Compressive Strength Prediction of Cemented Backfill Containing Phosphate Tailings Using Extreme Gradient Boosting...

Area of Science:

• Geotechnical Engineering
• Materials Science
• Artificial Intelligence

Background:

• Unconfined compressive strength (UCS) is critical for cemented backfill, typically determined by mechanical tests.
• Accurate UCS prediction is essential for backfill performance and safety.

Purpose of the Study:

• To develop a hybrid machine learning model for predicting the UCS of cemented backfill.
• To optimize the extreme gradient boosting (XGBoost) model using the whale optimization algorithm (WOA).

Main Methods:

• Input variables included PT proportion, OPC proportion, FA proportion, solid concentration, and curing age.
• A hybrid WOA-XGBoost model was developed and compared with original XGBoost, PSO-XGBoost, and Decision Tree (DT) models.
• Performance was evaluated using root mean square error (RMSE), coefficient of determination (R²), and mean absolute error (MAE).

Main Results:

• The WOA-XGBoost model achieved superior prediction accuracy with RMSE=0.241, R²=0.967, and MAE=0.184.
• WOA-XGBoost outperformed original XGBoost (RMSE=0.426), PSO-XGBoost (RMSE=0.316), and DT (RMSE=0.464).

Conclusions:

• The WOA-XGBoost model demonstrates enhanced prediction accuracy for cemented backfill UCS.
• This hybrid model offers a fast and accurate alternative to traditional mechanical tests for UCS prediction.