Oil-water identification method based on machine learning and related equipment
By using machine learning-based methods, a neural network-like prediction model and intersection maps are employed to identify oil and water types, solving the problem of low accuracy in identifying oil and water in low-permeability reservoirs in traditional methods and improving oilfield development efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- PETROCHINA CO LTD
- Filing Date
- 2024-10-18
- Publication Date
- 2026-04-21
AI Technical Summary
Traditional binary parameter charts based on electrical and physical properties have low accuracy in identifying oil and water in low-permeability and low-resistivity reservoirs, resulting in low oilfield development efficiency and increased losses of human, material, and financial resources.
A machine learning-based approach was adopted to collect well logging data, perform preprocessing and annotation interpretation, and establish a neural network-like prediction model. The intersection of the first and second principal variables was used to identify oil and water types. The model integrates the powerful representation learning ability of machine learning models and the formation water salinity characterization parameters, which are key influencing factors of low-permeability oil reservoirs and low-resistivity reservoirs.
It improves the scientific rigor and accuracy of fluid identification in low-permeability and low-resistivity reservoirs, thereby enhancing the success rate of oilfield exploration, reserve enhancement, and rolling development.
Smart Images
Figure CN121901809A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of oil-water layer logging characterization technology, specifically a machine learning-based oil-water identification method and related equipment. Background Technology
[0002] As oilfield development deepens, low-permeability reservoirs have become the main type of reservoir for production. However, these reservoirs are affected by various factors, including high content of interstitial material, relatively high formation water salinity, high bound water content, and hydrophilic reservoir rocks, resulting in low reservoir resistivity. This increases the difficulty of identifying fluids in these reservoirs, severely hindering efficient oilfield development. In particular, unclear oil-water identification leads to a high failure rate in production capacity construction, resulting in significant losses of human, material, and financial resources.
[0003] Due to the influence of low resistivity characteristics, the traditional method of identifying oil reservoirs based on the idea that the better the oil content, the higher the resistivity is no longer applicable. The traditional binary parameter chart based on electrical and physical properties has low accuracy in identifying oil and water in low-permeability reservoirs with low resistivity. Summary of the Invention
[0004] This invention provides a machine learning-based oil-water identification method and related equipment, which solves the problem of low accuracy in identifying oil and water in low-permeability, low-resistivity reservoirs using traditional binary parameter charts based on electrical and physical properties.
[0005] To achieve the above objectives, the present invention provides the following technical solution:
[0006] A machine learning-based method for oil-water identification includes:
[0007] Collect logging data, oil testing data, and production test data of the target formations in the completed drilling in the study area, and compile them into a database;
[0008] The well logging data in the database is preprocessed, and the database data is annotated and interpreted to obtain the interpretation results;
[0009] A neural network-like prediction model was established by using preprocessed well logging data as input variables and interpretation results as output variables.
[0010] The key influencing factors for different input quantities are obtained through the results of machine learning modeling with neural networks, and the first principal variable is established based on the key influencing factors for different input quantities.
[0011] A second main variable is established based on database data;
[0012] Draw the intersection map of the first and second principal variables to identify the oil-water type.
[0013] Preferably, the logging data includes resistivity, sonic transit time, natural gamma, porosity, permeability, and oil saturation.
[0014] Preferably, the preprocessing step for the well logging data in the database specifically includes:
[0015] The Z-Score normalization method is used to normalize the data, making the data range between ±1, while screening for outliers.
[0016] Preferably, the steps for establishing a neural network-like prediction model are as follows:
[0017] Using preprocessed logging data as input variables and interpretation results as output variables, filter parameters are set to exclude outliers, and a neural network model is trained. When the overall interpretation accuracy is greater than 80%, the model is successfully established. If the overall interpretation accuracy is less than 80%, the model parameters are reset and retraining is performed until the overall interpretation accuracy is greater than 80%.
