Method and system for predicting initial productivity of offshore high-water-content reservoir production well

Abstract

The invention discloses an offshore high-water-cut reservoir production well initial productivity prediction method and system, and the method comprises the steps: obtaining dynamic and static factor data affecting the initial productivity of a production well in an offshore high-water-cut reservoir and the initial productivity data of the high-water-cut reservoir production well, and carrying out the preprocessing of the dynamic and static factor data and the initial productivity data; determining a main control factor based on the dynamic and static factor data and the initial productivity data which are subjected to combined analysis and preprocessing by multiple methods; a COA-RF model is constructed, and hyper-parameter combination assignment of the COA-RF model is obtained based on the main control factors; a COA-RF model is trained based on hyper-parameter combination assignment, the trained COA-RF model is tested, and a prediction result of the initial productivity of the high-water-content oil reservoir production well is output; and based on a prediction result, evaluating the performance of the COA-RF model by adopting different indexes. According to the method, the main control factors of the initial productivity of the high-water-content oil reservoir production well can be accurately identified, and the prediction precision is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of offshore oil and gas field development technology, and in particular to a method and system for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir. Background Technology

[0002] In the development of offshore high water-cut oil reservoirs, accurately predicting the initial production capacity of production wells is crucial for formulating reasonable development plans, optimizing resource allocation, controlling development costs, and improving recovery rates and economic benefits.

[0003] However, existing methods for predicting the initial production capacity of wells mainly rely on numerical simulation and well logging interpretation models. Both methods have significant limitations. While numerical simulation can predict production without actual production data, it involves large computational loads, requires high data accuracy, and suffers from complex fluid motion physical mechanisms, resulting in long simulation times and low efficiency. Furthermore, well logging interpretation models calibrate logging curves based on actual core test data, extending their application to non-core wells. However, they primarily provide static geological parameters and cannot fully reflect the dynamic impact of multiple factors on well production capacity. As the amount of data increases, shallow neural network models struggle to handle complex nonlinear relationships, exhibiting limited prediction accuracy and generalization ability. Summary of the Invention

[0004] This invention provides a method and system for predicting the initial production capacity of production wells in offshore high water-cut oil reservoirs, in order to solve the technical problems of low accuracy in identifying the main controlling factors of production capacity of production wells in high water-cut oil reservoirs and low accuracy in handling complex nonlinear relationships in the prior art.

[0005] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a method for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir, comprising:

[0006] S10, acquire dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and preprocess the dynamic and static factor data and initial production capacity data.

[0007] S20, based on the combined analysis of the preprocessed dynamic and static factor data and initial production capacity data using multiple methods, determine the main controlling factors;

[0008] S30, construct the COA-RF model, and assign hyperparameter combinations to the COA-RF model based on the main controlling factors;

[0009] S40, Train the COA-RF model based on the hyperparameter combination assignment, test the trained COA-RF model, and output the prediction results of the initial production capacity of the production well in the high water-cut reservoir.

[0010] S50, Based on the prediction results, the performance of the COA-RF model is evaluated using different metrics.

[0011] Optionally, step S10 includes:

[0012] S101, to obtain dynamic and static data on factors affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs;

[0013] S102, to obtain initial production capacity data of production wells in high water-cut reservoirs;

[0014] S103, perform missing value and outlier detection and standardization processing on the dynamic and static factor data and initial production capacity data.

[0015] Optionally, the dynamic and static factors affecting the initial production capacity of production wells in the offshore high water-cut oil reservoirs include single-well sand body vertical thickness, single-well perforation vertical thickness, logging porosity, original oil saturation, clay content, logging permeability, surface crude oil viscosity, mobility, formation flow coefficient, reservoir variation coefficient, reservoir permeability surge coefficient, production pressure differential, well inclination angle, and average water cut.

[0016] Optionally, step S20 includes:

[0017] S201, the preprocessed dynamic and static factor data and initial production capacity data are analyzed using a combination of linear regression, grey relational analysis, SHAP method, Pearson correlation coefficient method, and MDI method.

[0018] S202, Based on the analysis results, the controlling factors are determined by majority voting.

