首页|基于机器学习的煤与瓦斯突出危险性评估和预警研究进展

基于机器学习的煤与瓦斯突出危险性评估和预警研究进展

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
煤与瓦斯突出是威胁煤矿安全生产的重大灾害之一.近年来机器学习的兴起及最新理论成果为评估煤与瓦斯突出危险性、超前预警煤与瓦斯突出灾害、减少或避免事故危害提供了新思路和有力手段.突出危险性评估主要针对待开采煤层进行突出危险性整体分析,而在线监测预警是在采掘过程中实时监测煤层突出危险性.总结了突出机理、危险性评估和预警的多源信号类型及预警指标体系,并给出了基于多源时空信息融合的突出前兆特征识别与危险性分级预警步骤.基于突出危险性评估和预警研究现状及当前煤炭安全智能开采需求,指出当前研究存在的问题,针对评价预警指标灵活性不足的问题,提出知识数据联合驱动、动态调整和实时分析提高指标灵活性的方法;针对现场干扰事件影响预警准确性的问题,提出更优数据处理与多目标优化策略的方法,提高模型抗噪声干扰能力;针对多源信息融合预警研究薄弱的问题,提出先进信号处理方法结合融合模型的思路,保证融合的准确性;针对突出危险预警超前性不足的问题,提出时序分析融合深度学习的方法;针对突出小样本导致特征挖掘不充分的问题,提出扩充不平衡样本的方法,以提高预警准确率.
Research progress on coal and gas outburst risk assessment and early warning based on machine learning
Coal and gas outburst is one of the major disasters threatening the safety of coal mine production.The rise and latest theoretical achievements of machine learning in recent years have provided novel ideas and powerful means for the risk assessment and early warning of coal and gas outburst,which helps to reduce or even avoid accident losses.The outburst risk assessment mainly focuses on the overall analysis of outburst risk in the coal seam to be mined,while the online monitoring and early warning is the real-time monitoring of coal seam outburst risk during the mining process.The outburst occurrence mechanism and multi-source signal types and indicator system for risk assessment and early warning are summarized,and the steps for identifying precursor features and grading risk early warning of outburst hazard based on multi-source information fusion are also given.Based on the current research status of outburst risk assessment and early warning research and the current needs for coal safety intelligent mining,the existing problems of research are pointed out.In order to solve the problem of insufficient flexibility of evaluation early warning indicators,the methods for improving the flexibility of indicators are proposed by combining knowledge and data driving,dynamic adjustment and real-time analysis.Aiming at the problem of field interference events affecting the accuracy of early warning,better data processing and multi-objective optimization strategy is proposed to improve the anti-noise interference ability of the model.For the weak research of multi-source information fusion early warning,the idea of advanced signal processing method combined with fusion model is proposed to ensure the accuracy of fusion.Aiming at insufficient advance warning of outburst risk,time series analysis integrated with deep learning is proposed.Aiming at insufficient feature mining caused by small samples of outburst,expanding unbalanced samples is proposed to improve the accuracy of early warning.

coal and gas outburstmachine learningoutburst risk assessmentmulti-source information fusionearly warning for out burst

李宛桐、夏方方、朱旖旎、赵胜磊、杨紫云、王金鑫

展开 >

中国矿业大学安全工程学院,江苏省徐州市,221116

煤与瓦斯突出 机器学习 突出危险性评估 多源信息融合 突出预警

"十四五"国家重点研发计划国家自然科学基金江苏省基础研究计划(自然科学基金)徐州市前沿引领技术基础研究项目中国矿业大学国家级大学生创新创业训练计划

2022YFC300470552304271BK20221117KC22001202310290013Z

2024

10.19880/j.cnki.ccm.2024.07.007

中国煤炭
煤炭信息研究院

中国煤炭

CSTPCD北大核心
影响因子：0.736
ISSN：1006-530X
年,卷(期)：2024.50(7)
  • 36