[0018] Preferably, establishing the first main variable specifically involves:
[0019] Multiply the key influencing factors of the input quantities by the corresponding preprocessed data to obtain the variable results, and sum the variable results of all input quantities to obtain the first principal variable.
[0020] Preferably, the step of establishing the second main variable based on database data is as follows:
[0021] Based on the formation water resistivity calculated from the database, the formation water resistivity based on spontaneous potential and reservoir resistivity is calculated separately:
[0022] The formula for calculating reservoir resistivity based on spontaneous potential is:
[0023] Rw(SSP) = Rmf × 10SSP / K;
[0024] The formula for calculating formation water resistivity based on reservoir resistivity is:
[0025] Rw(RT)=Rt×φm / a
[0026] The second principal variable is obtained based on formation water resistivity, and the calculation formula is as follows:
[0027] C2 = Rw(RT) / Rw(SSP),
[0028] C2 is the second principal variable, where Rmf is the resistivity of mud filtrate, SSP is the spontaneous potential, K is the electrical permeability, φ is the electrical porosity, m is the pore structure index, and a is the lithology coefficient.
[0029] Preferably, the steps for drawing the intersection map of the first and second main variables and identifying the oil-water type are as follows:
[0030] Draw the intersection map of the first and second principal variables, using oil layers, oil-water co-layers, and water layers as distinctions to divide different identification areas, and map the new unknown layer logging data on the map to identify oil and water types.
[0031] An oil-water identification system for low-permeability, low-resistivity reservoirs includes:
[0032] Database module: Used to collect logging data, oil testing data, and production testing data of the target formations of completed drilling in the study area, and organize them into a database;
[0033] Preprocessing module: Used to preprocess the well logging data in the database, and to annotate and interpret the database data to obtain interpretation results;
[0034] Modeling module: Used to establish a neural network-like prediction model by taking preprocessed well logging data as input variables and interpretation results as output variables;
[0035] First main variable module: used to obtain the importance weights of different input quantities through the results of neural network machine learning modeling, and to establish the first main variable based on the importance weights of different input quantities;
[0036] Second main variable module: Used to establish a second main variable based on database data;
[0037] Identification module: used to draw the intersection map of the first and second main variables and identify the oil and water types.
[0038] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement a machine learning-based oil-water identification method.
[0039] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a machine learning-based oil-water identification method.
[0040] Compared with existing technologies, this invention has the following advantages: This invention proposes a machine learning-based oil-water identification method, integrating the advantages of machine learning models. Specifically, it leverages the powerful representational learning capabilities of neural networks to process complex and massive multidimensional well logging data, establishing a strong nonlinear mapping relationship. It also integrates key influencing factors of low-permeability, low-resistivity reservoirs, namely the formation water resistivity ratio factor, a characterizing parameter of formation water salinity. This improves the scientific rigor and accuracy of the method in identifying oil and water in reservoir fluids. This method can accurately determine the oil-water properties of fluids in low-permeability, low-resistivity reservoirs, significantly improving the success rate of oilfield exploration, reserve enhancement, and rolling development. Attached Figure Description
[0041] Figure 1 This is a flowchart of an oil-water identification method based on machine learning according to the present invention;
[0042] Figure 2 This is a neural network machine learning flow graph of the present invention;
[0043] Figure 3 This is a layout diagram showing the intersection of C1 and C2 in an embodiment of the present invention;
[0044] Figure 4 This is a block diagram of an oil-water identification system based on machine learning according to the present invention. Detailed Implementation
[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0046] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0047] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0048] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0049] like Figure 1 As shown, this invention provides a machine learning-based oil-water identification method, comprising:
[0050] Collect logging data, oil testing data, and production test data of the target formations in the completed drilling in the study area, and compile them into a database;
[0051] The well logging data in the database is preprocessed, and the database data is annotated and interpreted to obtain the interpretation results;
[0052] A neural network-like prediction model was established by using preprocessed well logging data as input variables and interpretation results as output variables.