[0019] Optionally, step S30 includes:

[0020] S301, based on the dynamic and static factors affecting the initial production capacity of production wells in the offshore high water-cut oil reservoir and the initial production capacity data of production wells in the high water-cut oil reservoir, data samples are randomly extracted with replacement to generate multiple subset datasets;

[0021] S302, Construct an RF model based on the aforementioned multiple subset datasets;

[0022] S303, using the mean absolute error (MAE) as the fitness function value, the hyperparameters of the RF model are optimized using the COA algorithm to obtain the COA-RF model, and the hyperparameter combination of the COA-RF model is obtained through iterative optimization.

[0023] Optionally, the formula for calculating the mean absolute error (MAE) is as follows:

[0024]

[0025] Among them, y i This is the true value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs at sea. is the predicted value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs, and n is the total number of samples in the initial production capacity sample dataset of high water-cut oil reservoirs.

[0026] Optionally, obtaining the hyperparameter combination values ​​of the COA-RF model through iterative optimization includes:

[0027] S3001, Initialize the crayfish population and calculate the fitness value of each crayfish;

[0028] S3002, Update the location and status of individual crayfish;

[0029] S3003, solve the crayfish position solution space and evaluate the fitness function of individual crayfish based on the current position;

[0030] S3004. The iteration terminates when the optimal value of the hyperparameters of the RF model is found or the maximum number of iterations is reached.

[0031] Optionally, the hyperparameters of the COA-RF model are the number of trees, the maximum depth of the trees, the minimum number of samples for splitting a tree node, the minimum number of samples for leaf nodes, and the maximum number of features considered during splitting.

[0032] Optionally, the different metrics include MAE, MSE, RMSE, and R².

[0033] Secondly, this invention provides a system for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir, the system comprising:

[0034] The data acquisition module is used to acquire dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and to preprocess the dynamic and static factor data and initial production capacity data.

[0035] The main control factor identification module is used to determine the main control factors by jointly analyzing the preprocessed dynamic and static factor data and initial production capacity data using multiple methods.

[0036] The model building module is used to build a COA-RF model and obtain the hyperparameter combination values ​​of the COA-RF model based on the main control factors.

[0037] The prediction result output module is used to train the COA-RF model based on the hyperparameter combination assignment, test the trained COA-RF model, and output the prediction result of the initial production capacity of the production well in the high water-cut reservoir.

[0038] The performance evaluation module is used to evaluate the performance of the COA-RF model using different metrics based on the prediction results.

[0039] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0040] This application employs a combination of methods to jointly analyze the preprocessed dynamic and static factor data and initial production data, avoiding the limitations of a single method and ensuring accurate identification of the main controlling factors for the initial production capacity of wells in high water-cut reservoirs. Furthermore, this application uses the COA algorithm to optimize the hyperparameters of the RF model, significantly improving prediction accuracy. Attached Figure Description

[0041] Figure 1 This is a flowchart illustrating the method for predicting the initial production capacity of offshore high water-cut oil reservoirs provided in this embodiment of the invention.

[0042] Figure 2 This is a schematic diagram of the structure of the offshore high water-cut oil reservoir production well initial production capacity prediction system provided in an embodiment of the present invention;

[0043] Figure 3 Heatmap of missing values ​​for the dataset of initial production capacity prediction of offshore high water-cut oil reservoirs provided in this embodiment of the invention;

[0044] Figure 4 A flowchart illustrating the integrated feature engineering method provided in this embodiment of the invention;

[0045] Figure 5 The experimental results of the linear regression method provided in the embodiments of the present invention are shown in the figure.

[0046] Figure 6 The experimental results of the grey relational analysis method provided in the embodiments of the present invention are shown in the figure.

[0047] Figure 7 The experimental results of the SHAP method provided in the embodiments of the present invention are shown in the figure.

[0048] Figure 8 The experimental results of the Pearson correlation coefficient method provided in the embodiments of the present invention are shown in the figure.

[0049] Figure 9 The experimental results of the MDI method provided in this embodiment of the invention are shown in the figure.

[0050] Figure 10 A diagram showing the majority voting experiment results of the integrated feature engineering method provided in this embodiment of the invention;

[0051] Figure 11 This is a diagram illustrating the iterative optimization process of the COA-RF model provided in this embodiment of the invention.

[0052] Figure 12 The experimental results of the COA-RF model for predicting the initial production capacity of production wells in offshore high water-cut oil reservoirs provided in this embodiment of the invention are shown in the figure.

[0053] Figure 13 The experimental results of predicting the initial production capacity of a production well in a high water-cut offshore reservoir using the standard random forest model provided in this embodiment of the invention are shown in the figure.

