Delineating homogeneous zones for rock joint wall mechanical properties in open-pit mine slope based on a multi-indicator stacked generalization model.

Rock joints significantly influence slope stability, making it essential to delineate homogeneous zones. Traditional empirical methods are often subjective and ineffective when the mechanical properties of the joint walls are similar. To address this limitation, this study introduces a data-driven Stacked Generalization (SG) model that captures relationships between data points. Based on this model, a method for homogeneous zones division is proposed, incorporating five key indicators representing the mechanical properties of joint walls as feature parameters: joint wall compressive strength, disintegration resistance, wave velocity ratio, weathering coefficient, and linear density of the rock joints. These parameters can be quantified through field measurements and laboratory testing of rock joints that govern slope stability. The preprocessing stage integrates quadratic polynomial feature expansion and mutual information analysis to identify and select significant nonlinear relationships among these indicators. A case study demonstrates that the SG model achieves 94% balanced accuracy, outperforming traditional classifiers. The classification results of the model output are mapped onto an engineering geological plan of the study area, dividing it into four homogeneous zones. This approach provides an effective framework for delineating homogeneous zones of joint wall mechanical properties in open-pit mines and assessing slope stability under complex geological conditions.