OPTIMIZATION OF TBM MAINTENANCE USING MACHINE LEARNING

This thesis explores the optimization of maintenance strategies for Tunnel Boring Machines (TBMs) through the application of advanced machine learning techniques. Given the complexity and operational significance of TBMs in major infrastructural projects, enhancing maintenance efficiency is crucial. This research utilized a comprehensive dataset provided by WeBuild, covering extensive operational parameters and downtime events associated with the tunnel excavation of the TELT project. A neural network model, specifically a Multi-Layer Perceptron (MLP), was developed and trained to predict potential downtimes based on operational data. The model architecture included multiple layers with dropout for regularization and utilized ReLU and softmax activation functions to handle the multi-label classification task effectively. The performance of the model was evaluated based on accuracy, precision, recall, and F1-score, demonstrating significant predictive capabilities with an emphasis on the practical application in real-time maintenance decision-making. The findings suggest that machine learning can significantly enhance predictive maintenance strategies, reducing downtime and improving operational efficiency. This study contributes to the field by demonstrating how machine learning can be integrated into the maintenance processes of heavy machinery, with implications for cost savings and operational reliability.