[0054] Figure 14 The experimental results of the standard decision regression tree model for predicting the initial production capacity of offshore high water-cut oil reservoirs are shown in the figure provided in the embodiments of the present invention.

[0055] Figure 15 The experimental results of the standard XGBoost model for predicting the initial production capacity of production wells in high water-cut offshore reservoirs, provided for embodiments of the present invention, are shown in the figure.

[0056] Figure 16 The experimental results of the standard support vector regression model for predicting the initial production capacity of production wells in offshore high water-cut oil reservoirs, provided in the embodiments of the present invention, are shown in the figure.

[0057] Figure 17 The experimental results of the Bayesian-RF model for predicting the initial production capacity of production wells in high water-cut offshore oil reservoirs, provided in this embodiment of the invention, are shown in the figure.

[0058] Figure 18 The experimental results of the PSO-RF model for predicting the initial production capacity of offshore high water-cut oil reservoirs, provided in this embodiment of the invention, are shown in the figure. Detailed Implementation

[0059] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] In the relevant descriptions of this embodiment, the terms "including," "containing," and "possessing" are all open terms and are generally understood to include but not be limited to; the term "at least one" is generally understood to mean one or more, where "multiple" refers to two or more; the term "at least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items, for example, "at least one of a, b, or c", or "at least one of a, b, and c", which can all mean: a, b, c, ab (i.e., a and b), ac, bc, or abc, where a, b, and c can be single or multiple; the symbol "A / B" is used to describe the selection relationship of associated objects, generally indicating an "or" relationship.

[0061] In the following description of the embodiments, the terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms "a" and "the" as used in the embodiments of this application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0062] Those skilled in the art should understand that, in the following description of the embodiments of this application, the sequence of numbers does not imply the order of execution. Some or all steps may be executed in parallel or sequentially. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0063] Those skilled in the art will understand that the numerical ranges in the embodiments of this application should be understood to specifically disclose each intermediate value between the upper and lower limits of the range. Any stated value or intermediate value within a stated range, as well as any other stated value or each smaller range between intermediate values ​​within a range, are also included within this invention. The upper and lower limits of these smaller ranges may be independently included or excluded from the range.

[0064] Unless otherwise stated, the technical / scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. While this application describes only preferred methods and materials, any methods and materials similar or equivalent to those described herein may be used in the implementation or testing of this application. All references to this specification are incorporated by way of citation to disclose and describe the methods and / or materials associated with those references. In the event of any conflict with any incorporated reference, the content of this specification shall prevail.

[0065] This invention provides a method for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir. This method can be executed by a computer. (Refer to...) Figure 1 The method may include the following steps:

[0066] S10: Obtain dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and preprocess the dynamic and static factor data and initial production capacity data.

[0067] High water-cut reservoirs refer to reservoirs with a water cut between 60% and 90%.

[0068] In an exemplary embodiment, step S10 may include the following steps:

[0069] S101, to obtain dynamic and static data on factors affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs;

[0070] S102, to obtain initial production capacity data of production wells in high water-cut reservoirs;

[0071] S103, perform missing value and outlier detection and standardization processing on the dynamic and static factor data and initial production capacity data.

[0072] Specifically, the dynamic and static factors affecting the initial production capacity of production wells in the aforementioned high water-cut offshore oil reservoirs include: single-well sand body vertical thickness, single-well perforation vertical thickness, logging porosity, original oil saturation, clay content, logging permeability, surface crude oil viscosity (standard conditions), mobility, formation flow coefficient, reservoir variability coefficient, reservoir permeability surge coefficient, production pressure differential, well inclination angle (angle-based well inclination angle), and average water cut (12 months); the initial production capacity data of the high water-cut oil reservoir production wells includes... The missing value heatmap of the initial production capacity prediction dataset for high water-cut offshore oil reservoir production wells is shown below. Figure 3 As shown.

[0073] Specifically, vertical thickness refers to the vertical thickness of a single-well sand body, that is, the thickness of the sand body in a single well in the vertical direction; perforation thickness refers to the vertical thickness of a single-well perforated layer, that is, the thickness of the oil layer opened up in a single well through perforation technology in the vertical direction; porosity refers to logging porosity, that is, the rock porosity value indirectly measured by logging methods (such as sonic logging); oil saturation refers to the original oil saturation; clay content is the clay content; permeability refers to logging permeability, which is the formation rock permeability indirectly estimated by core calibration; crude oil viscosity refers to the surface crude oil viscosity (standard conditions); formation flow coefficient is the formation flow coefficient; the remaining terms are consistent. The well inclination angle uses an angle-based well inclination angle.

