Stiffness-Conditioned Neural Operator for Fluid-Structure Interaction

A hybrid CNN-Transformer architecture uses a stiffness-conditioned neural evolution operator to predict fluid-induced vibrations in flexible plates. The model employs 101 structural tokens and bidirectional cross-attention to capture three distinct response regimes based on nondimensional bending stiffness.