Improved FRF Estimators for MIMO Sine Sweep Data

The dynamic identification of large aerospace structures like satellites, launcher components or aircraft prototypes, requires a great experimental effort in order to assure high quality test results. At the same time, because of the high development costs, a reduction of the test time is needed. In this paper, a test strategy capable to reduce the overall time required to perform dynamic tests aimed to the estimate of the Frequency Response Functions, FRFs, is introduced. A sine sweep excitation is the key point of the approach because of the good compromise between the excitation level and the testing time. Nevertheless, the accuracy of the dynamic identification strongly depends on the non-parametric estimation method used to compute the FRFs from the sine-sweep input-output data. In this paper, the accuracy of the estimates of both the frequency response functions and the modal parameters due to different sweep rates and different FRY estimation methods will be analyzed. These methods include: harmonic estimator, single-block Discrete Fourier Transform (DFT), frequency averaging technique, Welch's method and Reduced Discrete Fourier Transform (RDFT). The performance of the different estimators is critically assessed by a numerical analysis performed on a lumped parameter system and by experimental investigations carried out on both a GARTEUR aircraft scaled model and the Lambert aircraft M212