Prediction of Rock Mass Classification Using Machine Learning and the Q-System

• A machine learning model (Random Forest) accurately predicts Q-system rock mass class with 92.3% accuracy, using only five easily obtainable parameters. • RQD and joint spacing are the most critical factors influencing rock mass classification, as identified by feature importance analysis. • The proposed model offers a rapid and objective alternative to traditional Q-system assessments, reducing subjectivity and field investigation time. • The approach can be integrated into early-stage design and real-time tunneling operations to enhance safety and cost-effectiveness.