Research On The Control System Of Oil Artificial Intelligence-based Identification Method Of Heterogeneous Reservoir Fluid Properties In XD Transition Zone

The lithology of the XD transition zone is mainly argillaceous siltstone,siltstone and fine sandstone.The porosity and permeability of sandstone reservoirs are high,and the permeability difference of the same porosity is complicated,the difference of oil content is obvious,the difference of electrical and physical properties between oil and water reservoirs is not obvious,and the heterogeneity is strong.The reservoir thickness ranges from 0.6m to4.4m.There are both thin and thick layers in the XD transition zone,and thin layers and thin interlayers lead to the phenomenon of low oil and high water resistance layers in the XD transition zone.It is more difficult to identify the fluid properties of heterogeneous reservoirs.Compared with the pure oil zone,the interpretation standards and models for the identification of oil and water layers can no longer meet the requirements of fine and accurate interpretation.Therefore,it is of great significance to identify the heterogeneous fluid properties of XD transition zone reservoir.At present,the main method to identify reservoir flow properties is to establish the conventional graph method based on logging data,but it has been applied to the problem of low coincidence rate and heavy workload in the identification of heterogeneous reservoir flow properties.When using artificial intelligence method to distinguish the properties of heterogeneous reservoir flow,it can not only extract the nonlinear relationship between the data,but also reduce the labor cost,which has obvious advantages compared with the manual interpretation of logging data.Based on conventional logging and core analysis data,this paper analyzes the reservoir characteristics of the XD transition zone.In view of the heterogeneity characteristics of the XD transition zone,it analyzes the four relationships of the reservoir,establishes the reservoir parameter model,and combines with the oil test data based on artificial intelligence to predict the R₀（saturated pure water resistivity）curve and distinguish the reservoir flow properties.It reflects the advantages of high resolution in artificial intelligence technology,solves the problem that conventional chart method cannot identify the heterogeneous reservoir fluid properties,improves the identification accuracy of reservoir fluids with strong heterogeneity caused by thin layers and thin interlayers,and meets the needs of XD transition zone development.At the same time,by comparing three methods based on artificial intelligence（DNN）,K proximity（KNN）and support vector machine（SVM）to learn the fluid properties and logging data in the study area and verify other test Wells in the study area,an artificial intelligence architecture method with a high consistency rate is obtained.Based on the in-depth research on the identification method of low resistivity oil layer based on artificial intelligence technology,deep neural network method is determined to solve the problem that the fluid property identification of low resistivity oil layer and high resistivity water layer exists in the heterogeneous reservoir of XD transition zone.The application of this discriminant method in production is very important to improve the discriminant coincidence rate of heterogeneous strong reservoir fluid properties and the working efficiency of researchers.