Predictive Maintenance Framework for Urban Metro Vehicles

Our study emphasizes the critical role of Predictive Maintenance (PdM) in safeguarding companies against system failures and accidents. By timely identifying and classifying potential failures, PdM can help reduce the risk of accidents, enhance safety measures, minimize downtime, and improve vehicle maintenance. To this end, we utilized two Machine Learning algorithms, XG Boost and Logistic Regression, to predict the occurrence of failures and determine the superior framework for specific types of failure prediction. Our proposed method is based on the MetroPT dataset and brings numerous advantages to metro and rail operators. By proactively predicting maintenance requirements, management can address potential system failures before they become serious issues. This approach significantly bolsters vehicle reliability and reduces downtime in the metro network, ensuring a more resilient and dependable transportation system. The analog sensors data, including air tank pressure, compressor oil temperature, and flowmeter values, play a vital role in this framework.