A linear system identification and validation of an AH-64 Apache aeroelastic simulation model

This study addresses a linear math model validation of a global, nonlinear real-time rotary wing flight simulator at hover and 130 knots. The simulation is the AH-64 attack helicopter with a rotor blade element model. The main rotor dynamics have been formulated to provide a consistent matching between structural and aerodynamic theory. The structural model is a representation of the flexible blade based on a priori mode shape data. The dynamic inflow model is an adaptation of the Peters/He theory. Frequency response testing of the simulation model permits an evaluation of the mid to high frequency range. The Comprehensive Identification from FrEquency Responses (CIFER) program is used to (1) extract a complete set of non-parametric input-to-output frequency responses that fully characterize the coupled helicopter dynamics, and (2) conduct a nonlinear search for a state space model that matches the frequency response data set. CIFER is used to extract state space stability derivative models from both flight test data and simulator generated responses. These linear models are compared with the vehicle response in both the frequency domain and in the time domain to confirm predictive capabilities to dissimilar input forms. The hover results obtained showed that a high fidelity identification of the state space linear models was achieved. Both linear models proved an excellent match with the frequency response set and demonstrated excellent predictive capabilities. The 6-DOF model fit was significantly improved by inclusion of the higher order rotor terms. The effects of these higher order terms were most apparent on the pitch rate, roll rate, and collective response of the aircraft. The forward flight data set was not as high quality as the hover data set and low signal to noise ratios hindered the identification process. Nevertheless, it was possible to identify a 6-DOF model that matched the aircraft frequency response set quite closely. Of extreme significance is confirmation that the FLIGHTLAB solution technique used to model the Apache, correctly predicts the off-axis response of the aircraft. The methodology described herein explores a new approach to the rotorcraft system identification problem, simplifying the selection of model parameters to be changed and aiding in determining an adequate model form and structure.