A new, fast, and accurate algorithm for predicting soil slope stability based on sparrow search algorithm-back propagation

Slope stability prediction is one of the most essential and critical tasks in mining and geotechnical projects. A fast and precise slope stability prediction is crucial for safe operations and cost-effective slope maintenance. In this work, a back propagation (BP) neural network based on a sparrow search algorithm (SSA) is developed to predict the slope safety coefficients by using five input features, including unit weight (γ), cohesion (c), friction angle (φ), slope angle (α), and slope height (H). The proposed model is trained and simulated using 55 data samples. The regression coefficient of the proposed SSA-BP neural network is 0.9405, with a mean relative error (MAE) of 0.1684. Compared with fusion algorithms, such as Ridge Regression (RR), Decision Tree (DT), Random forest (RF), Support Vector Regression (SVR), and Light Gradient Boosting Machine (lightGBM), the proposed method yields more accurate and robust prediction results. Furthermore, a multivariate function relationship between the slope safety coefficient and the five variables is constructed based on the relationship between five independent input variables and the variation of the safety coefficient. The proposed method introduces a novel approach for calculating the slope safety coefficient.