An Online Data-Driven Predictive Maintenance Approach for Railway Switches

Tomé, Emanuel Sousa; Ribeiro, Rita P.; Veloso, Bruno; Gama, João

doi:10.1007/978-3-031-23633-4_27

Cited by 2 publications

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results of two long-span bridges show that transfer learning is effective in improving detection performance with very few labeled data, many better than classical methods. In one of the rst works on the topic, Sousa Tom´e et al [14] developed an online, data-driven predictive maintenance framework for railway switches, an essential element within the railway infrastructure. Their approach uses real-time data logging from the railway's interlocking system to predict maintenance needs, aiming to improve maintenance e ciency and railway switch reliability.…”

Section: Introductionmentioning

confidence: 99%

Enhancing Metro Rail Efficiency: A Predictive Maintenance Approach Leveraging Machine Learning and Deep Learning Technologies

Nair,

et al. 2024

Preprint

View full text Add to dashboard Cite

This paper looks into the modeling and implementation of a predictive maintenance system of an air production unit for a metro rail designed to suit the challenges detailed by the industrial sector. Using modern machine learning, deep learning, and AI techniques, the system identifies the faulty equipment well in advance when applied to the huge volume of sensor data. One of the major functionalities of the system is an interface designed to alert users, whereby the instant alerts are made to the maintenance personnel for faster intervention, minimization of the possible downtime. The basis of the study is the application of the predictive maintenance system within the unit of air production. It indicates great efficacy toward the prediction of a failure. A wide variety of ML and deep learning models were experimented with and fine-tuned carefully by training and evaluation over the training set and also over the testing set to ensure predictive accuracies. For example, from the above comparative model analysis, the most suitable predictive approach was indicated through the use of accuracy. Deep Learning Models, including LSTM, RNN, and BiLSTM, have been exceedingly good, with all the above models giving an accuracy of above 99.7 percent. Notably, Adaboost, a Boosting technique also has performed well. The culmination of this project highlights the pivotal role of AI and ML technologies in advancing predictive maintenance strategies within the industrial sector. The findings illustrate the potential of these technologies to transform maintenance practices, optimize operational processes and contribute to the overall sustainability of industrial operations. This paper contributes valuable insights into the feasibility and effectiveness of AI-driven predictive maintenance systems.

show abstract

Section: Introductionmentioning

confidence: 99%