Métodos iterativos lineares projetados através da teoria de controle e suas aplicações

Abstract

It is shown how to built a family of new iterative methods using a feedback control viewpoint together with appropriate choices of control Liapunov function, leading to a family of time varying PI/PD controllers. The new iterative methods are used to determine the optimal coefficients of adaptive filters. Simulations are performed in order to show that the new methods have the same performance as the RLS algorithm.

Control Liapunov function; iterative methods; stability; PI/PD controller; conjugate gradient algorithm; convergence; adaptive filters