O uso de análise de componentes independentes na extração de características dos sinais transitórios de faltas em linhas de transmissão de energia elétrica

Detalhes bibliográficos
Ano de defesa: 2017
Autor(a) principal: Almeida, Aryfrance Rocha
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Tese
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Não Informado pela instituição
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: http://www.repositorio.ufc.br/handle/riufc/27255
Resumo: Several methods for localization and classification of faults in electric power transmission lines, using conventional techniques, computational intelligence and digital signal processing techniques have been proposed, intensively, on literature in the last three decades. These methods have improved the process of characterization of faults in various aspects. However, even the methods based on Wavelet Transform (WT), Artificial Neural Networks (ANN) and other techniques derived from smart Computing, do not have convenient and systematic way treaty faults in transmission systems whose data are contaminated by noise. Based on this evidence, this paper proposes a combination of methods using Independent Component Analysis (ICA), the Theory of the Travelling Waves (TTW) and Support Vector Machine (SVM) effective approaches to extracting characteristics of transient signals of fault even before signs considerably contaminated by noise. The approach was applied to locate and recognize faults in a transmission line 500 kV high voltage that connects the substation of President Dutra - MA to the substation of Boa Esperança - PI. The experiment was carried out for different types of faults that have occurred in different locations. The use of these methods applied to a real transmission line model has proven that the proposed methods, in combination, result in superior performance on location and classification of faults. The obtained errors are less than 1% to the location and accuracy of 100% for the classification of faults with noise. The proposed approach has shown performance best when compared to major conventional techniques, as well as when compared to techniques using Artificial Neural Networks and other computational intelligence techniques