[0053] The key influencing factors for different input quantities are obtained through the results of machine learning modeling with neural networks, and the first principal variable is established based on the key influencing factors for different input quantities.
[0054] A second main variable is established based on database data;
[0055] Draw the intersection map of the first and second principal variables to identify the oil-water type.
[0056] Specifically:
[0057] Step 1: Collect logging data, oil testing data, and production test data of the target formations in the completed drilling in the study area, and organize them into a basic database. The logging data mainly includes resistivity, sonic transit time, natural gamma, porosity, permeability, oil saturation, etc. The data points are selected as the average value of each formation.
[0058] Comprehensive collection of well logging, oil testing, and production testing data ensured data diversity and integrity. This data forms the basis for subsequent analysis and modeling, and is crucial for improving the accuracy of oil-water identification. Organizing the data into a basic database facilitates unified management and subsequent processing.
[0059] Step 2: Preprocess the well logging data in the database using Z-Score normalization to avoid biases in the neural network calculation process caused by different well logging values of varying magnitudes, ensuring the data falls within ±1. Simultaneously, anomalies are screened to ensure data accuracy and reliability.
[0060] The standard deviation method was used for normalization, which resolved the computational bias caused by differences in the magnitude of well logging data. This ensured that all input variables were compared on the same scale, improving the model's convergence speed and stability. Meanwhile, outlier screening ensured data accuracy and reliability, preventing erroneous data from affecting model training.
[0061] Step 3: Label the dataset. The task types are divided into oil layer, oil-water co-layer, and water layer. In order to further ensure the accuracy of the labeling, the data labeling needs to be checked. The test production data is used as the primary basis for labeling, the oil test data is used as the secondary basis for labeling, and the well logging interpretation data is used as the tertiary basis for labeling.
[0062] Clear task type labeling provides a clear objective for model training. A multi-layered label verification mechanism (production test data, oil test data, well logging interpretation data) ensures the accuracy and reliability of the labeling results. Accurate labeling data is crucial for the model to correctly learn and identify oil and water types.
[0063] Step 4: Use the preprocessed logging data as input variables, defined as prediction variables; use the interpretation results (oil layer, oil-water co-layer, and water layer) as output variables, defined as targets. Establish a neural network-like prediction model. During this process, multiple classifier models and different target benchmark models are established to improve the accuracy of machine learning and the overall interpretation results of the model. The model establishment process is as follows: Figure 3 As shown, the preprocessed logging data is used as the input variable and the interpretation result is used as the output variable. Filter parameters are set to exclude outliers and train a neural network model. When the overall interpretation accuracy is greater than 80%, the model is successfully established. If the overall interpretation accuracy is less than 80%, the model parameters are reset and the model is retrained until the overall interpretation accuracy is greater than 80%.
[0064] Choosing neural networks as the prediction model can handle complex nonlinear relationships and improve the model's predictive ability. By building multiple classifier models and selecting different target benchmark models, the performance of different models can be compared, and the optimal model can be selected for subsequent analysis. This multi-model comparison method helps improve the overall accuracy of the model's interpretation results.
[0065] Step 5: Obtain the importance weights, i.e., correlation coefficients, of different input quantities through the modeling results of neural network-like machine learning, and establish the first principal variable C = b1×Z (resistivity) + b2×Z (sonic transit time) + b3×Z (natural gamma) + b4×Z (porosity) + b5×Z (permeability) + b6×Z (oil saturation), where b* is the importance weight, i.e., correlation coefficient, of a certain input quantity, and Z(*) is the normalized data of a certain type of logging data.
[0066] By calculating the importance weights (i.e., correlation coefficients) of different input quantities and combining them with normalized logging data, the first principal variable C was established. This step helps to integrate information from multiple logging parameters to form a comprehensive index, facilitating subsequent oil-water type identification. The first principal variable C reflects the comprehensive impact of different logging parameters on oil-water type identification.
