Spatio-temporal Swin Transformer-based Flow-solid Coupling Interaction Sequence Image Prediction Network

Authors

Changjun Zou, School of Information and Software Engineering, East China Jiaotong University, Nanchang 330013, China
Zhiyu Ge, School of Information and Software Engineering, East China Jiaotong University, Nanchang 330013, China
Chenxi Zhong, School of Information and Software Engineering, East China Jiaotong University, Nanchang 330013, China

Abstract

Abstract: To address limitations in modeling long-term dependencies and multi-scale features in fluidstructure interaction scenarios, a spatiotemporal deep learning model (SwinLSTM) integrating ConvLSTM and Swin Transformer is proposed. The model employs a gated spatiotemporal attention mechanism that dynamically embeds Swin Transformer's window-based multi-head self-attention into ConvLSTM's output gate, enabling adaptive temporal-spatial feature coupling, and designs a multi-level ConvLSTM framework to hierarchically capture complex spatiotemporal correlations. Experiments on a self-built fluid-interaction dataset show that our method achieves the highest PSNR and leading SSIM scores, with superior performance in preserving vortex details and boundary consistency. This work provides an efficient solution for fluid dynamics prediction in interactive scenarios.

Recommended Citation

Zou, Changjun; Ge, Zhiyu; and Zhong, Chenxi (2026) "Spatio-temporal Swin Transformer-based Flow-solid Coupling Interaction Sequence Image Prediction Network," Journal of System Simulation: Vol. 38: Iss. 1, Article 9.
DOI: 10.16182/j.issn1004731x.joss.25-0824
Available at: https://dc-china-simulation.researchcommons.org/journal/vol38/iss1/9

First Page

112

Last Page

124

CLC

TP391.41

Recommended Citation

Zou Changjun, Ge Zhiyu, Zhong Chenxi. Spatio-temporal Swin Transformer-based Flow-solid Coupling Interaction Sequence Image Prediction Network[J]. Journal of System Simulation, 2026, 38(1): 112-124.

DOI

10.16182/j.issn1004731x.joss.25-0824