摘要
储层的微孔隙结构是影响高含水期油田剩余油分布的主要因素。提出了基于神经网络技术对测井资料处理以识别储层孔隙结构类型的方法。介绍了BP神经网络原理,该方法利用人工神经网络技术所具有的非线性、容错性和较强的模式识别能力实现了综合推理,进行储层孔隙结构类型预测。选取反映孔隙结构类型特征的自然电位、自然伽马,声波时差等7条常规测井曲线建立样本模式,并统一刻度,进行归一化处理,建立了神经网络模型。对大庆油田采油五厂储层样本进行了处理,符合率达80%以上,表明该方法用于预测储层微孔隙结构类型是可行和有效的。
The micropore structure of the reservoir is fundamental factor that influences the residual oil distribution of high water-cut oilfield.Put forward is a method that estimates the micropore structure types according to the BP neural network.Introduced are the principle of the BP neural network,which uses the artificial neural network technique that has nonlinearity,fault-tolerant ability and stronger pattern recognition capability to implement estimating the micropore structure types through synthetical reasoning.Selected are the 7 logging curves which reflect the characteristics of the reservoir porosity type,including SP,GR,AC,etc.,to build sample mode.The neural network is found through calibration and normalization.Practical application in reservoir samples from the 5th Plant of Daqing Oilfield shows that the prediction coincidence rate reaches 80%.It is concluded that this method is feasible and effective in reservoir micropore structure prediction.
出处
《测井技术》
CAS
CSCD
北大核心
2009年第4期355-359,共5页
Well Logging Technology
关键词
测井资料
神经网络
孔隙结构
BP算法
log data
neural network
micropore structure
BP algorithm
