Prediction of incompetent rock mass based on characteristic parameters and neural network

Abstract

TBM is famous for its safety and efficiency of tunnel boring, but TBM is also easy to encounter geological disasters in incompetent rock mass with its poor geological adaptability. Unfortunately, due to the shelter from cutterhead and shield, it’s hard to diagnose and identify the quality of rock mass. Based on this, this paper uses the torque penetration index TPI to predict incompetent rock mass, the main steps can be summarized as follows: firstly, we use linear fitting to calculate TPI, secondly, build the prediction model respectively with II-III grade of surrounding rocks and IV-V grade of surrounding rocks based on neural network, thirdly, predict the current TPI by the TPI of last boring cycles, finally, calculate the predicted torque based on the penetration and predicted TPI, then the incompetent rock mass is quickly identified through the error analysis between actual torque and predicted torque