Machine Learning-Based Prediction and Optimization of Slurry Shield’s Key Tunneling Parameters

Investigating the impact of key tunneling parameters, such as cutter￣head rotation speed, main thrust, and cutter-head torque, on the slurry support effect at the tunnel face and energy consumption during slurry shield construction is a crucial requirement to ensure efficient and rapid tunneling while minimizing the shield’ s mechanical losses. The tunneling parameters from the shield tunneling project of Jinan East Line Tunnel across Yellow River were used to calculate the field penetration index(FPI)and the torque penetration index(TPI)for each tunneling ring. The tunneling parameter set was divided into the optimal data set and the data set to be optimized using the excavation specific energy, and the prediction models of key tunneling parameters were established based on the support vector regression and artificial neural network methods respectively. The results showed that FPI and TPI can effectively describe the homogeneity of the excavated strata. The shield’ s excavation specific energy is log-normally distributed, which can be used to describe the shield’s excavation working condition and assess the configuration level among the slurry shield’ s tunneling parameters. The artificial neural network prediction model is suitable for optimizing the cutter-head rotation speed and the shield’s jacking force when the energy consumption level of shield tunneling fluctuates significantly in the homogeneous strata. © 2023 Northeastern University. All rights reserved.