Examining Vortex-Induced Vibration through Convolutional Neural Networks

Abstract

This paper aims to apply CNNs to fluid-structure interaction (FSI) problems. It is worth noting that most prior research utilizing neural networks for fluid flow prediction assumed stationary boundaries. However, for FSI applications, CNNs must predict flow fields with moving boundaries. To address this challenge, we have designed and trained a CNN specifically tailored to predict unsteady, incompressible fluid flow with moving boundaries.