Advanced hybrid machine learning models with explainable AI for predicting residual friction angle in clay soils.

Accurately estimating the residual strength friction angle of clay soils is vital for the design and stability evaluation of geotechnical structures such as slopes, retaining walls, and foundations, especially in regions susceptible to landslides and ground instability. Traditional methods for determining the residual strength friction angle are often labor-intensive, time-consuming, and costly. This study explores three advanced hybrid machine learning models: Gradient Boosting Neural Network (GrowNet), Reinforcement Learning Gradient Boosting Machine (RL-GBM), and a Stacking Ensemble to predict the residual friction angle of clay soils, addressing a critical gap in current predictive methodologies. The research utilized a carefully harmonized dataset of 400 samples from global studies, considering various soil parameters (liquid limit, plasticity index, change in plasticity index, and clay fraction) as inputs for the prediction models. Various evaluation metrics, including coefficient of determination (R(2)), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE), were used to assess model performance. The results show that GrowNet achieved the largest R(2) of 0.94 and the lowest RMSE and MAE values of 1.87 and 1.17 on the testing dataset, representing a substantial improvement over traditional empirical correlations and previous machine learning approaches. To address model transparency, Explainable Artificial Intelligence (XAI) techniques, including SHAP (Shapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations), were applied to the GrowNet model. These techniques consistently identified Clay Fraction as the most influential variable, followed by Plasticity Index. Integrating high-performing machine learning models with interpretability tools significantly improves the accuracy and reliability of residual friction angle predictions, offering practical value for geotechnical engineering applications.