机器学习技术在煤层底板突水预测中的研究现状
Research Status of Machine Learning Technology in Prediction of Water Inrush from Coal Seam Floor
摘要: 为了对煤层底板突水实现更加精准可靠的预测,减少矿井水害的发生。机器学习在煤层底板突水预测得到了广泛应用,对目前煤层底板突水预测的应用效果进行汇总分析,整理出当下煤层底板突水预测所存在的主要问题。底板突水预测技术存在的主要问题在于水文地质条件复杂且突水数据较少;煤层底板突水主控因素的选取对基于机器学习的煤层底板突水预测十分重要;算法模型的选取在泛化能力与过拟合问题起很大作用;深度学习的快速发展为底板突水预测技术提供新的方向,但在训练数据上仍然有较大的影响。对未来煤层底板突水预测技术发展进行了思考,对近些年煤层底板突水预测方法汇总并进行比较,发现目前技术所存在的利弊,对未来精准、高效地煤层底板突水预测进行展望。
Abstract: In order to achieve more accurate and reliable prediction of water inrush from coal seam floor and reduce the occurrence of mine flood. Machine learning has been widely used in the prediction of water inrush from coal seam floor. This paper summarizes and analyzes the application effect of current coal seam floor water inrush prediction, and sorts out the main problems existing in the current coal seam floor water inrush prediction. The main problem of the water inrush prediction technology is that the hydrogeological conditions are complex and the water inrush data are few. The selection of main controlling factors of water inrush from coal seam floor will have a great impact on the prediction of water inrush from coal seam floor based on machine learning. The se-lection of algorithm model plays an important role in generalization ability and over-fitting prob-lem. Although the rapid development of deep learning provides a new direction for floor water inrush prediction technology, it still has a great impact on training data. At the end, the future de-velopment of coal seam floor water inrush prediction technology is considered and explored. This paper summarizes and compares the prediction methods of water inburst from coal seam floor in recent years, and finds out the advantages and disadvantages of the current technology. The accu-rate and efficient prediction of water inrush from coal seam floor in the future is prospected.
文章引用:李晨曦, 鲁海峰. 机器学习技术在煤层底板突水预测中的研究现状[J]. 矿山工程, 2023, 11(3): 362-366. https://doi.org/10.12677/ME.2023.113045

参考文献

[1] 李忠建, 魏久传, 郭建斌, 徐建国, 隋岩刚. 运用突水系数法和模糊聚类法综合评价煤层底板突水危险性[J]. 矿业安全与环保, 2010, 37(1): 24-26.
[2] 李白英. 预防矿井底板突水的“下三带”理论及其发展与应用[J]. 山东矿业学院学报(自然科学版), 1999(4): 11-18. [Google Scholar] [CrossRef
[3] 李万军, 杨家兵. “下三带”理论和“P-h”临界曲线法预测底板突水[J]. 煤矿开采, 2010, 15(5): 45-47. [Google Scholar] [CrossRef
[4] 代革联, 薛小渊, 许珂, 牛超, 杨韬. 基于脆弱性指数法的韩城矿区11号煤层底板突水危险性评价[J]. 煤田地质与勘探, 2017, 45(4): 112-117, 125.
[5] 靳德武. 我国煤层底板突水问题的研究现状及展望[J]. 煤炭科学技术, 2002, 30(6): 1-4. [Google Scholar] [CrossRef
[6] 王国瑞, 冯书顺, 马自强, 翟延亮, 张维. 突水系数法的演化及应用[J]. 内蒙古煤炭经济, 2015(7): 123, 147. [Google Scholar] [CrossRef
[7] 武强, 刘守强, 贾国凯. 脆弱性指数法在煤层底板突水评价中的应用[J]. 中国煤炭, 2010, 36(6): 15-19, 22. [Google Scholar] [CrossRef
[8] 李颖. 基于支持向量机的煤层底板突水预测方法研究[D]: [硕士学位论文]. 北京: 煤炭科学研究总院, 2007.
[9] 刘再斌, 靳德武, 刘其声. 基于二项logistic回归模型与CART树的煤层底板突水预测[J]. 煤田地质与勘探, 2009, 37(1): 56-61.
[10] 潘晖, 王继尧. 基于粒子群优化神经网络的煤层底板突水预测[J]. 山西焦煤科技, 2009(1): 34-36, 42.
[11] 孟祥瑞, 王军号, 高召宁. 基于IoT-GIS耦合感知的煤层底板突水预测研究[J]. 中国安全科学学报, 2013, 23(2): 85-91. [Google Scholar] [CrossRef
[12] 刘雪艳, 张雪英, 李凤莲. 基于万有引力的煤层底板突水预测算法[J]. 煤炭学报, 2015(S2): 458-463. [Google Scholar] [CrossRef
[13] 温廷新, 孙雪, 田洪斌, 孔祥博. 基于PCAFuzzyRF模型的煤层底板突水预测[J]. 安全与环境学报, 2017, 17(3): 855-858. [Google Scholar] [CrossRef
[14] 陈建平, 王春雷, 王雪冬. 基于CNN神经网络的煤层底板突水预测[J]. 中国地质灾害与防治学报, 2021, 32(1): 50-57. [Google Scholar] [CrossRef
[15] 尹会永, 周鑫龙, 郎宁, 张历峰, 王明丽, 吴焘, 李鑫. 基于SSA优化的GA-BP神经网络煤层底板突水预测模型与应用[J]. 煤田地质与勘探, 2021, 49(6): 175-185.