Estimation of the TBM advance rate under hard rock conditions using XGBoost and Bayesian optimization

The advance rate (AR) of a tunnel boring machine (TBM) under hard rock conditions is key parameter in the successful implementation tunneling engineering. In this study, we improved accuracy prediction models by employing hybrid model extreme gradient boosting (XGBoost) with Bayesian optimization (BO) to TBM AR. To develop proposed models, 1286 sets data were collected from Peng Selangor Raw Water Transfer project Malaysia. database consists mass and intact features, including rating, quality designation, weathered zone, uniaxial compressive strength, Brazilian tensile strength. Machine specifications, revolution per minute thrust force, considered predict accuracies predictive examined using root mean squares error (RMSE) coefficient determination (R2) between observed predicted yield five-fold cross-validation procedure. Results showed that BO algorithm can capture better hyper-parameters for XGBoost than default model. robustness generalization BO-XGBoost yielded prominent results RMSE R2 values 0.0967 0.9806 (for testing phase), respectively. demonstrated merits addition, variable importance through mutual information tests was applied interpret parameters have greatest impact as compared material properties.