Abstract:
Frequency-response function calculation is an essential element in rotorcraft data analysis, especially for system identification, model validation and model analysis. This article analyzes the performance of a recently developed method for the calculation of frequency-response functions, called the local polynomial method (LPM), when applying it to rotorcraft systems. The local polynomial method is presented as an alternative to the methods that are based on segmenting and windowing the data and is particularly effective in reducing the leakage error. The performance of both methods is first compared based on simulated data from linear helicopter models in various conditions. The simulated data allows to separately investigate the influence of: signal-to-noise ratio, signal length, frequency range, type of excitation and input correlation. The conditions under which the local polynomial method is superior to the windowing methods are given. Finally, flight test data are used to validate the results observed in the simulations.