测录井结合神经网络流体识别技术在高邮凹陷阜宁组的应用  

Application of Fluid Identification Based on the Neural Networks by Combination of Wireline and Mud Logging in Funing Formation,Gaoyou Sag

作　　者：任培罡[1,2] 尹军强[1] 杨加太[1] 曹书坡[1] 雷磊[1] 

机构地区：[1]中国石化江苏石油工程有限公司地质测井处,江苏扬州225007 [2]江苏油田博士后工作站,江苏扬州225007

出　　处：《测井技术》2015年第2期242-246,260,共6页Well Logging Technology

基　　金：江苏省企业博士后集聚计划资助项目(JD11002)

摘　　要：针对高邮凹陷阜宁组低孔隙度低渗透率储层成因复杂,非均质性强,流体识别困难的特点,提出了测录井结合的神经网络流体识别技术。对传统的BP神经网络算法进行改进,加入动量项提高收敛速度和避免陷入局部极小,自适应调整学习参数,优化网络结构确定隐层神经元个数;将测井资料、录井资料相结合,把3个曲线参数和10个录井参数综合应用于算法之中,在高邮凹陷阜宁组低孔隙度低渗透率储层解释中取得较好的效果,解释结论与试油结论有较高的一致性。The low porosity and low permeability reservoirs of Funing Formation in Gaoyou bag represent the features of anisotropy with complex forming reasons, which is more difficult for the fluid identification in this area. Therefore, the fluid identification based on the neural networks by combination of wireline and mud logging, which was improved based on the traditional BP Neural Network, has been used in this study to get high convergence speed and avoid local minimum points by adding momentum item and adaptive adjustment algorithm. Meanwhile, the number of hide-layer is confirmed by optimizing the structure. Lastly, combined with wireline and mud logging, three wireline and ten mud logging parameters are used in this algorithm. The interpretation application of low porosity and low permeability reservoirs in Funing Formation of Gaoyou Sag illustrates that the method is available and effective. The interpretation conclusions are consistent with the oil testing results.

关 键 词：测井解释 低孔隙度 低渗透率 流体识别 BP神经网络 录井资料 江苏油田 

分 类 号：P631.84[天文地球—地质矿产勘探]