Stationary behavior of an anti-windup scheme for recursive parameter estimation under lack of excitation
Abstract
Stationary properties of a recently suggested windup preventionscheme for recursive parameter estimation are investigated in thecase of insufficient excitation. When the regressor vectorcontains data covering the whole parameter space, the point ofconvergence of the Riccati equation is uniquely defined by thechoice of the weighting matrix. If the excitation is insufficient,the algorithm is shown to possess a manifold of stationary pointsand a parametrization of this manifold is given. However, if thepast excitation conditions already caused the algorithm toconverge to a certain point, the stationary solution would not beaffected by current lack of excitation. This property guaranteesgood anti-windup properties of the studied parameter estimationalgorithm.
