Abstract: To predict the rock mass class of a shaft boring machine (SBM) in real-time.This study proposed a SBM rock mass classification prediction model based on the BO-LightGBM algorithm.Collecting operational data from the Volcano Tunnel’s SBM and on-site geological exploration data,a combined knowledge-driven and data-driven approach was employed to select 31 200 sets of real-time data across five feature parameters as input features.The Shapley Additive Explanations (SHAP) method was used to analyze the interpretability of the algorithm.The prediction results show that the BO-LightGBM model has a better performance than Bayesian optimization random forest (RF) model and the support vector machine (SVM) model,the Bo-LightGBM model has an accuracy of 94%.The prediction time consumed of Bo-LightGBM model for the validation set is 0.61s,enabling real-time prediction of rock mass classification.The accuracy and interpretability of the proposed algorithm were verified through model performance comparison and SHAP method analysis.