering Committee of the CEFC Conference and he has been the general chairman and technical program chairman of numerous international conferences.

P. Di Barba, Dept. of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy

Paolo Di Barba is a full professor of electrical engineering in the Department of Electrical, Computer, and Biomedical Engineering, University of Pavia, Pavia, Italy. His current research interests include the computer-aided design of electric and magnetic devices, with special emphasis on the methods for field synthesis and automated optimal design. He has authored or coauthored more than 240 papers, either presented to international conferences or published in international journals, the book Field Models in Electricity and Magnetism (Springer, 2008), the monograph Multiobjective Shape Design in Electricity and Magnetism (Springer, 2010) and the book Optimal Design Exploiting 3D Printing and Metamaterials (2022).

A. Formisano, Dept. of Engineering, University of Campania “Luigi Vanvitelli,” Aversa, Italy

Alessandro Formisano is a full professor at Università della Campania “Luigi Vanvitelli.” His scientific activity started in 1996, in cooperation with several research groups active in the fields of electromagnetic fields and devices (e.g., EdF Paris, TU-Graz, TU-Budapest, TU-Ilmenau, TU-Bucharest, Slovak Academy of Science, Grenoble Univ.), and thermonuclear controlled fusion (KIT, ITER, Fusion For Energy, EURATOM). His interests are electromagnetic fields computation, neural networks, robust design and tolerance analysis, thermonuclear plasmas identification, optimal design, and inverse problems in electromagnetism. He serves as editorial board member or reviewer for the most prestigious journals (IEEE Trans. On Magn., Compel, Sensors, ACES Journal) in the field of numerical computation of electromagnetic fields.

M. E. Mognaschi, Dept. of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy

Maria Evelina Mognaschi is associate professor at the University of Pavia (Italy), Department of Electrical, Computer and Biomedical Engineering. Her scientific interests are inverse problems, in particular multi-objective optimization and identification problems in electromagnetism and biological systems. Recently, she investigated the solution of forward and inverse problems in electromagnetics by means of deep learning techniques. She has authored or co-authored more than 120 ISI- or Scopus-indexed papers, either presented to international conferences or published in international journals.

M. Tucci, DESTEC, University of Pisa, Pisa, Italy

Mauro Tucci received the Ph.D. degree in applied electromagnetism from the University of Pisa, Pisa, Italy, in 2008. Currently, he is a full professor with the Department of Energy and Systems Engineering, University of Pisa. His research interests are machine learning, data analysis, and optimization, with applications in electromagnetism, nondestructive testing, and forecasting.

References

A. Kovacs, L. Exl, A. Kornell, J. Fischbacher, M. Hovorka, M. Gusenbauer, L. Breth, H. Oezelt, D. Praetorius, D. Suess, and T. Schrefl, “Magnetostatic and micro magnetism with PINNs,” J. of Magnet. and Mag. Mat., vol. 548, 2022.

A. Khan and D. A. Lowther, “Physics informed neural networks for electromagnetic analysis,” IEEE Transactions on Magnetics, vol. 5, no. 9, 2022.

S. Barmada, P. Di Barba, A. Formisano, M. E. Mognaschi, and M. Tucci, “Learning inverse models from electromagnetic measurements data,” Proc. of IGTE Symp. 2022, Graz (Austria), Sep. 18-21, 2022.

M. Raissi, P. Perdikaris, and G. E. Karniadakis, “PINNs: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations,” J. Comput. Phys., vol. 378, pp. 686707, 2019.

P. Rumuhalli, L. Udpa, and S. Udpa, “Finite element neural networks for elec. inverse problems,” Rev. Q. Nondestr. Eval., vol. 21, p. 28735, 2002.

W. Tang, T. Shan, X. Dang, M. Li, F. Yang, S. Xu, and J. Wu, “Study on Poissons equation solver based on deep learning technique,” 2017 IEEE El. Design of Adv. Pack. and Syst. Symp., Haining, China, 2018.

