Prediction of Grouting Volume for Shield Tunnel Construction Based on Integrated Learning Model

Accurately forecasting grouting volume is crucial in shield tunneling to ensure tunnel stability and construction safety. This paper proposes a grouting volume prediction method based on an Integrated Learning Model (ILM), with Bayesian Optimization (BO) used to optimize the model’s hyperparameters. The ILM is constructed by combining three base learners: Random Forest (RF), Support Vector Machine (SVM), and Gradient Boosting Decision Tree (XGBoost), with the optimal model serving as the meta-learner. The prediction model is built using real-world shield construction data and its performance is evaluated through comparative analysis. The results show that the ILM achieves an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> value of 0.92, with optimal performance in RMSE and MAE, significantly outperforming individual models. Furthermore, feature importance analysis reveals that face earth pressure and excavation rate are the key factors influencing grouting volume, offering valuable insights for optimizing grouting strategy during construction.