基于信息熵的煤自燃标志气体优选  

Optimization of Index Gas for Coal Spontaneous Combustion Based on Information Entropy

作　　者：韩文强 HAN Wen-qiang(Energy Service Center,Luzhou District Energy Bureau,Changzhi 046000,China)

机构地区：[1]山西省长治市潞州区能源局能源服务中心,山西长治046000

出　　处：《煤》2021年第9期24-27,共4页Coal

摘　　要：为更好预测预报王庄煤矿3号煤层自然发火进程,通过程序升温氧化实验并采用基于信息熵的C4.5决策树算法,分析煤自燃不同阶段标志气体及相关比值的信息增益率。结果表明:CO的信息增益率在整个升温氧化过程中都很高,而C_(2)H_(4)只有在进入高危险期时信息增益率才较高,因此CO应作为煤自燃低危险期—较高危险期的主要标志气体,C_(2)H_(4)作为进入高危险期的主要标志气体;同时φC_(2)H_(4)/φC_(2)H_(6)和φC_(2)H_(4)/φCH4在进入高危险期时信息增益率也较高,其分裂阈值分别为0.042和0.026,该阈值有效避免了现场风量等因素的影响,可作为高危险期的辅助指标。In order to predict the spontaneous combustion process of No. 3 coal seam in Wangzhuang Coal Mine, this paper analyzed the information gain-ratio of index gases and their ratios based on the programmed temperature oxidation experimental and C4.5 decision tree algorithm of information entropy. The results shows that the information gain rate of CO is very high in the whole process of heating and oxidation, while the information gain rate of C_(2)H_(4) is very high only when it enters the high risk period. Therefore, CO should be used as the main marker gas for the low risk period and high risk period of coal spontaneous combustion, and C_(2)H_(4) should be used as the main marker gas for the high risk period. And φC_(2)H_(4)/φC_(2)H_(6) and φC_(2)H_(4)/φCH4 into the crisis, high information gain rate is higher, the threshold value were 0.042 and 0.026, the threshold effectively avoid air volume, the influence of such factors as the scene can be used as auxiliary indexes in high danger.

关 键 词：煤自燃 标志气体 预测预报 信息熵 信息增益率 

分 类 号：TD752.2[矿业工程—矿井通风与安全]