A Model of Second Order Hammerstein Series for Nonlinear System

Abstract

Abstract. In this paper, modeling a weakly nonlinear system whose nonlinearity is up to the second order is studied. During this task a second order Hammerstein series model is used because it can lead to a trade-off between computational cost and generalization ability. To extract such model form measured input-output data, a nonparametric algorithm based on the minimum mean square error criterion is proposed. The polyspectrum up to order 4 is used to determine the Hammerstein kernels, while the input is a random multi-sine signal. Finally the proposed modeling method is validated on simulated system.