LAUSR

Abstract In hard rock TBM tunneling, the loss caused by disc cutter wear accounts for a large proportion of time and cost for the entire project. However, existing disc cutter wear prediction models mainly focus on predicting cutter consumption before construction and cannot predict the wear of each disc cutter. Moreover, the accurate rock parameters required in these models are challenging to obtain. Hence, these models are not capable of determining which cutter on cutterhead should be replaced during construction. To solve the problems mentioned above, this paper presents a novel field parameters-based method for estimating the wear of each disc cutter in real-time. The proposed method is implemented through the following steps. To begin with, a new health index is constructed and defined as the ratio of the rolling distance of a cutter in a small excavated section to its maximum rolling distance. Then, specific field parameters related to the new health index are analyzed and selected. Thereafter, the mapping model between the new health index and the specific field parameters is established based on a one-dimensional convolutional neural network. Finally, on the basis of the established model, the estimated health indices corresponding to all excavated sections of a disc cutter are accumulated to obtain its health status. The field data obtained from Mumbai metro tunnel was utilized to verify the effectiveness of the proposed method, which demonstrates that the proposed method can estimate the wear of each disc cutter in real-time with average accuracy as high as 87.8% on the test set. Therefore, the proposed method is capable of significantly reducing the time and cost of cutter inspection, replacement, and repair for TBM, thereby improve tunneling efficiency and reduce construction cost.