資源・素材学会 年次大会

Tunnel Boring Machines (TBMs) have been widely used in tunnel excavation in recent years due to their high efficiency and low environmental load. However, TBMs are highly sensitive to the surrounding geological conditions during excavation. Encountering unexpected geological changes such as faults and fractured zones can pose significant risks and challenges to TBM excavation. Therefore, predicting the rock mass conditions ahead of a tunnel face during TBM excavation is crucial for safe and efficient operation. With the development of artificial intelligence technologies, machine learning and deep learning methods have been increasingly researched and applied in various fields. The data obtained in the TBM excavation process are well-suited for machine learning modeling. Previous studies have applied machine learning methods to TBM operation data for the prediction of rock mass conditions, but there is still room for improvement on the accuracy. In this study, advanced machine learning and deep learning methods were applied to the actual data obtained during TBM excavation. The focus was on predicting the rock mass conditions ahead of a tunnel face, such as rock mass strength and its changes. The study explored which methods are suitable for TBM data modeling and how to improve the accuracy of these models to meet engineering requirements.