任怀伟, 刘凯, 李建, 等.超大采高工作面煤壁稳定性分析及片帮预测研究[J].河南理工大学学报（自然科学版）,2025,44(3):1-11.

REN H W, LIU K, LI J, et al. Stability analysis and rib spalling prediction of coal wall in ultra-high mining working faces[J]. Journal of Henan Polytechnic University(Natural Science) , 2025, 44(3): 1-11.

超大采高工作面煤壁稳定性分析及片帮预测研究

任怀伟1,2, 刘凯3, 李建1,2, 赵叔吉1, 韩存地4, 王龙4, 李济洋1

1.中煤科工开采研究院有限公司 智能开采装备分院，北京  100013；2.煤炭科学研究总院 开采研究分院，北京  100013；3.中国矿业大学（北京） 能源与矿业学院，北京  100083；4.陕西陕煤曹家滩矿业有限公司，陕西 榆林  719099

摘要: 目的 为了解决超大采高工作面煤壁片帮难题， 方法 通过理论分析与现场实测研究煤壁片帮机理，采用数值模拟对不同采高条件下煤壁的破坏过程进行模拟，探讨采高和顶板压力与煤壁片帮之间的关系，并基于现场数据，利用机器学习方法进行片帮预测，比较并选择最优的预测模型。  结果 研究发现，超大采高工作面煤壁片帮主要发生在煤壁的中上部，随着采高和支承压力的增加，片帮破坏程度呈加剧趋势。分析得到片帮稳定性系数计算公式，发现稳定性系数随顶板压力和采高的增加而增大。数值模拟结果表明，当采高从6 m增至10 m时，煤壁破碎程度和片帮深度显著增加，且片帮深度在顶板来压时达到峰值。数据分析结果表明，煤壁片帮是煤壁、液压支架和顶板相互作用的结果，合理的支护高度和良好的液压支架支撑力是降低片帮风险的关键。最后，采用机器学习方法对煤壁片帮进行预测，比较多种常见的机器学习算法，发现KNN方法的预测准确率最高，达77.46%。然而，现有机器学习方法在片帮预测上的准确性仍需进一步提高。 结论 研究结果揭示了超大采高工作面煤壁片帮的发生机理，提出了片帮预测的新思路，即随着采高和顶板压力增加，煤壁片帮风险加大，采用机器学习方法可以有效预测片帮发生的可能性，这为超大采高工作面煤矿安全生产提供了理论支持和技术保障。

关键词:超大采高工作面;煤壁片帮;稳定性分析;机器学习;片帮预测

doi: 10.16186/j.cnki.1673-9787.2024110015

基金项目:国家自然科学基金资助项目（52274207）

收稿日期:2024/11/08

修回日期:2025/01/25

出版日期:2025-04-18

Stability analysis and rib spalling prediction of coal wall in ultra-high mining working faces

REN Huaiwei1,2, LIU Kai3, LI Jian1,2, ZHAO Shuji1, HAN Cundi4, WANG Long4, LI Jiyang1

1.Smart Mining Branch， CCTEG Coal Mining Research Institute Co.， Ltd.， Beijing  100013， China；2.Coal Mining and Designing Branch， China Coal Research Institute， Beijing  100013， China；3.China University of Mining and Technology-Beijing， School of Energy and Mining Engineering， Beijing  100083， China；4.Shaanxi Caojiatan Coal Mining Co.， Ltd.， Shaan Mei Coal Group Corp.， Yulin  719099， Shaanxi， China

Abstract: Objectives The problem of coal wall rib spalling in ultra-high mining height working faces was addressed. Methods The mechanism of rib spalling was studied through theoretical analysis and field measurements. Numerical simulations were conducted to model coal wall failure under different mining height conditions. The relationship between mining height， support pressure， and rib spalling was analyzed. Machine learning methods were applied to predict rib spalling based on field data. Several prediction models were compared to select the most optimal one.  Results It was found that rib spalling occurred mainly in the middle and upper parts of the coal wall. As mining height and support pressure increased， rib spalling worsened. A stability coefficient formula for rib spalling was derived. It was found that the stability coefficient increased with the increase of the roof pressure and mining height. Numerical simulations showed that as mining height increased from 6m to 10m， the degree of coal wall fragmentation and rib spalling depth increased significantly， and the depth of rib spalling reached the peak point when the roof was pressed. Data analysis revealed that rib spalling resulted from the interaction between the coal wall， hydraulic supports， and the roof. Finally， maintaining proper support height and sufficient hydraulic support force was critical to reduce rib spalling risk. Machine learning methods were applied to predict rib spalling. Among the common machine learning algorithms， the KNN method achieved the highest prediction accuracy， 77.46%. However， the prediction accuracy of current machine learning methods still required improvement.  Conclusions The mechanism of rib spalling in ultra-high mining height working faces was revealed. A new approach to predicting rib spalling was proposed. As mining height and roof pressure increased， the risk of rib spalling grew. Machine learning methods could effectively predict rib spalling risks， providing theoretical support and technical assurance for the safe operation of ultra-high mining height coal mines.

Key words: ultra-high mining height working face; coal wall rib spalling; stability analysis; machine learning; rib spalling prediction