[0067] Step Six: The main influencing factors of low-resistivity oil layers include high bound water saturation, the additional conductivity of clay minerals, conductive minerals, oil-water differentiation, and high formation salinity, etc. Among these, formation water salinity is the main influencing factor, which is mainly reflected in the formation water resistivity value in well logging data. Formation water resistivity can be calculated using both spontaneous potential and reservoir resistivity. The formation water resistivity calculated using spontaneous potential is Rw(SSP) = Rmf × 10SSP / K, which approximately reflects the actual formation water environment. However, the formation water resistivity based on reservoir resistivity is Rw(RT) = Rt × φm / a, and its value is easily affected by oil content. Let C2 = Rw(RT) / Rw(SSP) to construct a new sensitive parameter to highlight the differences between oil layers, oil-water co-layers, and water layers. Where Rmf is the resistivity of mud filtrate, SSP is the spontaneous potential, K is the electrical permeability, φ is the electrical porosity, m is the pore structure index, and a is the lithology coefficient.
[0068] A new sensitive parameter, C2, was constructed to address the main influencing factors of low-resistivity oil reservoirs, particularly formation water salinity. C2 highlights the differences between oil reservoirs, oil-water co-containment layers, and water layers by comparing formation water resistivity calculated based on spontaneous potential and reservoir resistivity. This parameter provides an important reference for subsequent oil-water type identification, contributing to improved accuracy and reliability.
[0069] Step 7: Draw the C1 and C2 intersection chart, using different target values—oil layer, oil-water co-containment layer, and water layer—to distinguish and divide different identification areas. Mapping the new unknown layer logging data onto the chart allows for intuitive and accurate identification of oil and water types.
[0070] By drawing a cross-plot of C1 and C2, and using known data on oil layers, oil-water co-layers, and water layers to delineate different identification areas, new unknown layer logging data can be quickly and accurately mapped and identified on the plot. This method is intuitive, convenient, and has high identification accuracy, providing strong support for oil-water identification in actual production.
[0071] Another embodiment of the present invention provides a machine learning-based oil-water identification method, which specifically includes the following seven steps:
[0072] Step 1: Collect logging data samples, oil testing data, and production test data from completed drilling and appraisal wells in the shallow Jurassic strata of the oilfield, and compile them into a basic database of the shallow Jurassic strata of Oilfield A. The logging data mainly includes resistivity, sonic transit time, natural gamma, porosity, permeability, oil saturation, spontaneous potential, etc. A total of 202 logging data points were obtained for the oilfield, including 85 oil layers, 26 oil-water co-layers, and 91 water layers.
[0073] Step Two: Utilizing geophysical logging knowledge, select the approximate range of input logging data. Sonic transit time, porosity, and permeability reflect the pore throat development degree of low-permeability reservoirs; higher values indicate better reservoir properties. Resistivity and electrical logging oil saturation reflect the oil-bearing capacity of the reservoir; within the same region and stratigraphic level, higher values indicate better oil-bearing capacity. Natural gamma ray reflects the clay content of the reservoir; the greater the deviation below the benchmark value, the lower the clay content. Select logging data such as resistivity, sonic transit time, natural gamma ray, porosity, permeability, and oil saturation as the screening range. Preprocess the logging data in the basic database using standard deviation normalization to unify the data to the same dimension, ensuring the rationality and accuracy of the data during neural network processing.
[0074] Step 3: Relabel the output values in the dataset. The task types are divided into oil layers, oil-water co-layers, and water layers. The relabeling is based on the oil testing results, production testing, and basic knowledge of oilfield development of the corresponding layers. The main basis is the oil testing and production testing of the layers. After the correction, the types of the 202 logging data strips are changed to 82 oil layers, 30 oil-water co-layers, and 90 water layers.
