نوع مقاله : مقاله پژوهشی
عنوان مقاله English
نویسندگان English
Abstract Accurate prediction of surface settlement induced by urban tunneling is essential for mitigating geotechnical risks and protecting adjacent structures. This study investigates the prediction of maximum surface settlement caused by Earth Pressure Balance (EPB) tunneling in Mashhad Urban Railway Lines 2 and 3 using machine learning techniques. A database consisting of 292 samples was compiled from geotechnical investigations, geometric characteristics, and Tunnel Boring Machine (TBM) operational parameters, and 20 key input variables were selected. Three predictive models, including Artificial Neural Network (ANN), Deep Neural Network (DNN), and Random Forest (RF), were developed and compared. The results indicated that the DNN model with three hidden layers achieved the best performance, yielding a coefficient of determination (R²) of 0.9896 and the lowest mean squared error. Sensitivity analysis based on the Garson and Olden methods was further conducted to identify the most influential parameters affecting settlement behavior. The novelty of this research lies in the development of a data-driven model based on real-world tunneling data from the Mashhad Metro, the integrated use of geotechnical and TBM operational parameters, and the quantitative comparison of machine learning predictions with the conventional Peck empirical method. The proposed DNN model demonstrated substantially higher accuracy than the Peck method and can serve as a reliable tool for settlement risk assessment and management in urban tunneling projects.
کلیدواژهها English
