Parameter Identification ofHammerstein Modelsusing Elimination Theory KaiyuWang,MarcBodson, JohnChiasson, andLeonM.Tolbert
Marc Bodson · 2005
A Hammerstein modelisa systemmodelinwhich theinputs go through a static nonlinearity followed bya linear time-invariant system. Oftenthestatic nonlinearity ismodeled as a polynomial nonlinearity intheinputs or as apiecewise con- stantnonlinearity. Suchmodels are nonlinear intheunknown parameters andtherefore present a challenging identification problem. Inthis work,theauthors showthatelimination theory can beusedtosolve exactly forparameter values thatminimize a least-square criterion. Thus,theapproach guarantees the minimumcan befoundina finite numberofsteps, unlike iterative methodsthatare currently used. IndexTerms-Hammerstein models, Parameter identifica- tion, Nonlinear least-squares, Resultants.