Prediction of TBM advance rate based on Bootstrap method and SVR-ANN algorithm

The reasonable prediction and evaluation of advance rate is related to the success and benefit of TBM construction. The existing prediction models for TBM advance rate mostly use parameters of rock mass and TBM tunneling to predict the instantaneous/average advance rate. To solve the problem that these models do not consider the influences of uncertainty and risk during TBM tunneling process, a prediction model for TBM advance rate based on the Bootstrap method and SVR-ANN algorithm is proposed by introducing the idea of interval prediction. Based on the project of Lanzhou water sources water conveyance tunnel constructed by double-shield TBM, the shortcomings of some single input parameters are analyzed, and the rationality of two selected comprehensive parameters, namely rock mass quality classification index (RMR) and TBM working condition class (TWCR), is pointed out. In addition, the validity of the developed interval prediction model for TBM advance rate is verified. The results show that the developed interval prediction model for TBM advance rate provides relatively accurate point prediction results and constructs a clear and reliable AR prediction interval to cover the actual TBM advance rate completely. The MPIW of model test set at confidence levels of 90% and 95% is 9.84, 11.73 m/d, respectively. With the improvement of the confidence level, the uncertainty that can be contained in the prediction interval also increases. Moreover, the possible risk of TBM tunneling process and the abnormal interval width confirm each other, which verifies that the interval prediction model can quantitatively explain the characteristics of uncertainty in the construction process. The research results may provide a new idea for the forecasting of TBM tunneling efficiency, the estimation of construction schedule as well as the optimization of tunneling parameters.