Civil and Geoengineering Letters Editorial Office info@jcivilgeo.com mEDRA 20260525 DOI Registration 01

06 10.22034/CGEL.3.1.e100054 https://jcivilgeo.com/article/e100054 Civil and Geoengineering Letters mEDRA 01 01 3115-8129 Civil and Geoengineering Letters Editorial Office IR 01 3115-8129 JB 3 1 01 202605 1 1 A01 Hamed Mahmodi Mahmodi, Hamed Hamed Mahmodi 01 eng 20260525 03 00 Surface subsidence is a critical geotechnical concern in shallow foundation systems, especially in urban environments with heterogeneous soil conditions and variable groundwater levels. Excessive or uneven settlement can compromise structural integrity, reduce service life, and increase maintenance costs. Therefore, timely identification and prediction of surface subsidence risk are essential in foundation design and site planning. In this study, a data-driven approach using Multi-Layer Perceptron (MLP) neural networks is proposed to predict the risk of surface subsidence associated with shallow foundations. A comprehensive dataset was compiled, incorporating soil physical and mechanical properties, foundation geometry, groundwater depth, load characteristics, and historical subsidence measurements from multiple construction sites. The MLP model was trained using supervised learning techniques and validated with a test dataset to ensure accuracy and generalization capability. Performance metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R² were used to evaluate the model's predictive capability. The results show that the MLP model successfully captures the complex nonlinear interactions among the influencing factors and achieves high predictive accuracy. This research demonstrates the effectiveness of MLP neural networks as a practical tool for geotechnical risk assessment and decision-making. The proposed method can support engineers and planners in identifying high-risk zones early in the design process and implementing appropriate mitigation strategies. 20 Subsidence; Shallow foundation; Multi-layer perceptron (MLP); Settlement prediction; Machine learning 01 00 20260119 02 00 20260331 2026 Hamed Mahmodi