Toward state-of-the-art techniques in predicting and controlling slope stability in open-pit mines based on limit equilibrium analysis, radial basis function neural network, and brainstorm optimization

This study aims to propose state-of-the-art techniques in predicting and controlling slope stability in open-pit mines based on limit equilibrium analysis, artificial neural networks, and optimization algorithms. Accordingly, the simplified Bishop method was used to analyze the slope stability of an open-pit coal mine through the limit equilibrium analysis method. Various rock mass properties and the geometrical parameters of the slopes were considered, such as bench height, slope angle, unit weight, cohesion, and friction angle. Finally, 495 cases were analyzed to compute the factor of safety (FOS). Subsequently, the radial basis function neural network (RBFNN) model was applied to predict FOS. In order to optimize the RBFNN model, the brainstorm optimization (BSO) algorithm was applied to train the RBFNN model, named as BSO-RBFNN model. The genetic algorithm (GA)-RBFNN, RBFNN (without optimization), and multiple layers perceptron (MLP) neural network were also developed to predict FOS and compared with the proposed BSO-RBFNN model as part of the study. The results revealed that the optimization of the BSO algorithm and RBFNN model provided a state-of-the-art technique (i.e., BSO-RBFNN) for predicting and controlling slope stability with high accuracy (i.e., mean absolute error (MAE) = 0.047, root-mean-squared error (RMSE) = 0.057, determination coefficient (R2) = 0.929, variance accounted for (VAF) = 92.948), and reliability (i.e., absolute error of 5.89% for 80% of cases in practice). Comparisons also indicated that the proposed BSO-RBFNN model is the most dominant model for predicting slope stability in this study (i.e., MAEGA-RBFNN = 0.048, RMSEGA-RBFNN = 0.060, R2GA-RBFNN = 0.927, VAFGA-RBFNN = 92.534; MAERBFNN = 0.064, RMSERBFNN = 0.081, R2RBFNN = 0.925, VAFRBFNN = 89.189; MAEMLP = 0.065, RMSEMLP = 0.081, R2MLP = 0.873, VAFMLP = 85.724). Furthermore, the slope angle and bench height should be taken into account to control slope stability in practical engineering based on the proposed BSO-RBFNN model.