This paper compares the
performances of the unscented filter (UF) against the extended Kalman filter
(EKF) for aeroelastic systems with polynomial restoring forces. Using data
resulted from numerical simulations and wind tunnel experiments, both methods are
applied for system identification, filtering and long term prediction. In all
case studies reported here, the unscented filter gives better or comparable
results with the extended Kalman filter Moreover, as an alternative to the
smoothing based on the EKF, we propose the unscented smoothing which does not
require the computation of the Jacobians.