Prediction of TBM Tunneling Parameters Based on RF

With the continuous proficiency of TBM technology, predicting and optimizing TBM performance becomes increasingly important. Due to the complexity and difficulty in predicting geological conditions, it is preferred to adjust key excavation parameters based on measured data. This paper takes the TBM3 bid section excavation data of Jilin YS project as the research object, introduces knowledge driven vectors TPI and FPI, integrates time series into the random forest algorithm (RF), and utilizes the TBM circular tunneling process start-up loading and stably advancing segments excavation parameters to establish a cutter head thrust and torque prediction model. The results show that the introduction of TPI and FPI keeps the input and output of the model consistent, achieving the goal of simplifying the algorithm; The evaluation indicators R and R
²
for predicting cutter head thrust and cutter head torque are both above 0.8, with high prediction accuracy.