Detecção e Diagnóstico de Falhas em Sistemas de Arrefecimento de Motor Diesel Ferroviário
Ano de defesa: | 2018 |
---|---|
Autor(a) principal: | |
Orientador(a): | |
Banca de defesa: | |
Tipo de documento: | Dissertação |
Tipo de acesso: | Acesso aberto |
Idioma: | por |
Instituição de defesa: |
Universidade Federal do Espírito Santo
BR Mestrado em Engenharia Elétrica Centro Tecnológico UFES Programa de Pós-Graduação em Engenharia Elétrica |
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://repositorio.ufes.br/handle/10/10711 |
Resumo: | Diesel engines are used in a wide range of industrial applications, from power generators tomechanical motors. The detection of incipient faults is very important for safety and costreduction in any of these applications. This work presents two approaches for detectingfailures in the cooling system of a railway diesel engine and, in one of them, the diagnosis isalso made. The studied engine operates with different rotations and powers that affect thevariables used in its control and protection system. Data from normal and faulty situationsof a real diesel engine were collected through a system developed for data acquisition. Inthe first approach, a classifier is proposed to identify the operating point from rotationmeasurements. In this case, models based on the data are created to estimate the internalpressure of the cooling system, and with them generate residuals used for fault detection.The multiple modes of operation require the use of a classifier to select one of the multiplemodels and corresponding residual normalization. The application of this methodologyallowed to detect the fault caused by water leakage with great anticipation, which is a verydesirable feature in fault detection systems. The second approach uses kNN classifiers andartificial neural networks to detect and diagnose two types of failure: leakage of water andreduction of pressure in the water circuit. Statistical characteristics of the motor speedsignals, pressure and water temperature are used by the classifiers. The classifiers wereevaluated by the mean square error and the classification errors in the confusion matrix.Both classifiers presented good performance for the detection and diagnosis of the twofaults. The methodology allows to increase the number of failures considered. |