Predictive Maintenance of Urban Metro Vehicles: Classification of Failures Using Machine Learning

Predictive maintenance (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$P$</tex>dM) is crucial for com-panies to avoid unplanned outages, increase overall reliability, and lower operating costs. Failure detection and classification is a key element of predictive maintenance. In this study, a machine learning framework for identifying failures in the Air Production Unit (APU) of metro vehicles in real time was proposed. The framework was built to have the potential to create a recommendation system for APU failures. On one hand, we proposed a binary classification that can be used in real-time to immediately inform the metro operators of a potential future failure. On the other hand, we proposed a multiclassification algorithm that can identify the exact type of a failure and inform the metro operators to take the proper action. The proposed framework yielded promising results with a relatively high F1-score of about 85% for the binary classifier and 97% for the multiclassification using RF algorithm on the MetroPT dataset. We also found that analog sensors were more discriminant features than digital sensors in classifying APU failures. The pressure of the trains' installed air tanks, the oil temperature on the compressor, and flowmeter values were found to be highly important. By predicting when maintenance is needed, operators can address potential safety issues before they become serious problems, improve the reliability of the metro system, and reduce disruptions for passengers. The proposed framework is applicable in the field and can help operators make more informed decisions about when to repair or replace assets.