Identificação de Harmônicas em Sistemas Elétricos de Potência Utilizando o Algoritmo NLMS Tensor

  • Thiago Silva Fernandes Laboratório de Sistemas Adaptativos e Processamento de Sinais-LSAPS, Programa de Pós-Graduação em Engenharia Elétrica, Universidade Federal do Maranhão - UFMA, São Luís-MA
  • Francisco das Chagas de Souza Laboratório de Sistemas Adaptativos e Processamento de Sinais-LSAPS, Programa de Pós-Graduação em Engenharia Elétrica, Universidade Federal do Maranhão - UFMA, São Luís-MA
Keywords: Adaptive filter, harmonic components, linear operators, stochastic gradient, tensor adaptive algorithms

Abstract

A new approach to the identification of harmonic components in electrical power systems using adaptive tensor algorithms is presented in this article. The proposed method uses the separability property of linear operators as tool for the derivation of a stochastic gradient algorithm that is used to update the weights of the adaptive filter. Numerical experiments show that the proposed tensorial method has lower computational cost and higher convergence speed when compared to classic adaptive algorithms, in cases where linear operators are perfectly separable.
Published
2021-10-20
Section
Articles