Research on intelligent identification and resource utilization of TBM muck in breezy granite strata

Abstract

Abstract Tunnel boring machines (TBMs) excavating breezed granite strata produce a lot of muck, including rock powder and gravel. To promote waste recycling and environmentally friendly muck utilization, a series of studies was conducted on the intercity railway section from ShenZhen Airport East Station to HuangMaBu Station in China. First, the muck image dataset was obtained using a camera, and the SAM-YOLO neural network model was adopted to intelligently identify the features of the particle size and content of the muck. Subsequently, the muck was used as an alternative material for the backfill slurry raw material, and the optimal slurry ratio was determined through laboratory tests. Finally, a new backfill slurry was applied on-site. The results demonstrated that the established SAM-YOLO model can accurately and quickly identify muck characteristics, which can be roughly classified by particle size into three categories: rock powder (0-5 mm), gravel (5-10 mm), and crushed rock (10-31.5 mm). Rock powder was utilized to make the monomortar, gravel was used for backfilling, and the crushed rock was broken again to gravel particle size to be used for backfilling. The laboratory test and field application results indicated that the new monomortar can fulfil engineering requirements after the backfilling of gravel, and the optimal water, cement, and rock powder mix ratio is 0.9:1.0:1.0. This research thus provides effective guidance for the utilization of muck in similar projects.