Nonparametric Method for Identification of Mimo Hammerstein Models

A new nonparametric approach to identify multivariable Hammerstein models is presented in this paper. The linear dynamic subsystem is identified and represented by its finite impulse response (FIR) model, and, the static nonlinearity is identified and represented as an MIMO input-output mapping. By specially designed test signals, the estimation of FIRs for multivariable linear subsystems can be conducted under a SISO framework and can be decoupled from the identification of the static nonlinearity. Due to the nonparametric nature, the representation of MIMO Hammerstein model may not be unique. By making uses of this fact, several parameters can be adjusted to shape the model for achieving engineer’s requirement. These above-mentioned representations can be used to obtain an exact process model or an apparent model suitable for control design. Copyright © 2007 IFAC