High quality frequency response function measurements without user interaction

In this paper we propose a method to measure the frequency response function (FRF) of a system using arbitrary excitations without any user interaction. The choice of the excitation is left free, arbitrary, random, or periodic excitations are allowed depending on the preference of the user. The local polynomial method (LPM) is used to reduce the leakage errors on the FRF to extremely low levels compared with what is obtained with the classical windowing methods. The LPM provides also an estimate of the disturbing noise variance as a function of the frequency. The only parameter to be tuned in this method is the local bandwidth that sets how many neighbouring frequency lines are combined in a single point estimate. In this paper we propose a modified AIC-criterion to address this question so that the mean square error of the estimates is minimised.