
科学技术与工程

中国技术经济学会
Intelligent Prediction and Visual Analysis of Slope Stability via Ensemble Learning Combined with Multialgorithm Collaborative Optimization and Dual Interpretability Framework
Shun-chuan WU , Yi-fan CAO , Long-qiang HAN , Shi-jiang PU , Yu-nuo GAO , Shi-gui HUANG
Science Technology and Engineering ›› 2026, Vol. 26 ›› Issue (31) : 13423 -13436.
Slope stability analysis is regarded as a critical issue in geotechnical engineering. However, inherent limitations exist in traditional evaluation methods, and complex nonlinear relationships among influencing factors cannot be fully characterized. Therefore, an intelligent prediction model for slope stability based on the integration of multiple optimization algorithms was proposed to improve prediction accuracy and model interpretability. First, eight widely used machine learning models were compared, and the random forest model was selected as the optimal baseline model. Then, its hyperparameters were collaboratively optimized using the lung performance optimization algorithm, the golden sine algorithm, and the slime mould algorithm to further enhance predictive performance. Subsequently, an ensemble learning framework was constructed, whose hyperparameters were optimized with the successive halving gridsearch method. The ensemble model was compared with individual models, and it is indicated that superior performance in both accuracy and stability is achieved. To improve model interpretability, the SHAP(shapley additive explanations) and LIME(local interpretable model-agnostic explanations)methods were employed to quantify feature contributions to prediction results and to identify key influencing factors and their relative importance. It is shown that the optimized random forest model achieves an accuracy of 0.935 9, while better performance is achieved by the ensemble model, with an accuracy of 0.948 7 and favorable F1-score and precision. Finally, a software system integrating data visualization and slope stability prediction functions was developed. Efficient and reliable support was provided by the system for intelligent slope stability evaluation.
slope stability / optimization algorithms / ensemble learning / interpretability analysis / visualization system
/
| 〈 |
|
〉 |