Decoupling the linear and nonlinear parts in Hammerstein model identification (Q1433069)

It is known that in Hammerstein model identifications, a major difficulty is the coupling between the linear and the nonlinear parts. However, the author shows that, under the pseudo-random binary sequences input they can be decoupled and thus, identification of the linear part is essentially a linear problem. This greatly reduces the complexity and improves the efficiency of the identification algorithm.