End-to-end prediction and inverse identification of gear grinding and EOL performance based on physics-informed neural networks

Ziqian Liu

doi:10.1016/j.ymssp.2026.114720

End-to-end prediction and inverse identification of gear grinding and EOL performance based on physics-informed neural networks

2026-08-06

0

PRE

AI

Z

Ziqian Liu

D

Dong Guo *

Y

Yuan Cheng

Y

Yuji Zhou

X

Xuelai Liu

N

Nan Feng

DOI：10.1016/j.ymssp.2026.114720

Original

Original request for help

Save

Abstract

En 中文

• A compensation network maps vibrations for high-fidelity 3D waviness inversion. • A state-dependent neural operator compensates unmodeled dynamics in the ODE solver. • Empirical EOL vibrations enable closed-loop calibration without causal leakage.

Keywords:

Physics-informed neural networks

Gear grinding

Vibration prediction

Tooth surface waviness

Ghost noise

End-to-end prediction and inverse identification of gear grinding and EOL performance based on physics-informed neural networks

Abstract

Journal

Organization

Cited Papers

Citing Papers