[0075] Step 4: Use the preprocessed six logging data as input variables, defined as predictive variables; use the interpretation results (oil layer, oil-water co-layer, and water layer) as output variables, defined as targets. Establish a neural network-like prediction model. During this process, 10 classifier models were built, with enhancing model accuracy as the primary objective. The maximum training time was extended to 15-30 minutes. The overall model's interpretation accuracy reached 91.4%. The importance of the predictive variables was: resistivity 0.15, sonic transit time 0.29, natural gamma ray 0.09, porosity 0.02, permeability 0.27, and oil saturation 0.17.
[0076] Step 5: Establish the first principal variable C1 = 0.15×Z (resistivity) + 0.29×Z (sonic transit time) + 0.09×Z (natural gamma) + 0.02×Z (porosity) + 0.27×Z (permeability) + 0.17×Z (oil saturation) through the feedback results of neural network machine learning modeling, where Z(*) is the normalized data of a certain type of logging data, thereby calculating the C1 value corresponding to each well.
[0077] Step Six: Calculate the formation water resistivity Rw(SSP) using spontaneous potential = Rmf × 10 SSP / K This approximates the actual formation water environment, where Rmf is the resistivity of the mud filtrate, SSP is the spontaneous potential, and K is the electrical permeability. Based on the actual conditions of this oilfield, the formula Rw(SSP) = 2.02 × 10⁻⁶ is established. SSP / K This allows for the calculation of Rw(SSP) for each well. The formation water resistivity, derived from reservoir resistivity, is Rw(RT) = Rt × φ.m / a, whose value is easily affected by oil content, where φ is the electrical porosity, m is the pore structure index, and a is the lithology coefficient. Based on the measured data of this oilfield, the formula Rw(RT) = Rt × φ is established. 1.69 / 0.906, thus calculating Rw(RT) for each well. Let C2 = Rw(RT) / Rw(SSP), this sensitive parameter can highlight the differences between oil layers, oil-water co-layers, and water layers, and calculate the corresponding C2 value for each well.
[0078] Step 7: Draw the intersection chart of C1 and C2, using different target values (oil layer, oil-water co-layer, and water layer) to distinguish and divide different identification areas. See [link / details]. Figure 3 A chart for identifying fluid types in the Jurassic reservoirs of Oilfield A was created. Ten development wells in the Jurassic strata of this oilfield were selected for secondary interpretation of the low-resistivity Yan 9 and Yan 10 layers. Mapping these layers on the chart allows for intuitive and accurate identification of oil and water types. Three wells were identified as oil layers, four as oil-water co-layers, and three as water layers. Two wells with secondary interpretation of oil layers (A1-1 and A1-2) and one well with interpretation of oil-water co-layers (A2-1) were selected for repeated stimulation testing. The test results showed a daily oil / water production of 20.6 t / m³. 3 15.7t / 0m 3 10.4t / 10.2m 3 This indicates that the method has high accuracy.
[0079] like Figure 4 As shown, another embodiment of the present invention provides an oil-water identification system for low-permeability, low-resistivity reservoirs, comprising:
[0080] Database module: Used to collect logging data, oil testing data, and production testing data of the target formations of completed drilling in the study area, and organize them into a database;
[0081] Preprocessing module: Used to preprocess the well logging data in the database, and to annotate and interpret the database data to obtain interpretation results;
[0082] Modeling module: Used to establish a neural network-like prediction model by taking preprocessed well logging data as input variables and interpretation results as output variables;
[0083] First main variable module: used to obtain the importance weights of different input quantities through the results of neural network machine learning modeling, and to establish the first main variable based on the importance weights of different input quantities;
[0084] Second main variable module: Used to establish a second main variable based on database data;
[0085] Identification module: used to draw the intersection map of the first and second main variables and identify the oil and water types.
[0086] An embodiment of the present invention provides a terminal device. This terminal device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the various device embodiments described above.
[0087] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention.
[0088] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.
[0089] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
[0090] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.
[0091] If the modules / units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.
