Scholarlink

<p>Rock-mass classification systems provide a powerful approach for geomechanical characterization and risk assessment but face limitations in assessing roof water inrush risk in coal mines. A mechanism-oriented classification system is proposed for roof water inrush risk assessment in coal mines based on 81 inrush cases. The system first categorizes the inrush mechanism as either water-conducting fracture inrush (WCFI) or burst inrush from high-level separation layers (BIHSL) based on a discriminant criterion. A tailored evaluation index system is then applied, which includes a dedicated risk indicator specific to BIHSL. To construct a robust scoring table resilient to the limitations of a relatively small dataset, we integrated expert knowledge through the analytic hierarchy process (AHP) to establish stable indicator weights, thereby reducing reliance on purely data-driven fitting. Subsequently, the XGBoost algorithm coupled with SHAP (SHapley Additive exPlanations) analysis was used to objectively determine the grading thresholds, a process that leverages model interpretability rather than overwhelming data volume. This AHP-XGBoost-SHAP based scoring framework was applied to 81 documented cases. The resulting composite risk scores demonstrated a strong quantitative correlation with actual water inflows, achieving high goodness-of-fit (R<sup>2</sup> = 0.78 for WCFI under a linear model; R<sup>2</sup> = 0.90 for BIHSL under a polynomial model). This validates the system's capability for both qualitative risk classification into five levels (I-V) and for accurate quantitative inflow prediction. The final risk maps, defining permitted, restricted, and prohibited mining zones, provide a practical and mechanistic tool for groundwater hazard management.</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

A Classification System for the Risk Assessment of Roof Water Inrush of Coal Mines

  • 期刊论文

  • 发表日期：

    2026年4月10日

查看全文
  • Mengnan Liu,
  • Wei Qiao,
  • Xiangsheng Meng,
  • Peichao Feng,
  • Xianggang Cheng

摘要

Rock-mass classification systems provide a powerful approach for geomechanical characterization and risk assessment but face limitations in assessing roof water inrush risk in coal mines. A mechanism-oriented classification system is proposed for roof water inrush risk assessment in coal mines based on 81 inrush cases. The system first categorizes the inrush mechanism as either water-conducting fracture inrush (WCFI) or burst inrush from high-level separation layers (BIHSL) based on a discriminant criterion. A tailored evaluation index system is then applied, which includes a dedicated risk indicator specific to BIHSL. To construct a robust scoring table resilient to the limitations of a relatively small dataset, we integrated expert knowledge through the analytic hierarchy process (AHP) to establish stable indicator weights, thereby reducing reliance on purely data-driven fitting. Subsequently, the XGBoost algorithm coupled with SHAP (SHapley Additive exPlanations) analysis was used to objectively determine the grading thresholds, a process that leverages model interpretability rather than overwhelming data volume. This AHP-XGBoost-SHAP based scoring framework was applied to 81 documented cases. The resulting composite risk scores demonstrated a strong quantitative correlation with actual water inflows, achieving high goodness-of-fit (R2 = 0.78 for WCFI under a linear model; R2 = 0.90 for BIHSL under a polynomial model). This validates the system's capability for both qualitative risk classification into five levels (I-V) and for accurate quantitative inflow prediction. The final risk maps, defining permitted, restricted, and prohibited mining zones, provide a practical and mechanistic tool for groundwater hazard management.