Optimizing Metro Vehicle Maintenance: A Deep Learning Framework for Failure Prediction of Critical Components

Begena Araujo, Tulio (2023) Optimizing Metro Vehicle Maintenance: A Deep Learning Framework for Failure Prediction of Critical Components. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (3MB) | Preview

Abstract

Predictive Maintenance (PdM) is the state-of-the-art strategy for maintenance management. It is able to provide cost reduction in various sectors, including manufacturers, power plants, and transportation systems. This research proposes a new PdM framework that could be used by railway operations to enhance their vehicles maintenance strategy. This was done adapting the Cross Industry Standard Process for Data Mining (CRISP-DM) for the specificities of this project. The framework comprises four tiers: Data Acquisition, Data Transferring, Data Processing, and Decision Making. A Deep Learning model coupling Long Short-Term Memory with Autoencoder was developed and tested with a recent published dataset built with real data from sensors. The metrics reached in this study were 65% of Recall, 28% of Precision, and 40% F1 Score. These results, together with the method used to get them, mean that the proposed framework can be a viable strategy for implementing PdM.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Jilani, Musfira UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HD Industries. Land use. Labor > Specific Industries > Motor Industry
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	07 May 2025 10:58
Last Modified:	07 May 2025 10:58
URI:	https://norma.ncirl.ie/id/eprint/7498

Actions (login required)

View Item