Weighted random forest prediction model for TBM advance rate considering uncertainty

There are many factors affecting TBM construction speed, and they have significant uncertainty. In view of the ambiguity of geological parameters, the rock classification system RMR, rock abrasiveness CAI and rock hardness H are used to measure the geological conditions. For the randomness of mechanical parameters in the construction process, the active control parameters such as TBM cutter head thrust TF and rotational speed RPM are used for analysis. At the same time, it is proposed to quantify the uncertainty of human factors by the proportion of downtime of other factors. Based on the measured data of the double-shield TBM construction in the water conveyance tunnel of the Lanzhou water source construction project, the prediction database of the TBM advance rate and the weighted random forest algorithm model considering uncertainty are established. In addition, other models such as random forest, support vector regression and BP neural network are used to verify the prediction accuracy of the proposed model. The results show that the error of the root mean square and the determination coefficient of the predicted results of the test set in the weighted random forest model are 1.59 and 0.97, respectively, and the prediction accuracy and reliability of the weighted random forest prediction model are better than those of the other three models. The model adopts the method of optimizing hyperparameters by assigning different weights, which has the advantages of high accuracy, difficult overfitting and better generalization capability and robustness.