Session Description:

This session focuses on recent advances in data-driven machine learning methods for underground engineering. It aims to explore how the increasing geotechnical data are reshaping existing workflows in underground engineering, including project design, construction, deformation control, risk assessment, and operational management. By integrating artificial intelligence with multi-source geotechnical data, the session seeks to facilitate the transition of underground engineering toward more intelligent, adaptive, and data-centric paradigms. Topics of interest include but are not limited to:

• Prediction and control of excavation-induced deformation
• Uncertainty analysis and reliability assessment of underground space development
• Risk assessment and early warning technologies for underground defects, including tunnel cracks, water ingress, and cavity formation
• Data-driven performance evaluation of shield tunneling and geological classification
• Application of large AI models (e.g., LLMs, multimodal models) for decision support in underground engineering