A Covariance-Based Realization Algorithm for the Identification of Aeroelastic Dynamics from In-Flight Data

The unsteady, aerodynamically induced resonance of aircraft structures may lead to potentially destructive vibrations when left unaccounted for in flight control system design. Such aeroelastic modes are difficult to accurately predict analytically, and computational models require calibration, verification, and validation. The successful design of control systems that actively suppress aeroelastic vibrations thus requires the capability to identify unbiased parametric estimates of aeroelastic resonance modes. We present a novel subspace system identification method inspired by covariance estimates and classical realization techniques that constructs system estimates from measured input-output data. The resulting covariancebased realization algorithm allows for the identification of parametric system models from data sets of large signal dimension and is applicable to data perturbed by colored noise and acquired in closed-loop operation due to the unbiased estimation of cross-covariance functions, even in low signalto-noise conditions. The algorithm is applied to data measured on board the NASA Active Aeroelastic Wing F/A-18. The results demonstrate the effectiveness of the algorithm in efficiently computing accurate, unbiased linear dynamic models ∗Graduate Student, Dept. of Mechanical and Aerospace Engineering, [email protected], Student Member AIAA. †Professor, Dept. of Mechanical and Aerospace Engineering, [email protected]. ‡Aerospace Engineer, [email protected], Senior Member AIAA.