Diagnóstico de falhas em transformadores de potência através de análise de gases dissolvidos usando rede neural artificial

Power transformers are very important equipment in the operation of electrical systems, having the irreplaceable function of transforming voltage and current levels for transmission of electrical energy from generation center to end user. This importance is even greater from the economic point of vi...

Descripción completa

Detalles Bibliográficos
Autor: ENRIQUEZ, Alex Rogelio Soto
Tipo de recurso: tesis de maestría
Estado:Versión publicada
Fecha de publicación:2020
País:Brasil
Institución:Universidade Federal do Maranhão (UFMA)
Repositorio:Biblioteca Digital de Teses e Dissertações da UFMA
Idioma:portugués
OAI Identifier:oai:tede2:tede/3105
Acceso en línea:https://tedebc.ufma.br/jspui/handle/tede/3105
Access Level:acceso abierto
Palabra clave:Transformador de potência
BPSO
K-NN
Redes Neurais Artificiais
Power Transformer
Artificial Neural Network
Sistemas Elétricos de Potência
Descripción
Sumario:Power transformers are very important equipment in the operation of electrical systems, having the irreplaceable function of transforming voltage and current levels for transmission of electrical energy from generation center to end user. This importance is even greater from the economic point of view, since in the event of a failure condition with the consequent interruption of electrical service can lead to major economic losses for both the utility and the end user. An important amount of bibliographies oriented to the maintenance of power transformers in perfect operating conditions is presented in the updated literature. In this master's dissertation, a methodology for diagnosing power transformers failures is developed by applying Binary Particle Swan Optimization (BPSO) to adjust the K-NN classifier (k-Nearest Neighbor) selecting best grouping evaluation variables for a method (waterfall configuration). In the training and testing process for a method based on Artificial Neural Network (ANN) a performance of 100% is achieved, thus constituting a competitive alternative for power transformer fault diagnosis.