基于领域驱动数据挖掘技术定量描述油层水淹级别

Identifying water flooded layers based on the domain-driven data mining technique

为了准确识别水淹层、确定储集层的水淹程度,引入数据挖掘技术,结合水淹层评价领域的相关知识,从"领域驱动"的角度建立高效预测储集层水淹程度的模型。结合研究区块沉积相带特征,综合运用聚类分析、关联规则和决策树等方法,对水淹层的测井数据及储集层衍生参数进行分沉积微相、分参数组合挖掘,分别获得预测整个研究区块和各小层水淹层的模型;并给出挖掘参数对油层水淹情况敏感程度的大小关系,证实不同沉积微相对水淹情况敏感的参数不同。将所得预测模型程序化之后,对研究区180口老井水淹层进行复查、对10口新井水淹层进行预测,其结果与试油结论对比,准确率为91.03%,证明了方法的可靠性。

To identify water flooded layers and determine the flooded degree of reservoirs,the data mining technique is introduced and an efficient predictive model,combining with the relative knowledge of evaluating water flooded layers,is established to identify water flooded layers from the view of domain-driven.With the sedimentary facies characteristics of the study area,this paper uses the cluster analysis,association rule and decision tree to mine the logging data of the flooded layers and reservoir derived parameters by sedimentary microfacies and parameter combinations.As a result,the predictive models of the entire region and small layers are obtained.And the sensitivity of the parameters in identifying the water flooded layers is given,proving that different sedimentary microfacies have different parameters sensitive to flooded conditions.After the predictive models of the small layers are programmed,they are used to evaluate the 180 old wells and 10 new wells;the accuracy is 91.03 percent,which proved the reliability of this method.