Scholarlink

<p>The stability of pillars in underground mining is key to safety and efficient operations. Predicting pillar stability is a complex task, influenced by dynamic and nonlinear factors, which necessitates predictive methods with higher accuracy and practicality. This study introduces two innovative stability assessment methods—the hard-rock pillar stability index (HPSI) and a hybrid gene-expression programming (HGEP) model—developed using 195 pillar cases from multiple hard-rock mines. The HPSI provides a quantitative scoring indicator for stability, while HGEP offers an explicit discriminant function to capture complex nonlinear interactions. Validation results demonstrate the proposed models' superiority over traditional approaches. The HGEP model achieved an accuracy improvement of approximately 15%, reaching 88.4%, and both models exhibited strong generalization performance on external datasets. Additionally, SHapley Additive exPlanations (SHAP) analysis revealed the critical influence of geometric and stress parameters, enhancing the models' transparency and interpretability. Beyond technical contributions, the outcomes of this study hold significant practical relevance. The HPSI and HGEP models provide reliable and interpretable tools for assessing hard-rock pillar stability, supporting safer and more efficient underground mining operations.</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

Methodology for Constructing Explicit Stability Formulas for Hard Rock Pillars: Integrating Data-Driven Approaches and Interpretability Techniques

  • 期刊论文

  • 发表日期：

    2025年2月5日

查看全文
  • Yingui Qiu,
  • Jian Zhou

摘要

The stability of pillars in underground mining is key to safety and efficient operations. Predicting pillar stability is a complex task, influenced by dynamic and nonlinear factors, which necessitates predictive methods with higher accuracy and practicality. This study introduces two innovative stability assessment methods—the hard-rock pillar stability index (HPSI) and a hybrid gene-expression programming (HGEP) model—developed using 195 pillar cases from multiple hard-rock mines. The HPSI provides a quantitative scoring indicator for stability, while HGEP offers an explicit discriminant function to capture complex nonlinear interactions. Validation results demonstrate the proposed models' superiority over traditional approaches. The HGEP model achieved an accuracy improvement of approximately 15%, reaching 88.4%, and both models exhibited strong generalization performance on external datasets. Additionally, SHapley Additive exPlanations (SHAP) analysis revealed the critical influence of geometric and stress parameters, enhancing the models' transparency and interpretability. Beyond technical contributions, the outcomes of this study hold significant practical relevance. The HPSI and HGEP models provide reliable and interpretable tools for assessing hard-rock pillar stability, supporting safer and more efficient underground mining operations.