Improved SVR Method for Predicting the Cutting Force of a TBM Cutter Using Linear Cutting Machine Test Data

This research introduces a support vector regression (SVR) method to predict the cutting forces acting on the constant cross section (CCS) disc cutter, including the normal force (FN) and rolling force (FR), based on linear cutting machine (LCM) test data. To improve the prediction effect, an improved SVR-Outlier Detection (SVR-OD) method and an Additional Input Variable (AIV) method are proposed. After removing the outliers, 148 typical LCM test samples form the training set. Here, 70 samples from the Hangzhou No.2 Water Supply Channel constitute the test set. The prediction results show that the Root-mean-squared Relative Error (RMRE) values of the normal force and rolling force are 19.5% and 24.8%, respectively, and the corresponding determination coefficients are 0.845 and 0.807, respectively. For the prediction of the peak cutting force with an important reference to tunnel boring machine (TBM) design, the proportions of samples with an Absolute Relative Error (ARE) value of less than 20% for FN and FR are 9/10 and 7/7, respectively. The above prediction results are better than those of the common SVR method; thus, the developed method can effectively simulate the cutting force required by a rock mass with good integrity. The cutting force prediction using LCM test data is feasible and practical. In addition, the comparison of the prediction results between the improved SVR and common SVR methods shows that the improved SVR-OD and AIV methods play an active role in improving the prediction accuracy of the SVR method.