摘要
塔木察格盆地南贝尔油田构造条件复杂,孔隙结构特征差异性大,非均质性强,区内南屯组储集层流体性质识别难度较大,测井、录井解释符合率均不理想。为了探讨南贝尔油田南屯组储集层流体性质识别的有效方法,将基于主成分分析的马氏距离判别法引用到流体性质解释过程中,取该区22口探井、评价井已获试油验证的储集层流体10项参数的原始样本数据,以其通过降维得到的3个主成分作为判别因子,并求取经主成分计算后新样本的均值向量和样本协方差矩阵。在假设油层、干层、油水同层、水层4个不同类属总体的协方差矩阵相等的前提下,构建该区储集层流体性质的马氏距离判别模型,并依据判别函数值最大准则确定判别结果。应用结果表明:基于主成分分析的马氏距离判别模型的回判准确率为93.33%,预测正确率约为80%,可为该油田南屯组储集层流体性质的识别提供参考依据。
The South Buir Oilfield in Tumuchage Basin has a complex construction conditions,with big characteristics of pore structure difference and strong heterogeneity.In the target zone,it is difficult to identify the reservoir fluid property in Nantun formation,and the interpretation coincidence rates of well logging and mud logging were not ideal.In order to explore the effective methods of identifying the reservoir fluid property in Nantun formation of the South Buir Oilfield,Mahalanobis distance discriminant method based on principal component analysis was applied to the interpretation process of the fluid property.Adopt the original sample data of 10 reservoir fluid parameters validated by formation testing from 22 exploration and appraisal wells of the research area,the 3principal components obtained by dimension reduction are used as discriminant factors,and the mean vector and covariance matrix of the new samples were got after calculating the principal components.On the assumption that the covariance matrix is equal in the four different genera of oil layers,dry layers,oil-water layers and water layers,MahalanobisDistance Discriminant model for reservoir fluid properties in the study area was built.The discriminant results are based on maximum criteria of discriminant function value.The results showed that the regression accuracy of mahalanobis Distance Discriminant model based on principal component analysis was 93.33%,prediction accuracy is about 80%,this methed can provide some references for identifying reservoir fluid properties of Nantun formation in the South Buir Oilflied.
出处
《录井工程》
2017年第1期43-48,共6页
Mud Logging Engineering
关键词
南贝尔油田
主成分分析
马氏距离
流体性质
判别模型
Salth Buir Oilfield
principal component analysis
Mahalanobis distance
fluid property
