Design criteria for robust strict positive realness in adaptive schemes (Q1323770)

The authors investigate the property of a family of polynomials over specific regions of the complex plane, and they give criteria for design of filters ensuring convergence of recursive estimation algorithms in the presence of uncertainty in the plant model. The problem of enlarging the positive realness of the transfer functions in order to improve behaviour of the algorithm in broader operating conditions is treated. Also the problem of minimizing the phase of the transfer function at certain points \(z\) of the complex plane is treated. A geometrical approach for the filter design is given by introducing the lead and lag functions. The uncertainty is assumed to be both structured and unstructured. The latter is given by a bound on the frequency response magnitude, while the former is characterized in terms of regions of polynomial root locations. An analytical solution for a certain class of root location regions is given. A solution to devise a technique for tackling the problem when general regions of root locations are of concern is provided. A numerical example illustrates how the design procedure proposed by the authors can be used to improve the convergence of the recursive adaptive algorithms.