[0092] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art, guided by the specification, can make many other modifications without departing from the scope of the claims of the present invention, and all of these modifications are within the scope of protection of the present invention.
Claims
1. A machine learning-based method for oil-water identification, characterized in that, include: Collect logging data, oil testing data, and production test data of the target formations in the completed drilling in the study area, and compile them into a database; The well logging data in the database is preprocessed, and the database data is annotated and interpreted to obtain the interpretation results; A neural network-like prediction model was established by using preprocessed well logging data as input variables and interpretation results as output variables. The key influencing factors for different input quantities are obtained through the results of machine learning modeling with neural networks, and the first principal variable is established based on the key influencing factors for different input quantities. A second main variable is established based on database data; Draw the intersection map of the first and second principal variables to identify the oil-water type.
2. The oil-water identification method based on machine learning according to claim 1, characterized in that, The logging data includes resistivity, sonic transit time, natural gamma, porosity, permeability, and oil saturation.
3. The oil-water identification method based on machine learning according to claim 1, characterized in that, The specific steps for preprocessing well logging data in the database are as follows: The Z-Score normalization method is used to normalize the data, making the data range between ±1, while screening for outliers.
4. The oil-water identification method based on machine learning according to claim 1, characterized in that, The specific steps for establishing a neural network-like prediction model are as follows: Using preprocessed logging data as input variables and interpretation results as output variables, filter parameters are set to exclude outliers, and a neural network model is trained. When the overall interpretation accuracy is greater than 80%, the model is successfully established. If the overall interpretation accuracy is less than 80%, the model parameters are reset and retraining is performed until the overall interpretation accuracy is greater than 80%.
5. The oil-water identification method based on machine learning according to claim 1, characterized in that, The specific steps to establish the first master variable are as follows: Multiply the key influencing factors of the input quantities by the corresponding preprocessed data to obtain the variable results, and sum the variable results of all input quantities to obtain the first principal variable.
6. The oil-water identification method based on machine learning according to claim 1, characterized in that, The specific steps for establishing a second primary variable based on database data are as follows: Based on the formation water resistivity calculated from the database, the formation water resistivity based on spontaneous potential and reservoir resistivity is calculated separately: The formula for calculating reservoir resistivity based on spontaneous potential is: Rw(SSP) = Rmf × 10SSP / K; The formula for calculating formation water resistivity based on reservoir resistivity is: Rw(RT)=Rt×φm / a The second principal variable is obtained based on formation water resistivity, and the calculation formula is as follows: C2 = Rw(RT) / Rw(SSP), C2 is the second principal variable, where Rmf is the resistivity of mud filtrate, SSP is the spontaneous potential, K is the electrical permeability, φ is the electrical porosity, m is the pore structure index, and a is the lithology coefficient.
7. The oil-water identification method based on machine learning according to claim 1, characterized in that, The specific steps for drawing the intersection map of the first and second principal variables and identifying the oil-water type are as follows: Draw the intersection map of the first and second principal variables, using oil layers, oil-water co-layers, and water layers as distinctions to divide different identification areas, and map the new unknown layer logging data on the map to identify oil and water types.
8. An oil-water identification system for low-permeability, low-resistivity reservoirs, characterized in that, include: Database module: Used to collect logging data, oil testing data, and production testing data of the target formations of completed drilling in the study area, and organize them into a database; Preprocessing module: Used to preprocess the well logging data in the database, and to annotate and interpret the database data to obtain interpretation results; Modeling module: Used to establish a neural network-like prediction model by taking preprocessed well logging data as input variables and interpretation results as output variables; First main variable module: used to obtain the importance weights of different input quantities through the results of neural network machine learning modeling, and to establish the first main variable based on the importance weights of different input quantities; Second main variable module: Used to establish a second main variable based on database data; Identification module: used to draw the intersection map of the first and second main variables and identify the oil and water types.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the machine learning-based oil-water identification method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the machine learning-based oil-water identification method as described in any one of claims 1 to 7.