| Summary: | Natural fiber-reinforced concrete (NFRC) is gaining attention for its sustainability, cost-effectiveness, and biodegradability, making it a promising material for construction and repair. Betel nut husk fiber (BNHF) is being incorporated into concrete due to its eco-friendly, nontoxic, and biodegradable properties. However, traditional methods for measuring slump, compressive strength (CS), and split tensile strength (STS) are often time-consuming, labor-intensive, and costly. To address this, machine learning (ML) models offer an efficient alternative for predicting these properties, enabling faster and more economical adjustments to BNHF-reinforced concrete mixes. This study explores the use of five ensemble ML models: Random Forest (RF), XGBoost, CatBoost, AdaBoost, and LightGBM to predict slump, CS, and STS. Model performance was evaluated using five metrics: coefficient of determination (R²), root mean square error (RMSE), normalized root mean square error (NRMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The results showed that the CatBoost model performed best in predicting slump and CS, with R² values of 0.989 and 0.918, and RMSEs of 2.116 mm and 1.986 MPa, respectively. For STS, XGBoost outperformed other models, achieving an R² of 0.862 and an RMSE of 0.406 MPa on the test set. SHapley Additive exPlanations (SHAP) analysis indicated that BNHF percentage and fiber content had the greatest influence on slump, while curing time was the most significant factor affecting both CS and STS. The findings demonstrate that CatBoost and XGBoost can accurately predict the mechanical properties, offering a practical alternative to extensive laboratory testing and enabling time and cost savings in construction.
|
|---|