Z. Zhang, L. Zhang, Z. Sun, N. Erickson, R. From, and J. Fan, “Study on a Poissons equation solver based on deep learning technique,” Proc. of J. Int. Symp. on Elect. Compat., Sapporo, p. 305308, 2019.

B. Bartlett, “A generative model for computing electromagnetic field solutions,” Stanford CS229 Projects, 233 Stanford, CA, 2018.

J. Lim and D. Psaltis, “MaxwellNet: Physics-driven deep neural network training based on Maxwells equations,” APL Photonics, vol. 7, 2022.

Y. Chen, L. Lu, G. Karniadakis, and L. Negro, “Physics-informed neural net. for inv. problems in nano-optics and metamat,” vol. 28, no. 8, 2020.

L. Lu, R. Pestourie, W. Yao, Z. Wang, F. Verdugo, and S. Johnosn, “Physics-informed neural networks with hard constraints for inverse design,” J. Sci. Comput., vol. 43, no. 6, pp. B1105-B1132, 2021.

S. Wang, Z. Pend, and C. Christodoulou, “Physics-informed deep neural networks for transient electromagnetic analysis,” Antennas and Propagation, vol. 1, pp. 404-412, 2020.

M. Baldan, G. Baldan, and B. Nacke, “Solving 1D non-linear magneto quasi-static Maxwells equations using neural networks,” IET Sci. Meas. Technol., vol. 15, pp. 204217, 2021.

A. Kovacs, L. Exl, A. Kornell, J. Fischbacher, M. Hovorka, M. Gusenbauer, L. Breth, H. Oezelt, D. Praetorius, D. Suess, T. Schref, “Magnetost. and microm. with PINNs,” J. of Magnet. and Magnetic Materials, vol. 548, 2022.

A. Beltran-Pulido, I. Bilionis, and D. Aliprantis, ”PINNs for Solving Parametric Magnetostatic Problems,” IEEE Trans. Energy Conversion, 2022.

M. Baldan, P. Di Barba, and D. A. Lowther, “Physics-informed neural networks for solving inverse electromagnetic problems,” IEEE Trans. Magnetics, vol. 59, no. 5, 2023.

S. H. Rudy, S. L. Brunton, J. L. Proctor, and J. N. Kutz, “Data-driven discovery of partial differential equations,” Sci. Adv., vol. 3, no. 4, Art. no. e1602614, 2017.

Z. Long, Y. Lu, X. Ma, and B. Dong, “PDE-Net: Learning PDEs from data,” in Proc. 35th Int. Conf. Mach. Learn., pp. 3208-3216, 2018.

V. Dwivedi and B. Srinivasan, “Physics informed extreme learning machine (PIELM)—A rapid method for the numerical solution of partial,” Neurocomputing, vol. 391, pp. 96-118, 2020.

S. Mishra, “A machine learning framework for data driven acceleration of computations of differential equations,” Math. Eng., vol. 1, no. 1, pp. 118-146, 2018.

A. G. Baydin, B. A. Pearlmutter, A. A. Radul, and J. M. Siskind, “Automatic differentiation in machine learning: a survey,” J. of Machine Learning Res., vol. 18, no. 153, pp. 1-43, 2018.

O. C. Zienkiewicz and R. Taylor, The Finite Element Method (4th ed.), New York: McGraw-Hill Book Co., 1989.

COMSOL Multiphysics® v. 6.2. www.comsol.com. COMSOL AB, Stockholm, Sweden.

Downloads

Published

2023-11-30

How to Cite

[1]
S. Barmada, P. D. Barba, A. Formisano, M. E. Mognaschi, and M. Tucci, “Physics-informed Neural Networks for the Resolution of Analysis Problems in Electromagnetics”, ACES Journal, vol. 38, no. 11, pp. 841–848, Nov. 2023.

Issue

2023: Vol 38 Iss 11

Section

General Submission

Categories