【摘要】 应用遗传神经网络模式识别方法,以文72块沙三中为例,在取芯关键井流动单元聚类分析的基础上,选用流动层指数、孔隙度、渗透率、粒度中值、泥质含量、最大孔喉半径6项参数,将取芯井流动单元划分为4类,采用神经网络模式识别方法,通过建立遗传神经网络的学习及预测模型,对文72块沙三中油藏进行了流动单元识别,阐述了各类流动单元的特征,并应用序贯指示模拟,获得了流动单元的时空展布。流动单元与沉积微相空间分布的对比表明,物性和储集能力都较好的流动单元大部分位于水下分支水道微相中部及河口坝微相,水道和河口坝沉积是控制物性较好流动单元的主要沉积微相。储集层流动单元比沉积微相更精细地刻画了影响储集层流体流动的地下结构,通过流动单元研究可以预测剩余油的可能分布。更多还原

【Abstract】 Taking the middle member of Shahejie Formation in Wen72 block as an example,the pattern recognition method based on genetic artificial neural network is used to study flow units of the clastic rock reservoir.Based on the cluster analysis of core wells,six parameters including flow zone index,porosity,permeability,median grain size,mud content and max pore throat radius,are selected to classify the flow units into four types.Then the neural network pattern recognition method is used to identify the flow unit types in all wells of Wen72 block by building learning and predicting models.At the same time,the analysis of the characteristics of each flow unit and the simulation of the sequential indicator simulation make it possible to realize the time-space distributions of four flow unit types.A comparison of the spatial distribution of reservoir flow units with the sedimentary micro-facies shows that the main flow units with superior reservoir petrophysical properties lie in the middle of the channels and mouth bars,the main sedimentary micro-facies with superior controlling properties.The flow units show better underground structure of the reservoir fluid flowing than the sedimentary micro-facies.It is believed that the distribution of remaining oil can be forecasted by studying the flow units.更多还原