煤层含气量预测的BP神经网络模型与应用  

Prediction Models of Coal Bed Gas Content Based on BP Neural Networks and Its Applications

作　　者：孟召平[1] 田永东[1] 雷旸[1] 

机构地区：[1]中国矿业大学资源与地球科学系,北京100083

出　　处：《中国矿业大学学报》2008年第4期456-461,共6页Journal of China University of Mining & Technology

基　　金：国家自然科学基金项目(40772100);国家重点基础研究发展计划(973)项目(2007CB209405);全国优秀博士学位论文作者专项资金项目(200247);高等学校博士学科点专项科研基金项目(20050290009);教育部留学回国人员科研启动基金项目

摘　　要：为了对煤层含气量进行定量预测,采用BP神经网络预测方法,建立了煤层含气量预测的BP神经网络模型.以沁水盆地南部主采煤层为对象,分析得出了影响沁水盆地南部煤层含气量分布的主要控制因素有煤层有效埋藏深度、煤变质程度和煤岩、煤质特征等,选择了煤层有效埋藏深度、水分与灰分以及镜质组最大反射率3参数作为BP神经网络模型的基本特征量,建立了煤层含气量与这些因素之间的相关关系和BP神经网络预测模型,对煤层含气量进行预测分析.结果表明:BP神经网络模型具有极强的非线性逼近能力,能真实反映煤层含气量与主控因素之间的非线性关系,预测结果与实测值之间误差小,相对误差小于10%,预测效果明显地优于基于朗格缪尔方程的煤层含气量预测模型.In order to predict coal bed gas content quantitatively, a BP neural networks model was established using a BP neural networks prediction method. Based on the primary mineable coal bed in the southern Qinshui basin, the main controlling factors, including the effective buried depth of coal seam, the metamorphic degree of coal, the coal rock and the coal quality that affect the distribution of coal bed gas content were analyzed. Taking the effective buried depth.of coal seam, moisture and ash, maximal reflectance mean value of vitrinite as the basic characteristic quantity of BP neural networks models, the correlation between the gas content of the coal seam and the factors and the BP neural networks prediction model were proposed for prediction of the coal bed gas content. The results show that the BP neural networks model has strong ability for nonlinear approach which can actually reflect the nonlinear relationship between the coal bed gas content and main controlling factors. The small errors of less than 10% between the prediction values and measured values were achieved, which is obviously superior to that with the Lang-muir equation.

关 键 词：煤层气 含气量 BP神经网络模型 

分 类 号：P618.11[天文地球—矿床学]