Physics-Informed Data-Driven Modelling of Nonlinear Aerodynamic Forces of the Pazy Wing

A reduced-order model of the Pazy wing that builds stability and steady-state deformation into the identification itself, capturing nonlinear aerodynamic forcing across a wide range of angles of attack without interpolating between linearized models.

1. 01

   Extends the Dynamic Mode Decomposition with Control framework with a conic optimization problem that embeds physical constraints - stability and steady-state deformation - directly into the model.

2. 02

   Identified from time-domain nonlinear aeroelastic simulation data of the Pazy wing.

3. 03

   Captures the nonlinear aerodynamic forcing across a wide range of angles of attack, removing the need to interpolate across separately linearized models.

4. 04

   Validated both within and beyond the training range, giving a compact, robust representation suited to control applications.

Abstract

Highly flexible wings may exhibit significant geometrical nonlinearities and prediction of their dynamic aeroelastic characteristics poses a substantial computational challenge for wing and control design. To address this, we present a novel data-driven method for nonlinear reduced-order modelling of such wings. Our approach extends the Dynamic Mode Decomposition with Control (DMDc) framework by introducing a conic optimization problem that embeds physical constraints, namely stability and steady-state deformation, directly into the model. Using time-domain nonlinear aeroelastic simulation data of the Pazy wing, a data-driven model is generated that accurately captures the nonlinear aerodynamics forcing across a wide range of angles of attack. Our method ensures adherence to known aerodynamic behaviour and eliminates the need for interpolation across linearized models. Validation has been carried out and demonstrates the effectiveness of the method in modelling aerodynamic forces and moments both within and beyond the training range, providing a compact, robust representation suitable for control applications.