Diagnóstico de falhas em motores de tração de locomotivas diesel-elétricas baseado em sistemas inteligentes

Abstract

Railway transport is an important mode of transport and the vehicle used to move freight or passenger trains is the locomotive. One of the most used types of locomotive is the diesel-electric locomotive, characterized by having a diesel engine that drives an electric generator to power the electric motors, called traction motors. Faults in the traction motors has an impact on the operation of the locomotives and affect the maintenance area ofthe companies, since their maintenance represents a significant cost. In this context, fault diagnosis in locomotive traction motors is relevant and several approaches have been proposed in the literature. This work proposes and evaluates three models of intelligent systems applied to the fault diagnosis in traction motors: a multilayer artificial neural network, a neurofuzzy network and an evolving fuzzy classifier. The results of the computational experiments performed demonstrate that all models achieve good performance in the fault diagnosis, with better results presented by the neurofuzzy network. The results also demonstrate that the artificial neural network allows the diagnosis to be carried out more quickly and that the evolving fuzzy classifier allows the faults to be learned online and in real time.