[0074] After step S10, the dynamic and static factor data and the initial production capacity data need to be divided into training set and test set: 70% of the dataset is used as the training set and the remaining 30% of the dataset is used as the test set.

[0075] The well inclination angle (angle-based well inclination angle) is calculated through the following steps:

[0076] (1) The vertical thickness of the sand body in a single well was obtained through experimental measurement. .

[0077] (2) The thickness of the sand body in a single well was obtained through experimental measurement. .

[0078] (3) Through the formula Calculate the radian well inclination angle .

[0079] (4) Through the formula Calculate the wellbore inclination angle .

[0080] The experimental measurement steps described above are as follows:

[0081] 1. Obtain physical samples of underground rock strata through drilling core sampling, directly measure the sand body thickness, and obtain the vertical thickness of the sand body in a single well. .

[0082] II. By utilizing the response characteristics of well logging curves (such as natural gamma, resistivity, sonic transit time, etc.) to identify sand bodies and calculate their thickness, the thickness measurement of sand bodies in a single well is obtained. .

[0083] Furthermore, the preprocessing in step S103 employs the following methods:

[0084] Missing value handling was performed as follows: A heatmap was used to analyze the distribution of missing values. The heatmap uses color intensity to represent numerical magnitude; in the missing value analysis, dark red represents missing values ​​and blue represents non-missing values. Rows represent sample data from individual wells, and columns represent data features from individual wells, forming a two-dimensional matrix. The color distribution quickly located concentrated areas of missing values. Features with severe missing values ​​were then removed.

[0085] Outlier detection is performed using the following methods: box plots or 3D curves. Abnormal samples are removed as a rule.

[0086] The standardization process uses Z-score standardization to eliminate the influence of dimensions.

[0087] S20. Based on the combined analysis of the preprocessed dynamic and static factor data and initial production capacity data using multiple methods, the main controlling factors are determined.

[0088] In an exemplary embodiment, reference is made to Figure 4 Step S20 may include the following steps:

[0089] S201, the preprocessed dynamic and static factor data and initial production capacity data are analyzed using a combination of linear regression, grey relational analysis, SHAP method, Pearson correlation coefficient method, and MDI method.

[0090] S202, Based on the analysis results, the controlling factors are determined by majority voting.

[0091] Specifically, Figure 4 For a flowchart of the integrated feature engineering method (using majority voting to determine the controlling factors), refer to... Figure 4 The principle for determining the controlling factor through majority voting is as follows: if a feature is selected by three or more methods, it is considered the controlling factor for the initial production capacity (12 months) of a high water-cut reservoir well. The experimental results of majority voting using the comprehensive feature engineering method are shown in the figure below. Figure 10 As shown, by Figure 10 It can be seen that the main controlling factors finally determined include: perforation thickness, porosity, oil saturation, permeability, mobility, formation flow coefficient, and average water cut (12 months).

[0092] The following sections will elaborate on linear regression, grey relational analysis, SHAP method, Pearson correlation coefficient method, and MDI method:

[0093] Linear regression method: By calculating the fitting coefficients between each feature and the initial production capacity (12 months) of a well in a high water-cut reservoir, the top 70% of important features are selected. The formula for linear regression is as follows:

[0094]

[0095] Where y is the initial production capacity of a production well in a high water-cut reservoir (12 months). These are characteristics such as vertical thickness, perforation thickness, porosity, oil saturation, clay content, permeability, crude oil viscosity, mobility, formation flow coefficient, well inclination angle, and average water cut (12 months); This is the intercept value, representing the predicted value when all features are 0. These are the fitting coefficients for each feature, reflecting the degree of influence of each feature on the initial production capacity (12 months) of production wells in high water-cut reservoirs; This is the error term, representing the part that the model cannot explain.

[0096] Grey relational analysis: By calculating the grey relational degree, the correlation between each feature and the initial production capacity (12 months) of production wells in high water-cut reservoirs is measured, and the top 70% of important features are selected.

