TABLE OF CONTENTS The frequency-response method for system identification, as illustrated in Fig. 2.1, starts with flight-test data collected in the time domain. From that point on, the identification method of Chapters 5 through 13 has been carried out entirely in the frequency domain, with the goal of identifying an accurate mathematical model of the system. The assessment of parametric identification results is based entirely on a comparison of the flight-data frequency responses with those of the model and the resulting frequency-domain cost function J ave. An important assessment of model fidelity, robustness, and the limitations of the linear model is provided by evaluating its predictive capability in the time domain for test inputs, such as steps or doublets, that are dissimilar from those used in the identification.
Time-domain verification is accomplished by the direct integration of the equations of motion, using the flight-test measurements of the control inputs. The identification parameters are held fixed at this point. Only the flight-data biases and reference shifts are determined in the time domain because these cannot be found from the frequency domain analysis. The predicted response of the model is then compared with the flight data.
The key topics to be covered in this chapter include the following: motivation for time-domain verification, time-domain verification method, time-domain verification in CIFER using VERIFY, and example results.
Time-domain verification is an important method for assessing the predictive accuracy and reliability of the identified model. Confidence must be gained that the model...
