Flutter Onset Prediction Based on Parametric Model Estimation

This paper presents a thorough evaluation of three flutter prediction techniques based on the well-understood autoregressive moving-average (ARMA) parametric model. Other than the original stability parameter Fz that is derived from the identified autoregressive coefficients, the flutter margin and the Houbolt-Rainey (H-R) method are also presented and adjusted to provide alternative flutter predictions. From the application aspects, the empirical flutter margin, which is insensitive to the damping estimation, is first proposed. Feasibility of the empirical flutter indictor is demonstrated in this study. Taking advantage of the inherent recursive nature, the recursive least-squares algorithm with forgetting factor is suggested to perform the linear extrapolation iteratively at different flow conditions to facilitate an accurate flutter prediction. To verify the performance of the inspected methods, a linear pitch-plunge aeroelastic system and a dedicated wind-tunnel flutter test of an elastic wing are adopted to carry out the analyses for flutter statistics. While the extensive simulations are easy to implement on the aeroelastic idealization, the challenge of generating an adequate sample size based on the limited experimental data arises. To this end, the concept of surrogate data is first introduced to the flutter statistical analysis. Surrogate datasets generated via the iterative amplitude adjusted Fourier transform algorithm manage to describe the statistical behavior of the evaluated stability parameters. Both numerical and experimental results further substantiate the validity of the ARMA parametric model for aeroelastic modeling and provide alternative flutter prediction parameters with their own scope of application.