Flutter simulation and prediction via identification of non-linear impulse response

Abstract

System identification techniques to model the unsteady, non-linear aerodynamics are considered in the present study for flutter simulation and prediction. In order to capture the behavior of a system, we excite it using two different input signals-the commonly used 3211 multistep function and the filtered impulse method-and observe the behavior of the system identification model in prediction. Compared to the usual multistep function method, the filtered impulse method is computationally efficient as the order of model required to capture the functional relation is lower. It also gives higher accuracy around the distinct frequencies where flutter is likely to occur. In this paper we propose two viable system identification models for flutter modeling, one of which is based on the linear autoregressive moving average model while the other is a non-linear model based on Radial Basis Function. Results of NACA0012 and NACA64a010 (Isogai Wing) for unsteady flow and flutter computations are presented in this study.