鄂东气田煤层含气量测井预测

Logging Predicting for Coalbed Gas Content in Eastern Block of Ordos Basin

煤层含气量计算的准确与否直接关系到煤层气开发方案的有效制定。在简要介绍了现有煤层含气量评价方法的基础上,利用测井资料和煤岩心含气量分析化验资料,采用统计回归方法优选了煤层含气量的敏感性测井参数,并基于优选的测井参数,运用多元回归和神经网络两种数学方法构建了鄂东气田的煤层含气量测井预测模型。利用所构建的模型对研究区内的煤层含气量进行了预测,预测结果与煤岩心含气量室内分析数据对比表明,多元回归法和神经网络法均能较好地对煤层含气量进行预测,但神经网络法的预测精度更高。

The accurate calculation of coalbed gas content is key important to effective development of coalbed methane development plan. This paper first briefly described the existing coalbed gas content evaluation method. Then based on logging data which hidden rich coalbed gas content and indoor analyze data which is the actual coal rock gas content, logging parameters which are sensitive to coalbed gas content are clarified by using statistical re gression. Based on the optimal logging parameters, the coalbad gas content logging prediction model are estab lished by application of multiple regression and neural network mathematical methods. This model was tested to predict the coalbed gas content in the study area. Comparison of the prediction with the actual coal gas content in the laboratory analysis show that multiple regression method and neural networks can both better predict coalbed gas content by using the selected log parameters, but the neural network has the higher prediction accuracy.