Hydraulic Performance Modeling of Inclined Double Cutoff Walls Beneath Hydraulic Structures Using Optimized Ensemble...

This study investigates the effectiveness of inclined double cutoff walls installed beneath hydraulic structures by employing five machine learning models: Random Forest (RF), Adaptive Boosting (AdaBoost), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Categorical Boosting (CatBoost). A comprehensive dataset of 630 samples was gathered from previous studies, including key input variables such as the relative distance between the cutoff wall and the structure's apron width (L/B), the inclination angle ratio between downstream and upstream cutoffs (θ₂/θ₁), the depth ratio of downstream to upstream cutoff walls (d₂/d₁), and the relative downstream cutoff depth to the permeable layer depth (d₂/D). Outputs considered were the relative uplift force (U/U_o), the relative exit hydraulic gradient (i_R/i_Ro), and the relative seepage discharge per unit structure length (q/q_o). The dataset was split with a 70:30 ratio for training and testing. Hyperparameter optimization was conducted using Bayesian Optimization (BO) coupled with five-fold cross-validation to enhance model performance. Results showed that the CatBoost model demonstrated superior performance over other models, consistently yielding high R² values, specifically surpassing 0.95, 0.93, and 0.97 for U/U_o, i_R/i_Ro, and q/q_o, respectively, along with low RMSE scores below 0.022, 0.089, and 0.019 for the same variables. A feature importance analysis is conducted using SHapley Additive exPlanations (SHAP) and Partial Dependence Plot (PDP). The analysis revealed that L/B was the most influential predictor for U/U_o and i_R/i_Ro, while d₂/D played a crucial role in determining q/q_o. Moreover, PDPs illustrated a positive linear relationship between L/B and U/U_o, a V-shaped impact of d₂/d₁ on i_R/i_Ro and q/q_o, and complex nonlinear interactions for θ₂/θ₁ across all target variables. Furthermore, an interactive Graphical User Interface (GUI) was developed, enabling engineers to efficiently predict output variables and apply model insights in practical scenarios.