基于瓦斯实时监测的炮掘工作面爆破自动识别与突出危险性预测  

Auto-identification of blasting and outburst risk prediction in the blasting driving face based on real-time gas monitoring

作　　者：彭玉杰 宋大钊 李振雷 何学秋 王洪磊[1] 邱黎明 PENG Yujie;SONG Dazhao;LI Zhenlei;HE Xueqiu;WANG Honglei;QIU Liming(School of Civil and Resources Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]北京科技大学土木与资源工程学院,北京100083

出　　处：《煤炭科学技术》2022年第5期171-178,共8页Coal Science and Technology

基　　金：国家自然科学基金资助项目(51634001,51774023,51904019)。

摘　　要：为实现应用炮掘工作面炮后瓦斯实时监测数据进行突出危险性动态预测的目标,提出了一种应用一阶差分、卷积运算和MATLAB中的findpeaks函数,自动识别瓦斯体积分数监测曲线中爆破事件的方法,应用提出的自动识别方法,对某炮掘工作面的爆破事件进行了自动识别与提取,并对识别结果进行了评价;基于爆破识别与提取的结果,分析了炮后瓦斯体积分数变化特征,计算了炮后瓦斯体积分数增长速率、峰值和衰减速率3个指标与K1值的相关性,据此建立了炮掘工作面突出危险性预测模型,并将模型应用于另一炮掘工作面,对工作面前方的突出危险性进行预测。研究结果表明:提出的自动识别方法实现了对爆破时刻、爆破时的瓦斯体积分数、炮后瓦斯体积分数峰值及其时刻的自动识别与提取,识别爆破事件的召回率平均为84.13%、精准率平均为77.21%,提取出的炮后瓦斯体积分数峰值和爆破时刻的平均绝对误差分别为0.0189%和2.3237 min;炮后瓦斯体积分数增长速率、峰值、衰减速率都与K1值强相关,相关系数分别为0.85、0.92、0.79;应用建立的突出危险性预测模型,成功地预测出了25次喷孔事件和1次小型压出事件,预报效能为60.48%。研究结果可作为现有突出预测方法的补充,辅助钻屑瓦斯解吸指标K1值进行突出危险性预测,并可为炮掘工作面瓦斯异常识别及突出危险性预测提供指导。In order to predict the outburst risk dynamically by using gas real-time monitoring data after blasting in blasting driving face,this paper proposed a method to automatically identify blasting event in the gas concentration monitoring curve by applying the first-order difference,convolution operation and the findpeaks function in MATLAB.By using the auto-identification method,the blasting event in a blasting driving face is automatically identified and extracted.Then the recognition result is evaluated.Based on the results of blasting i⁃dentification and extraction,analyzed the variation characteristics of the gas concentration after blasting,and calculated the correlations be⁃tween the growth rate,peak value,decay rate of gas concentration after blasting and K_(1) value,and based on these three indexes,a prediction model of outburst risk for blasting driving face is established and applied to another face to predict the outburst risk in front of the working face.The results show that the proposed auto-identification method realized the automatic identification and extraction of the blasting mo⁃ment,the gas concentration at the blasting,the peak value and the moment of the gas concentration after the blasting.The recall and preci⁃sion of the proposed auto-identification method are 84.13%and 77.21%respectively,and the mean absolute errors of the extracted peak gas concentration after blasting and the blasting time are 0.0189%and 2.3237 min,respectively.The growth rate,peak value,decay rate of gas concentration after blasting are all strongly related to the K_(1) value,and the correlation coefficients between three indexes and K_(1) val⁃ue are 0.85,0.92 and 0.79 respectively.Applying the established outburst risk prediction model,25 times jet hole and 1 time small out⁃burst were successfully predicted,with prediction efficiency of 60.48%.The results can be used as a supplement to the existing outburst risk prediction methods,assist K_(1) value in prediction of outburst risk,and provide guidance for the

关 键 词：炮掘工作面 瓦斯监测 爆破自动识别 相关性分析 瓦斯突出 

分 类 号：TD713[矿业工程—矿井通风与安全]