Estimation of Kautz Poles in Wiener-Volterra Models Using Levenberg-Marquardt Algorithm

This work approaches the problem of estimating Kautz optimal poles in kernel expansion Wiener-Volterra models. The analytical solution for suboptimal case is already established literature. However, two parameters that compose still open. In this paper, an optimization strategy using Levenberg-Marquardt presented. algorithm used to find parameters, with same base all dimensions. construction bases digital filter considered. To validate implemented algorithm, data collected from excitation electrically coupled drive system was analyze impact search space thresholds and behavior Levenberg-Marquardt’s parameters. It also analyzed on model accuracy, as number functions increased. As a result, models determined have achieved better results than works found