TransFlowNet: A physics-constrained Transformer framework for spatio-temporal super-resolution of flow simulations

Xinjie Wang, Siyuan Zhu, Yundong Guo, Peng Han, Yucheng Wang, Zhiqiang Wei, Xiaogang Jin

Abstract: Highlights•TransFlowNet is a novel physics-constrained deep learning framework.•TransFlowNet focuses on the spatio-temporal super-resolution of flow simulations.•TransFlowNet has a newly designed implicit feature encoder based on Swin Transformer.•Two flow simulations are studied to evaluate the performance of TransFlowNet.