Abstract: There is a lack of prediction and earlywarning methods for surrounding rock quality and collapse risk when using a tunnel boring machine(TBM). TBM boring big data are mined to address this, and the collapse data are analyzed to provide auxiliary judgment criteria for potential collapse risks. Redundant and continuous raw data are preprocessed to obtain highquality analytical data. A method for calculating characteristic rock parameters based on parameter correlation analysis is proposed. The rationality and applicability of these parameters are demonstrated through simplified theoretical derivations, indoor test results, and onsite boring tests. Finally, based on actual TBM collapse cases, the correlation between the characteristic parameters and surrounding rock geological conditions is analyzed, providing a basis for the rapid assessment of collapse risk. The results show that the characteristic parameters of the surrounding rock obtained from processing and analysis of the TBM boring data reflect the quality of the surrounding rock, and their values are positively correlated with the quality of the surrounding rock. When their values significantly decrease and the variation exceeds 69.2%, the current boring cycle is highly prone to potential collapse risk.

Key words: waterdiversion tunnel, tunnel boring machine, big data, surrounding rock quality, characteristic parameters; collapse analysis