Jul 2026
Label-Free Finite-Volume-Residual Training of Attention Graph Neural Networks for Coupled Thermo-Fluid Fields
This work proposes to train an attention graph neural network by minimizing the finite-volume method (FVM) residuals of the governing equations to provide a practical training signal for neural surrogates and reduce the model development cost.
Tian-Yu Li, Zhi-Ying Cao, Qin-Gang Zhang et al.
· arXiv.org · 0 citations