Fast Inverse Identification of the Coefficient of Friction in Cold Strip Rolling Using a Physics-Informed Neural Network
Materials Science Forum · 13.04.2026 · 1183 · 55-67
DOI: 10.4028/p-y3mapp ↗Kurz gefasst
Fachartikel in der Zeitschrift Materials Science Forum, online erschienen am 13.04.2026. Verfasst von Christian Overhagen und 1 weiteren Autor:innen. Die Arbeit nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Friction plays an important role in flat rolling processes for the force and power demands,kinematics and final product quality. In search for a method of in-situ characterization of the coefficientof friction (COF) by non-contacting measurements, a method for determination of the COF from highaccuracy forward slip measurements, i.e. by laser doppler velocimetry combined with a comprehensiveevaluation without simplifications is the method of choice. The evaluation must comprise the volumeflux equilibrium at the neutral point and the roll gap exit, as well as the connection of the neutral pointand COF by von Karman’s ODE. This iterative procedure involves multiple solutions of the ODE aswell as the nonlinear volume flux relation. In the present work, this problem is addressed by a physicsinformed neural network (PINN), providing a rapid connection between the forward slip and the COFbased on the mathematical rolling theory. The contribution shows that we can solve von Karman’sODE inversely by a PINN, enablind detection of the COF and the neutral angle from the measured forward slip.
Zitieren
APA
Overhagen, C., & Martin, R. (2026). Fast Inverse Identification of the Coefficient of Friction in Cold Strip Rolling Using a Physics-Informed Neural Network. Materials Science Forum, 1183, 55-67. https://doi.org/10.4028/p-y3mapp
Vancouver
Overhagen C, Martin R. Fast Inverse Identification of the Coefficient of Friction in Cold Strip Rolling Using a Physics-Informed Neural Network. Materials Science Forum. 2026;1183:55-67. doi:10.4028/p-y3mapp
BibTeX
@article{Overhagen2026,
author = {Overhagen, Christian and Martin, Robert},
title = {Fast Inverse Identification of the Coefficient of Friction in Cold Strip Rolling Using a Physics-Informed Neural Network},
journal = {Materials Science Forum},
year = {2026},
volume = {1183},
pages = {55--67},
doi = {10.4028/p-y3mapp}
}
EUVIMED