Deep learning based operating parameter decision-making method for optimal penetration rate

Abstract

Abstract TBM has become one of the most important equipment for underground excavation due to its effectiveness on both time and economic. However, the efficiency and safety of TBM excavation is highly dependent on the driver's experience. This study proposes a novel deep learning-based intelligent decision-making model for TBM operating parameters. The model consists of a deep learning based TBM operating parameters and TBM performance mapping algorithm and operating parameters decision-making method. The proposed model takes the historical mechanical data of the TBM as input and can suggest optimal operating parameters for TBM excavation in real time. The mapping algorithm can predict the performance of the TBM at any given operating parameter with an average percentage error accuracy of 4.06% for PR and 4.65% for torque. The results show that the proposed decision-making method for operating parameters can increase the PR for about 10% in high excavatability regions and increase about 3% PR in low excavatability regions while reduce energy cost and presumably reduce cutter weariness. This study also analyzes the logic behind the selection of operating parameters by the TBM driver and why the proposed method can achieve better TBM performance than manually driven TBMs.