arXiv：cs.LG· Weizheng Zhang, Xunjie Xie, Hao Pan, Lin Lu·· 4 小时前
Cova-PINN：面向复杂几何流固共轭传热的跨域守恒物理信息神经网络
Cova-PINN: Cross-Domain Conservation Physics-Informed Neural Network for Fluid-Solid Conjugate Heat Transfer in Complex Geometries
AI 导读
Cova-PINN 是一个多域 PINN 框架，将守恒支撑与复杂几何中的热交互路径对齐，联合优化局部跨域复合控制体积平衡与全局换热器尺度上的配对壁面闭合。在四个 TPMS 换热器及 DualMS 设计上，相比最接近的基线 MUSA-PINN-CHT，其平均出口温度误差与器件级闭合误差分别降低 37.7% 和 60.2%，全流场与热负荷精度也有提升。
正文
Abstract:Multi-domain physics-informed neural networks (PINNs) flexibly model medium-specific representations to solve fluid--solid conjugate heat transfer (CHT). However, standard multi-domain PINNs enforce governing equations and interface conditions on separately sampled domain supports, which can yield plausible temperature fields but inaccurate end-to-end energy transfer and outlet temperatures. We propose Cova-PINN, a multi-domain PINN framework that aligns conservation support with thermal interaction paths in complex geometries. Cova-PINN jointly optimizes cross-domain composite control-volume balances at the local scale and paired-wall closure at the global exchanger scale. We evaluate Cova-PINN on four triply periodic minimal surface (TPMS) heat exchangers and a geometrically distinct DualMS design against CHT-specific, optimization-oriented, and complex-geometry PINN baselines under a common protocol. Relative to the closest baseline, MUSA-PINN-CHT, Cova-PINN reduces average outlet-temperature and device-level closure errors across the four TPMS topologies by $37.7\%$ and $60.2\%$, respectively, while also improving full-field and heat-duty accuracy, with consistent gains on DualMS.
| Subjects: | Machine Learning (cs.LG); Fluid Dynamics (physics.flu-dyn) |
| Cite as: | arXiv:2610.11108 [cs.LG] |
| (or arXiv:2610.11108v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11108 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Weizheng Zhang [view email]
[v1]
Thu, 8 Oct 2026 02:28:18 UTC (12,450 KB)
来源：arXiv：cs.LG · arxiv.org