Predicting the Stability of Rock Slopes in the Presence of Diverse Joint Networks and External Factors Using Machine Learning Algorithms

The presence of joints in rocks significantly impacts the mechanical behavior and stability of a slope. A better comprehension of the relationship between jointed rock masses and slope stability has been made possible by recent advances in machine learning algorithms and numerical modelling. The purpose of this research is to predict the stability of slopes in the presence of different types of joints (parallel deterministic, cross jointed, Baecher, Veneziano, and Voronoi) with the help of classification-based machine learning algorithms. In order to achieve this goal, 40,290 different cases have been utilized, following numerical simulation using shear strength reduction (SSR) technique in RS2. Geomechanical properties, parameters defining slope geometry, structural properties of joints including properties of filling materials, and the influence of certain external factors have been considered. For these datasets, classification algorithms such as random forest, k-nearest neighbor, support vector machine, logistic regression, decision tree, and Naive Bayes have been utilized. Additionally, the synthetic minority oversampling technique (SMOTE) has been implemented in order to address imbalanced class problems. The results exhibit an encouraging level of accuracy, with random forest and decision tree both achieving 0.98 as an overall accuracy.