Regularization of the Bezier-Bernstein polynomial functions and its application in the problem of nonlinear system identification

Abstract

The problem of determination of optimal distribution of the basic control points is considered, where the inverse Casteljau mapping provides the best approximation of the Bezier-Bernstein polynomial functions by the Bezier curves and surfaces. A number of optimality criteria for the system of basic control points are introduced, and appropriate estimations are adduced. An evolutional construction method for the inverse Casteljau mapping in the problem of nonlinear dynamic system identification using neurofuzzy models is proposed.