Characterizing Morphing Wing States via Stochastic Functional Models for Fly-by-Feel Aircraft

Biological fliers continuously reshape their wings during flight, enabling them to optimize aerodynamic performance and maintain stability under widely varying conditions—capabilities that conventional unmanned aerial vehicles (UAVs) do not yet fully possess. Advancing morphing wing technology requires not only mechanically adaptive structures but also a unified modeling framework capable of sensing and tracking configuration-dependent aeroelastic behavior in real time. This work investigates the characterization of morphing wing structural states by leveraging dynamic strain measurements and stochastic global system identification. A custom spanwise camber-morphing wing testbed instrumented with piezoelectric strain sensors was developed, featuring a 45° dihedral adjustment envelope enabled through a dual-joint tip architecture inspired by avian wing biomechanics. Broadband white-noise excitation was applied across ten discrete morphing angles, and the resulting vibration response was analyzed using non-parametric spectral estimation to reveal configuration-dependent modal shifts associated with geometric morphing. Based on these observations, a global Vector-dependent Functionally Pooled AutoRegressive (VFP-AR) model was identified to represent the aeroelastic response as a smooth function of morphing configuration. Compared to a conventional multi-model AR baseline, the VFP-AR model achieved significantly improved prediction accuracy while requiring over 90% fewer parameters, demonstrating strong statistical parsimony and generalization across the morphing envelope. These results validate the suitability of stochastic global modeling for morph-aware state representation and provide a foundation for future real-time morphing state estimation and fly-by-feel control architectures in adaptive UAV platforms.