Prediction of slope failure in open-pit mines using a novel hybrid artificial intelligence model based on decision...

Area of Science:

• Geotechnical Engineering
• Artificial Intelligence in Mining
• Computational Geology

Background:

• Slope stability is critical in open-pit mining operations.
• Accurate prediction of slope failures is essential for safety and economic viability.
• Existing AI models have limitations in slope stability analysis.

Purpose of the Study:

• To develop a novel and highly accurate artificial intelligence model for predicting slope failures in open-pit mines.
• To introduce the M5Rules-GA hybrid model for enhanced slope stability estimation.
• To evaluate the performance of the M5Rules-GA model against established AI techniques.

Main Methods:

• Developed a hybrid M5Rules-GA model combining M5Rules algorithm and Genetic Algorithm (GA).
• Modeled 450 slope observations from an open-pit mine in Vietnam using Geo-Studio software.
• Compared M5Rules-GA with Artificial Neural Networks (ANN), Support Vector Regression (SVR), and other hybrid models (FFA-SVR, ANN-PSO, ANN-ICA, ANN-GA, ANN-ABC).
• Evaluated model performance using determination coefficient, variance account for, and root mean square error.

Main Results:

• The proposed M5Rules-GA model demonstrated robust performance in slope stability analysis.
• M5Rules-GA achieved higher accuracy and reliability compared to other evaluated AI models.
• Performance metrics confirmed the superiority of the M5Rules-GA approach.

Conclusions:

• The M5Rules-GA model is a powerful and accurate tool for open-pit mine slope failure prediction.
• This hybrid AI approach offers significant improvements over existing methods for slope stability assessment.
• The findings contribute to advancing AI applications in geotechnical engineering and mine safety.