An instrumental variable approach for the identification of time-varying, Hammerstein systems

Abstract

This paper introduces an iterative algorithm for the identification of time-varying (TV)-Hammerstein systems. This system is composed by a TV static nonlinearity followed by a TV Box-Jenkins linear model. The algorithm uses two basis function expansions: one to represent the TV parameters and a second to approximate the output of the static nonlinearity. A simulation study showed that the algorithm accurately identified the shape of the TV static nonlinearity and linear dynamic elements even though the noise model structure was unknown.