基于灰色关联分析和BP神经网络的煤层含气量预测研究

Prediction of methane gas content in coalbed based on the grey relation analysis and BP neural network

基于测井资料和煤岩心含气量化验分析资料,采用现代数理统计方法优选了煤层含气量的敏感性测井参数,在煤岩心含气量分析化验数据归位的基础上,建立了煤岩心含气量-测井相统计模式,并利用灰色关联法对测井相-煤岩心含气量统计模式进行了系统分析,进而筛选了有效的煤层含气量测井建模数据库。基于神经网络非线性数学方法,利用筛选后的有效煤层含气量测井建模数据构建了研究区内煤层含气量的多测井参数非线性预测模型,并利用所构建模型对研究区内的煤层含气量进行预测。煤岩心含气量室内分析数据与预测结果对比表明,该整套方法能较好地对煤层含气量进行预测,预测精度能够满足煤层气储层测井评价的要求。

Based on the logging data and the core analysis data,the logging parameters which are sensitive to the methane gas content in coalbed are optimized by using statistics regression,and the statistical mode of methane gas content in coalbed cores- logging facies in the studied area is established. Then the statistical mode is analyzed by using the grey correlation,and finally the effective logging database for the prediction of methane gas content in coalbed is obtained. The neural network model for the prediction of methane gas content in coalbed is established according to the effective logging database,and the methane gas content in coalbed in the studied area is predicted using the model. The comparison of the predicted result with core analysis result shows that the accuracy of the methane gas content prediction method can meet the need of the logging evaluation of coalbed gas reservoirs.