摘要
页岩气储层总有机碳(TOC)含量是页岩气评价的重要参数,如何准确确定TOC含量是页岩气勘探开发研究的一个关键问题。以黔江地区下志留统龙马溪组为研究对象,通过页岩气储层有机碳含量测试和钻井测井资料的统计分析,研究了TOC含量的测井响应特征,优选了体积密度(DEN)、自然伽马(GR)、自然电位(SP)和声波时差(AC)4条测井曲线作为特征向量,建立了TOC含量的BP神经网络预测模型,改进了BP神经网络算法,并对黔江地区1口页岩气井下志留统龙马溪组TOC含量进行了预测和对比分析。结果表明:基于测井参数的BP神经网络模型具有极强的非线性逼近能力,能真实反映页岩气储层TOC含量与测井参数之间的非线性关系,预测结果与实测值之间误差小,相对误差一般小于10%。
Total organic carbon （TOC） content of shale gas reservoir is an important parameter of shale gas assessment, and how to accurately determine the content of TOC is a key problem of shale gas exploration and development. The author used the Lower Silurian Longmaxi formation in Qianjiang area as the research object. Through the statistical analysis of TOC content testing of shale gas reservoir and drilling and logging data, the log response characteristics of TOC content were analyzed. Furthermore,the four logs which consist of volume density logging（DEN） ,gamma logging （GR） ,spontaneous potential logging（ SP ） and acoustic logging（AC） were selected optimally as the feature vector. Afterwards, the BP neural network prediction model of TOC content was established, the BP neural network algorithm was improved and the TOC content of Lower Silurian Longmaxi formation of two shale gas wells in Qianjiang area were predicted and compared. The results show that the BP neural network model based on logging parameters has strongly approximate nonlinearization, which can reflect the nonlinear relationship between the TOC content of shale gas reservoir and logging parameters. The error between prediction results and measured values is small, and the relative error is less than 10%.
出处
《煤炭学报》
EI
CAS
CSCD
北大核心
2015年第2期247-253,共7页
Journal of China Coal Society
基金
国家重点基础研究发展计划(973)资助项目(2012CB214705)
国家自然科学基金资助项目(41372163
41172145)
关键词
页岩气储层
测井参数
有机碳含量(TOC)
预测模型
shale gas reservoir
logging parameters
total organic carbon （ TOC ） content
prediction model
