Scholarlink

<p>Leveraging the superiority of data-driven methods for solving highly nonlinear prediction and multi-objective optimization problems, we constructed a dataset comprising 1,080 samples based on key geometric factors and mechanical parameters of mine slope stability. We subsequently developed an extreme gradient boosting regression model (XGBR) for slope stability prediction. Optimized using a meta-heuristic algorithm, the Cultural Algorithm (CA) based XGBR model performs best across multiple assessment dimensions. The model's interpretability was then analyzed using the SHAP method to quantify the impact of each input feature on stability. The CA-XGBR model, incorporating stability constraint equations, was integrated into the objective function of a multi-objective optimization problem and solved using the Non-dominated Sorting Genetic Algorithm with Elite Strategy (NSGA-II). The resulting intelligent analysis system for slope stability integrates prediction and optimization capabilities. A comparative analysis using a specific case study shows that the proposed method outperforms the traditional method, providing an innovative solution for geotechnical slope stability assessment and geometric parameter optimization.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interpretable slope stability evaluation and optimization method based on hybrid extreme gradient boosting regression

  • 期刊论文

  • 发表日期：

    2026年3月31日

查看全文
  • Liutao Wang,
  • Chao Zhang,
  • Yinxiang Cui,
  • Jin Hu

摘要

Leveraging the superiority of data-driven methods for solving highly nonlinear prediction and multi-objective optimization problems, we constructed a dataset comprising 1,080 samples based on key geometric factors and mechanical parameters of mine slope stability. We subsequently developed an extreme gradient boosting regression model (XGBR) for slope stability prediction. Optimized using a meta-heuristic algorithm, the Cultural Algorithm (CA) based XGBR model performs best across multiple assessment dimensions. The model's interpretability was then analyzed using the SHAP method to quantify the impact of each input feature on stability. The CA-XGBR model, incorporating stability constraint equations, was integrated into the objective function of a multi-objective optimization problem and solved using the Non-dominated Sorting Genetic Algorithm with Elite Strategy (NSGA-II). The resulting intelligent analysis system for slope stability integrates prediction and optimization capabilities. A comparative analysis using a specific case study shows that the proposed method outperforms the traditional method, providing an innovative solution for geotechnical slope stability assessment and geometric parameter optimization.