Filtro digital adaptivo integrado

This thesis presents a study of various techniques of digital signal filtering to determine which provides greater convergence when applied to time-invariant linear systems such as the least squares and the stochastic gradient method, using in all of them the ARMA (1) models (autoregressive moving average, first-order stochastic model). We have made emphasis in the analysis of adaptive filtering techniques to develop algorithms that allow us to identify and estimate parameters integrated within a system seen as a black box, in such a manner that it becomes possible to conceptualize their level of convergence and to improve algorithms that are currently used in this important area that is involved in both artificial vision and complex control systems, where information prediction, description and reconstruction are required. The algorithms presented here have been developed in an analytical manner on the basis of cited literature and the necessary mathematical tools. All of them were simulated using MathLab.