神经网络技术识别厚油层层内剩余油方法

以检查井资料为基础,利用统计分析的方法建立储层物性(孔隙度和渗透率)及岩性(泥质含量)的测井解释模型,并按厚油层内部存储性和渗流性质的差异,建立起三级流动单元的识别和划分标准。在此基础上,利用神经网络技术对密闭取心检查井资料进行学习训练,建立起原始含油饱和度、目前含油饱和度和残余油饱和度的测井解释模型,从而实现对厚油层层内剩余油的综合定量解释,为高含水期厚油层层内剩余油挖潜提供物质基础。

Research on neural network technology to identify remaining oil in thick oil layer

The remaining reserves of multiple-layer sandstone oilfield mainly exist in the thick oil layers at its late stage of high water cut. Considering the complex distribution and identification difficulty of the remaining oil in thick oil layers, firstly based on inspecing well data, establish the well logging interpretation model of reservoir physical properties (porosity and permeability) and lithology (shale content) by statistic analysis method, and constitute the three-level flow unit identification and division criteria according to the difference of thick oil layer interior accumulation and seepage features. Based on this, neural network technology is adopted to study and learn sealed coring inspection well data and to establish the well logging interpretation model of initial oil saturation, current oil saturation and residual oil saturation. Thus the comprehensive quantitative interpretation of remaining oil in thick oil layer is realized, which provide a material basis for remaining oil potential development in thick oil layer at high water cut stage.