Automated Real-Time Ground Pressure Prediction and Tunnel Stability Assessment Using Hybrid Deep Learning and Geomechanical Modeling for Open-Face TBMs in Soft Ground Conditions.

**Abstract:** This paper introduces a novel methodology for enhancing tunnel stability and optimizing excavation parameters in open-face Tunnel Boring Machines (TBMs) operating in challenging soft ground conditions. Our approach integrates advanced deep learning techniques with established geomechanical modeling methods to create a hybrid system capable of predicting ground pressure in real-time and assessing tunnel stability risks with unprecedented accuracy. This significantly reduces the potential for ground collapse, minimizes support requirements, and accelerates TBM advance rates, leading to substantial cost savings and improved safety in tunnel construction projects. The system leverages a multi-modal data ingestion pipeline and a recursive evaluation loop to continuously learn and adapt to varying geological conditions, offering a scalable and robust solution for real-time tunnel management.