[0097] SHAP Method: Based on game theory, this method calculates the feature contribution (SHAP value) and selects the top 70% of important features. The SHAP method treats model prediction as a cooperative game, with each feature considered a "player" and the prediction result (initial production capacity of a high water-cut reservoir well (12 months)) considered the "total revenue." The SHAP value is calculated by allocating the "revenue share" of each feature based on its marginal contribution across all possible combinations of feature subsets. The SHAP value calculation method is as follows:

[0098]

[0099] in, yes The SHAP value of the feature Except Let f(s) be the model's prediction on any feature subset S outside the given set. It is a weighting factor that ensures that the contributions of all combinations are weighted fairly. Features include vertical thickness, perforation thickness, porosity, oil saturation, clay content, permeability, crude oil viscosity, mobility, formation flow coefficient, well inclination angle, and average water cut (12 months). n is the number of features.

[0100] Pearson correlation coefficient method: By calculating the linear correlation (absolute value) between each feature and the initial production capacity (12 months) of a production well in a high water-cut reservoir, the top 70% of important features are selected.

[0101] MDI (Multi-Discrimination Method): This method calculates the importance score of each feature based on a decision tree, selecting the top 70% of important features. Specifically, the decision tree constructs the model by recursively splitting the dataset, selecting the feature that causes the greatest decrease in impurity of the child nodes at each split. The MDI method utilizes this property, calculating the weighted average of the impurity reduction for each feature across all split nodes, as its importance score. The formula for calculating the reduction in impurity after splitting is as follows:

[0102]

[0103] in, It is the amount of impurity reduced after splitting. It is the current split node. , These are the left and right child nodes after the split; , , It is a decision tree node , , The number of samples in It is a decision tree node Initial production capacity of production wells in medium-to-high water-cut reservoirs (12 months). It is the first One characteristic, The following parameters were collected: vertical thickness, perforation thickness, porosity, oil saturation, clay content, permeability, crude oil viscosity, mobility, formation flow coefficient, well inclination angle, and average water cut (12 months). It is the average initial production capacity (12 months) of high water-cut reservoir production wells in the left child nodes after the decision tree node t splits. It is the average initial production capacity (12 months) of high water-cut reservoir production wells in the right child nodes after the decision tree node t splits. It is the average initial production capacity (12 months) of the high water-cut reservoir production well in the decision tree node t.

[0104] The experimental results of the above methods (linear regression, grey relational analysis, SHAP method, Pearson correlation coefficient method, and MDI method) are shown in the figures below. Figures 5-9 As shown, Figures 5-9 In the figure, the horizontal axis represents the fitting coefficient, grey relational degree, SHAP value, Pearson correlation coefficient, and importance score, respectively.

[0105] S30, Construct an RF model, train the RF model using a training set, and assign hyperparameter combinations to the RF model based on the controlling factors.

[0106] In an exemplary embodiment, step S30 may include the following steps:

[0107] S301, based on the dynamic and static factors affecting the initial production capacity of production wells in the offshore high water-cut oil reservoir and the initial production capacity data of production wells in the high water-cut oil reservoir, data samples are randomly extracted with replacement to generate multiple subset datasets;

[0108] S302, Construct an RF model based on the aforementioned multiple subset datasets;

[0109] S303, using the Mean Absolute Error (MAE) as the fitness function value, the optimal hyperparameters of the RF model are optimized using the CoA algorithm to obtain the COA-RF model. The hyperparameter combinations of the COA-RF model are then obtained through iterative optimization. The iterative optimization process of the optimal hyperparameters of the COA-RF model (fitness change curve) is shown in the figure below. Figure 11 As shown, Figure 11 In the graph, the horizontal axis represents the number of iterations, and the vertical axis represents the optimal fitness.

[0110] The hyperparameters of the COA-RF model are: n_estimators (number of trees), max_depth (maximum depth of the tree), min_samples_split (minimum number of samples for splitting a tree node), min_samples_leaf (minimum number of samples for leaf nodes), and max_features (maximum number of features to consider during splitting).

[0111] In an exemplary embodiment, the formula for calculating the Mean Absolute Error (MAE) is as follows:

[0112]

[0113] Among them, y i This is the true value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs at sea. is the predicted value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs, and n is the total number of samples in the initial production capacity sample dataset of high water-cut oil reservoirs.

