Modified frequency identification method and its application to helicopter modeling

A frequency identification method for a six-DoF state-space model of helicopter is presented. The overall concept is to extract a complete set of non-parametric input-to-output frequency responses that fully characterizes the coupled helicopter dynamics. This set of frequency responses is the basis for the model simplification, and for conducting a nonlinear search based on K-means clustering algorithm for a six-DoF linear state-space model that matches the frequency-response data set. A new method for combining the results of multi-input frequency-response analyses obtained from a range of spectral windows into a single optimized response is presented. This method significantly improves the dynamic range and the accuracy of the identified frequency-response relative to single-window methods. The accuracy of the identified model is validated by comparing the model-predicted responses with the responses collected during flight experiments.