Aplicação de transformada wavelet e análise de componentes principais robusta em sinais acústicos para detecção de falhas de curto circuito de estator

Signal processing techniques applied to pattern recognition and fault diagnosis collaborate to avoid unplanned maintenance in the production process. Three-phase induction motors are widely used and exposed to various undesirable operational situations that can cause failures. In order to avoid unpl...

Descripción completa

Detalles Bibliográficos
Autor: Cinel, Murilo Monteiro
Tipo de recurso: tesis de maestría
Estado:Versión publicada
Fecha de publicación:2020
País:Brasil
Institución:Universidade Tecnológica Federal do Paraná (UTFPR)
Repositorio:Repositório Institucional da UTFPR (da Universidade Tecnológica Federal do Paraná (RIUT))
Idioma:portugués
OAI Identifier:oai:repositorio.utfpr.edu.br:1/5443
Acceso en línea:http://repositorio.utfpr.edu.br/jspui/handle/1/5443
Access Level:acceso abierto
Palabra clave:Processamento de sinais
Motores elétricos de indução
Localização de falhas (Engenharia)
Signal processing
Electric motors, Induction
Fault location (Engineering)
CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
Engenharia Elétrica
Descripción
Sumario:Signal processing techniques applied to pattern recognition and fault diagnosis collaborate to avoid unplanned maintenance in the production process. Three-phase induction motors are widely used and exposed to various undesirable operational situations that can cause failures. In order to avoid unplanned downtime, several failure detection and monitoring techniques were created. This work proposes a methodology for digital signal processing applied to acoustic emission signals from the operation of a motor capable of classifying stator short-circuit failures. The method is based on the application of the robust principal component analysis algorithm in audio signals. The signal is decomposed by the wavelet transform in order to obtain characteristics of different frequency ranges. Then calculates the attributes that represent the signal under analysis. For validation of the study, the classification of stator failures is proposed using a support vector machine that reached more than 99 % of correct calassifications.