[0114] In an exemplary embodiment, obtaining the hyperparameter combination assignment of the COA-RF model through iterative optimization includes:

[0115] S3001, Initialize the crayfish population and calculate the fitness value of each crayfish;

[0116] S3002, Update the location and status of individual crayfish;

[0117] S3003, solve the crayfish position solution space and evaluate the fitness function of individual crayfish based on the current position;

[0118] S3004. The iteration terminates when the optimal value of the hyperparameters of the RF model is found or the maximum number of iterations is reached.

[0119] Specifically, a Radio Frequency (RF) model contains multiple hyperparameters, and the crayfish optimization algorithm is an optimization algorithm that seeks an optimal combination of hyperparameters. A crayfish population simulates the numerous values ​​of these hyperparameters in the RF model, with each crayfish representing a set of values ​​for these hyperparameters.

[0120] S40, the RF model is trained based on the hyperparameter combination, the trained RF model is tested using a test set, and the predicted production capacity of the production well in the high water-cut reservoir is output.

[0121] Figure 12The figure shows the experimental results of the COA-RF model predicting the initial production capacity of production wells in high water-cut offshore reservoirs. In the figure, blue represents the training set and purple represents the test set. The figure shows that, regardless of whether it is in the training set or the test set, the difference between the actual and predicted initial production capacity of a single well is small, and the coefficient of determination R0 is relatively high. 2 The maximum accuracy of the COA-RF model is due to its high prediction accuracy.

[0122] S50, Based on the prediction results, the performance of the COA-RF model is evaluated using different metrics.

[0123] Specifically, the different metrics include MAE, MSE, RMSE, and R².

[0124] After evaluating the performance of the COA-RF model using different metrics, the model was compared with other machine learning models, including: standard random forest (RF), standard decision regression tree (DT), standard extreme gradient boosting tree (XGBoost), standard support vector regression (SVR), particle swarm optimization random forest (PSO-RF), and Bayesian optimization random forest (bayesian-RF). Figures 13-18 The diagrams illustrating the prediction results of the aforementioned machine learning models are shown below. Figure 12 and Figures 13-18 The comparison shows that, compared to Figures 13-18 Whether in the training set or the test set, Figure 12 The actual initial production capacity of a single well differs relatively little from the predicted initial production capacity. Compared to... Figure 13-18 The coefficient of determination R in the test set 2 ( Figure 13 It is 0.893. Figure 14 It is 0.859. Figure 15 It is 0.857. Figure 16 It is 0.743. Figure 17 It is 0.936. Figure 18 (0.945) Figure 12 The coefficient of determination R in the test set 2 The maximum value (0.951) indicates that the COA-RF model provided in this application significantly improves prediction accuracy.

[0125] In one or more embodiments of this application, the preprocessed dynamic and static factor data and initial production data are analyzed using multiple methods in combination, avoiding the limitations of a single method and ensuring accurate identification of the main controlling factors for the initial production capacity of production wells in high water-cut reservoirs. Furthermore, this application employs the COA algorithm to optimize the hyperparameters of the RF model, significantly improving prediction accuracy.

[0126] Based on the above embodiments, this application also provides a system for predicting the initial production capacity of production wells in offshore high water-cut oil reservoirs. Figure 2 This is a schematic diagram of the structure of a system for predicting the initial production capacity of a high water-cut offshore oil reservoir, as described in an embodiment of the present invention. Figure 2 As shown, the initial production capacity prediction system 200 for offshore high water-cut oil reservoirs may include:

[0127] The data acquisition module 210 is used to acquire dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and to preprocess the dynamic and static factor data and initial production capacity data.

[0128] The main control factor determination module 220 is used to determine the main control factors by jointly analyzing the preprocessed dynamic and static factor data and initial production capacity data using multiple methods.

[0129] The model building module 230 is used to build a COA-RF model and obtain the hyperparameter combination values ​​of the COA-RF model based on the main control factors.

[0130] The prediction result output module 240 is used to train the COA-RF model based on the hyperparameter combination assignment, test the trained COA-RF model, and output the prediction result of the initial production capacity of the production well in the high water-cut reservoir.

[0131] The performance evaluation module 250 is used to evaluate the performance of the COA-RF model using different metrics based on the prediction results.

