基于测井参数的煤储层含气量预测方法研究  

Research on Gas Content Prediction Method for Coal Reservoirs based on Logging Parameters

作　　者：白云思雨 谢非 王赐勋 BAIYUN Siyu;XIE Fei;WANG Cixun(School of Earth Sciences and Engineering,Xi'an Shiyou University,Shaanxi 710065;Shaanxi Provincial Key Laboratory of Oil and Gas Accumulation Geology,Xi'an Shiyou University,Shaanxi 710065)

机构地区：[1]西安石油大学地球科学与工程学院,陕西710065 [2]西安石油大学陕西省油气成藏地质学重点实验室,陕西710065

出　　处：《中国煤层气》2024年第3期29-32,共4页China Coalbed Methane

摘　　要：煤层气含气量测定通常受人为因素影响较大,且测定成本高时间长,而测井数据连续且易获取,通过与含气量建立线性关系,确定利用补偿中子、密度、自然伽马、电阻率、声波时差和深度这六种参数用来建立煤层含气量预测模型。采用多元线性回归、BP神经网络、随机森林这三种方法构建煤储层含气量预测模型。随后利用构建的模型进行盲井检验,结果表明以上三种预测模型均能较好预测煤层含气量,而随机森林模型在此区块预测效果最好。The determination of coalbed methane gas content is often significantly affected by human fac-tors,and the measurement cost is high and the time is long.However,the logging data is continuous and easy to obtain.Through the establishment of a linear relationship with the gas content,six parameters of compensation neutron,density,natural gamma,resistivity,sonic time difference and depth are identified and used to establish the prediction model for the gas content of the coal seam.Three methods,multiple linear regression,BP neural network and random forest,are used to construct a prediction model for coal reservoir gas content.Subsequently,the constructed model is used to test the blind well,and the results show that the above three prediction models can effectively predict the gas content in the coal seam,with the random forest model showing the best prediction effect in this Block.

关 键 词：煤层 煤层气 多元线性回归 BP神经网络 随机森林 含气量 

分 类 号：P631.81[天文地球—地质矿产勘探] P618.13[天文地球—地质学]