Least squares identification of Hammerstein model based on over-sampling scheme

A new least squares type of identification algorithm based on an over-sampling scheme is proposed for a Hammerstein model which consists of a nonlinear element followed by a linear dynamic model. By estimating the intermediate input to the linear part explicitly, we can identify an arbitrary continuous function type of nonlinear element as well as the unknown linear transfer function model. Furthermore, the consistency of the estimated parameters in the linear part can be assured unlike other previous methods even if there is an approximation error in the description of the nonlinear function.