Deep underground, tunnel-boring machines collect extensive operational data like thrust, torque, and penetration, monitored almost in real-time. Yet, geological information ahead of the tunnel remains limited. This creates a unique AI challenge: managing one highly detailed data source alongside a sparse one. The patent tackles this by predicting disc cutter wear in shield tunneling machines, using geological data to influence how dense operational data is processed and compressed. This approach connects two different data scales, the geological conditions and the machine responses, improving assessments of cutter wear, which affects tunneling efficiency and maintenance. Such predictions enable engineers to identify when cutters need servicing, moving beyond fixed schedules or raw data streams. This highlights a key AI principle: more data isn’t always better if datasets have incompatible resolutions. Often, the key lies in reshaping data based on the physical context, with geological insights guiding data representation for AI analysis.
Teaching a tunnelling machine to listen to the ground
Patent number: CN 119623293 A
Inventor(s): L. Bai, D. Mo, N. Wu, and W. Huang
Citation: L. Bai, D. Mo, N. Wu, and W. Huang, “用于滚刀磨损预测的盾构机运行数据处理方法、装置、计算机设备、可读存储介质和程序产品 [Shield-machine operating-data processing method, apparatus, computer device, readable storage medium and program product for cutter-wear prediction],” China Patent Application CN 119623293 A, Mar. 14, 2025.
