Eduardo Furtado

Predictive Maintenance of Mining Trains

Featured image
Image generated with FLUX.1 [schnell]

I developed a predictive maintenance model to reduce train breakdowns for a mining company’s railway system connecting their mine to the port. Overheating wheel bearings were the primary cause of costly on-track failures, with repairs requiring trains to be transported to a garage with limited capacity, causing further delays.

Using Python, Scikit-learn, and TensorFlow, I benchmarked machine learning models to classify overheating bearings and prioritize trains with a high failure risk for maintenance. The model was optimized for precision to avoid unnecessary jobs that could overwhelm the garage.

The model was deployed on the company’s servers with a dashboard to display the predictions, enabling maintenance staff to plan and schedule activities more effectively. This solution significantly reduced train delays, decreased unexpected breakdowns, and improved overall operational reliability.