Identification of Wiener-Hammerstein models using linear interpolation in the frequency domain (LIFRED)

A new method to identify the linear subsystems of a Wiener-Hammerstein model through the measurement of the second-order Volterra kernel is proposed. This technique makes use of the symmetry properties of the Volterra kernel and assumes that the frequency response gain and phase between estimated points can be reasonably well approximated by a straight line. The signal applied for the identification is a multisine with properties of no interharmonic distortion. Several advantages of the proposed method over existing ones are discussed, and two simulation examples are presented to illustrate the applicability of the technique. The method is also shown to be robust to noise and distortion in the input signal.