In ight estimation of bending modes of air vehicles, is useful whenadaptive notch ltering is required to avoid excitation of the bendingmodes by the stability augmentation system. While single-tone fre-quency estimation from noiseless data can be a relatively easy task,identifying say the 1st modal frequency from sensors data, while allother modes are active as well, and in the presence of control signalsand measurement noise is no longer an easy task. The present worksuggests addressing the problem of 1st modal frequency estimation byusing a team approach where several algorithms are teamed togetherto accomplish the task : Pre Processing to remove irrelevant frequencybands and envelope detection, Kalman Filter (Extended, Second OrderGaussian or Unscented), Robust Zero Cross Detection and a simpliedFunctional Neural Network realized using a Kalman lter as well, forAuto Regressive Moving Average system identication. The suggestedteam algorithm is applied on a synthetic as well as real data to demon-strate its performance.
展开▼
