Scholarlink

<p>A comprehensive understanding of water-conducting fracture zone development is crucial for ensuring coal mining safety and groundwater resource management. This study introduces an XGBoost model, optimized by the sparrow search algorithm (SSA), to predict fracture zone heights using a nationwide database of 114 measurement samples. The Shapley additive explanations (SHAP) model is applied to assess factor influence, while comparative analyses are conducted using support vector machine (SVM) and random forest (RF) models. To validate its effectiveness, the SSA-XGBoost model was applied to three mines in Hebei’s Kailuan mining area, and the results compared to China's “coal mining under buildings, railways and water-bodies” empirical formulas. Results demonstrate that the SSA-XGBoost model outperforms traditional methods, with prediction errors within 0–10% and superior accuracy on high-dimensional datasets. Compared to empirical formulas and other machine learning models, the SSA-XGBoost predictions aligned more closely with measured heights, yielding a maximum error of only 6.64%. In conclusion, this study provides a scientifically robust and effective tool for enhancing coal mine safety and protecting groundwater resources, with potential applications in related engineering fields.</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

Prediction of the Development Height of Fractured Water-Conducting Zones in Chinese Coal Mines

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Yuan Ji,
  • Lujia Yu,
  • Donglin Dong,
  • Zhonglin Wei

摘要

A comprehensive understanding of water-conducting fracture zone development is crucial for ensuring coal mining safety and groundwater resource management. This study introduces an XGBoost model, optimized by the sparrow search algorithm (SSA), to predict fracture zone heights using a nationwide database of 114 measurement samples. The Shapley additive explanations (SHAP) model is applied to assess factor influence, while comparative analyses are conducted using support vector machine (SVM) and random forest (RF) models. To validate its effectiveness, the SSA-XGBoost model was applied to three mines in Hebei’s Kailuan mining area, and the results compared to China's “coal mining under buildings, railways and water-bodies” empirical formulas. Results demonstrate that the SSA-XGBoost model outperforms traditional methods, with prediction errors within 0–10% and superior accuracy on high-dimensional datasets. Compared to empirical formulas and other machine learning models, the SSA-XGBoost predictions aligned more closely with measured heights, yielding a maximum error of only 6.64%. In conclusion, this study provides a scientifically robust and effective tool for enhancing coal mine safety and protecting groundwater resources, with potential applications in related engineering fields.