Prediction of slope stability in open-pit mines using intelligent algorithms: SVM and RF optimized by genetic algorithm

Abstract

Monitoring and predicting slope stability in open-pit mines plays a critical role in enhancing safety, minimizing losses, and improving operational efficiency. Slope instability can lead to severe and often irreversible economic, human, and environmental consequences. Traditional stability analysis methods, such as limit equilibrium and numerical modeling, face limitations due to geometric simplifications, linear assumptions, and their inability to capture complex patterns—factors that reduce their effectiveness in real-world conditions. In recent years, machine learning approaches have emerged as powerful tools in geotechnical analysis. This study aims to predict the stability status of open-pit mine slopes using machine learning models, specifically Support Vector Machine (SVM) and Random Forest (RF). To improve the accuracy of these models, their parameters were optimized using a Genetic Algorithm (GA). The dataset used includes geotechnical and geometric features influencing slope stability, obtained from field investigations and documented sources. The results indicate that the RF–GA hybrid model outperforms the SVM–GA model, achieving 93% accuracy with an AUC of 0.93, compared to 86% accuracy and an AUC of 0.86 for the SVM–GA model. Moreover, the RF model demonstrated higher sensitivity in identifying stable slopes and reduced the number of false negatives. These findings highlight the strong potential of the RF–GA model in delivering reliable predictions and supporting decision-making in slope stability management. The integration of intelligent algorithms with local data offers a robust alternative to traditional methods in geotechnical engineering.