关键词: 敞开式TBM, 硬岩破碎带, 卡机风险, 机器学习, 智能判识, 围岩压力

Abstract: The authors discuss the advance identification and pretreatment techniques for jamming risks of open tunnel boring machines (TBMs) when traversing fractured zones in hard rock. This issue is addressed by a case study conducted on a TBM tunnel project that statistically analyzes the TBM driving parameters and shield-monitoring data accumulated over 16 km of tunneling. The results reveal the mechanism that jams the TBM when boring through fractured hard rock and the typical parametric response. By analyzing the parameters, primary factors such as maximum TBM thrust, maximum cutterhead torque, maximum force on the shield, and maximum variation in shield displacement are selected to assess jamming risks. Machine learning algorithms are used to construct a model to predict jamming risks. Furthermore, a jamming-type classification method based on multiparameter weight analysis is proposed, enabling the accurate identification of various jamming scenarios and providing a theoretical basis for targeted on-site mitigation measures. Additionally, the effectiveness of various measures designed to reduce jamming risks is systematically analyzed using a discrete-element numerical simulation. Finally, optimized treatment solutions for various types of jamming are proposed based on engineering practice. The research results demonstrate the following: (1) During jamming incidents, TBM thrust and torque remain consistently high; the shield exhibits either significant inward displacement or notable outward displacement under active jacking by the operator, and the shield radial force increases sharply and remains high. (2) The neural network (random forest) prediction model achieves an accuracy of 80.8% (84.6%). (3) Densely arranged steel arch supports reduce surrounding rock pressure by approximately 10%, whereas advance support measures reduce it by over 50%. (4) When shield jamming is anticipated, the cutterhead and belt conveyor rescue mode should be activated while strengthening the arch support. When the shield and cutterhead both exhibit jamming risk, advance support must be adopted, and the support structure behind the shield should be reinforced.

Key words: open TBM, hard-rock fractured zone, jamming risk, machine learning, intelligent identification, rock pressure