Measuring a linear approximation to weakly nonlinear MIMO systems

The choice of the input signals has impact on the nonparametric frequency response function (FRF) measurements of a nonlinear MIMO system. It is shown that Gaussian noise, periodic noise, and random multisines are equivalent, yielding in the limit the same linear approximation to a nonlinear MIMO system. Even in the noiseless case, variability of the FRF is observed due to the presence of the nonlinearity and the randomness of the excitation. Based upon these effects a new class of equivalent input signals is proposed, decreasing essentially the variance of the nonlinear FRF measurements, while the same linear approximation is retrieved. Copyright © 2005 IFAC