Comment on 'Method for Real-Time Frequency Response and Uncertainty Estimation'

摘要

This comment has clarified misleading assertions from Grauer and Morelli's paper regarding CIFER" frequency domain identification and frequency sweep methods. CIFER frequency response identification techniques were shown to produce superior results to Welch's method shown in the subject paper. This was demonstrated for the F-16 nonlinear simulation using the same multisine input signal usedin the subject paper. In addition, a 25 s frequency sweep was used for frequency response identification. At the discrete multisine input frequencies, the CIFER results for a multisine input have improved accuracy over the 25 s sweep, although both inputs produce highly accurate results. Additionally, considering the reduced frequency resolution of the three-input multisine, the input design requires more a priori knowledge of the aircraft dynamics, especially for lightly damped modes. Using a sweep of the same record length T_(rec) = 25 s with CIFER processing provides for overall increased resolution (less dependency on a priori knowledge) and accuracy comparable to the subject paper's new realtime method, and much improved over Welch's method.