[0132] In the description of this application, it should be noted that the terms "first", "second", and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0133] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interface; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0135] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0136] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0137] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0138] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the technical scope disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

[0139] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

Claims

1. A method for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir, characterized in that, include: S10, acquire dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and preprocess the dynamic and static factor data and initial production capacity data. S20, based on the combined analysis of the preprocessed dynamic and static factor data and initial production capacity data using multiple methods, determine the main controlling factors; S30, construct the COA-RF model, and assign hyperparameter combinations to the COA-RF model based on the main controlling factors; S40, Train the COA-RF model based on the hyperparameter combination assignment, test the trained COA-RF model, and output the prediction results of the initial production capacity of the production well in the high water-cut reservoir. S50, Based on the prediction results, the performance of the COA-RF model is evaluated using different metrics.

2. The method according to claim 1, characterized in that, Step S10 includes: S101, to obtain dynamic and static data on factors affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs; S102, to obtain initial production capacity data of production wells in high water-cut reservoirs; S103, perform missing value and outlier detection and standardization processing on the dynamic and static factor data and initial production capacity data.

3. The method according to claim 1, characterized in that, The dynamic and static factors affecting the initial production capacity of production wells in the offshore high water-cut oil reservoirs include single-well sand body vertical thickness, single-well perforation vertical thickness, logging porosity, original oil saturation, clay content, logging permeability, surface crude oil viscosity, mobility, formation flow coefficient, reservoir variation coefficient, reservoir permeability surge coefficient, production pressure differential, well inclination angle, and average water cut.

4. The method according to claim 1, characterized in that, Step S20 includes: S201, the preprocessed dynamic and static factor data and initial production capacity data are analyzed using a combination of linear regression, grey relational analysis, SHAP method, Pearson correlation coefficient method, and MDI method. S202, Based on the analysis results, the controlling factors are determined by majority voting.

5. The method according to claim 1, characterized in that, Step S30 includes: S301, based on the dynamic and static factors affecting the initial production capacity of production wells in the offshore high water-cut oil reservoir and the initial production capacity data of production wells in the high water-cut oil reservoir, data samples are randomly extracted with replacement to generate multiple subset datasets; S302, Construct an RF model based on the aforementioned multiple subset datasets; S303, using the mean absolute error (MAE) as the fitness function value, the hyperparameters of the RF model are optimized using the COA algorithm to obtain the COA-RF model, and the hyperparameter combination of the COA-RF model is obtained through iterative optimization.

6. The method according to claim 5, characterized in that, The formula for calculating the Mean Absolute Error (MAE) is as follows: ； Among them, y i This is the true value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs at sea. is the predicted value of the i-th sample in the initial production capacity sample dataset of high water-cut oil reservoirs, and n is the total number of samples in the initial production capacity sample dataset of high water-cut oil reservoirs.

7. The method according to claim 5, characterized in that, The hyperparameter combinations for obtaining the COA-RF model through iterative optimization include: S3001, Initialize the crayfish population and calculate the fitness value of each crayfish; S3002, Update the location and status of individual crayfish; S3003, solve the crayfish position solution space and evaluate the fitness function of individual crayfish based on the current position; S3004. When the optimal value of the hyperparameters of the COA-RF model is found or the maximum number of iterations is reached, the iteration terminates.

8. The method according to claim 5, characterized in that, The hyperparameters of the COA-RF model are the number of trees, the maximum depth of the trees, the minimum number of samples required for a tree node to split, the minimum number of samples required for a leaf node, and the maximum number of features considered during splitting.

9. The method according to claim 1, characterized in that, The different metrics include MAE, MSE, RMSE, and R².

10. A system for predicting the initial production capacity of a production well in a high water-cut offshore oil reservoir, characterized in that, include: The data acquisition module is used to acquire dynamic and static factor data affecting the initial production capacity of production wells in offshore high water-cut oil reservoirs and initial production capacity data of production wells in high water-cut oil reservoirs, and to preprocess the dynamic and static factor data and initial production capacity data. The main control factor identification module is used to determine the main control factors by jointly analyzing the preprocessed dynamic and static factor data and initial production capacity data using multiple methods. The model building module is used to build a COA-RF model and obtain the hyperparameter combination values ​​of the COA-RF model based on the main control factors. The prediction result output module is used to train the COA-RF model based on the hyperparameter combination assignment, test the trained COA-RF model, and output the prediction result of the initial production capacity of the production well in the high water-cut reservoir. The performance evaluation module is used to evaluate the performance of the COA-RF model using different metrics based on the prediction results.

Images