A Physics-Driven Framework for Parametric Periodic-Flow Modeling and Finite-Amplitude Aeroelastic Response AnalysisNew

This paper addresses the computational challenge of parametric analysis for periodic unsteady flows commonly occurring in forced-motion and fluid-structure interaction problems. The authors propose a purely physics-driven framework that eliminates the need for pre-generated flow-field or aerodynamic data. The core innovation is a Periodic Physics-Informed Neural Network (P-PINN) that enforces periodic boundary conditions and physics constraints to solve parametric periodic flows and predict finite-amplitude aeroelastic responses efficiently.