应用遗传神经网络研究低渗透储层成岩储集相——以胜利渤南油田三区沙河街组为例

Study diagenetic reservoir facies of low permeability reservoir with genetic neural network; take Shahejie Fm in block 3 of Bonan oilfield as an example

以渤南油田三区沙河街组为例,应用遗传人工神经网络模式识别方法,开展了低渗透储层成岩储集相的研究。该工作是在储集层沉积相、成岩作用研究的基础上,选用流动层带指标、孔隙度、渗透率、粒度中值、泥质含量、孔喉半径均值和变异系数等7项参数,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,对渤南油田三区沙河街组进行了成岩储集相识别,识别出4类成岩储集相:不稳定组分强溶解次生孔隙成岩储集相、碳酸盐胶结物溶解次生孔隙成岩储集相、强压实强胶结残余粒间孔成岩储集相和极强压实强胶结致密成岩储集相。Ⅰ类储集相的储集性能最好,Ⅳ类最差,为非储层或差储层。

Taking Shabejie Fm in block 3 of Bonan oilfield as an example, the pattern recognition method based on genetic artificial neural network is used to study the diagenetic reservoir facies of low permeability reservoirs. Based on study of the sedimentary facies and diagenesis of reservoirs, 7 parameters including flow zone index, porosity, permeability, median grain diameter, shale content, mean radius of pore throat and variation coefficient etc. are selected and neural network pattern recognition method is used to identify the diagenetic reservoir facies of Shabejie Fm in block 3 of Bonan oilfield through building learning and predicting models of genetic neural network. Four diagenetic reservoir facies are recognized, namely secondary pore diagenetic reservoir fa- cies resulted from strong dissolution of unstable components （type Ⅰ ）, secondary pore diagenetic reservoir facies resulted from dissolution of carbonate cements （ type Ⅱ ）, relic intergranular pore diagenetic reservoir facies after strong compaction and cementation （type Ⅲ ） , and tight diagenetic reservoir facies after extremely strong compaction and strong cementation （type Ⅳ ）. The reservoir properties of the type Ⅰ reservoir facies are the best, while that of type Ⅳ are the worst and they are nonreservoirs or poor reservoirs.