Carbonate rock acid fracturing effect master control parameter analysis method and system

Abstract

The invention discloses a carbonate rock acid fracturing effect main control parameter analysis method and system, relates to the technical field of acidification, and solves the problem that main control parameters are difficult to identify in existing carbonate rock acid fracturing effect assessment. Calculating the correlation of each parameter based on the parameter data, and filtering out the parameters with extremely high correlation; executing a random forest algorithm and a recursive elimination algorithm based on the parameters to obtain a first weight and a second weight of the parameters; performing weighted average on the first weight and the second weight to obtain a comprehensive weight of the parameter; selecting parameters according to the comprehensive weight to construct a plurality of schemes, inputting the schemes into a BP neural network to simulate the yield to obtain the correlation between the actual yield and the predicted yield, and selecting the parameters contained in the scheme with the highest correlation as main control parameters of the acid fracturing effect; and analyzing carbonate rock acid fracturing effect master control parameters based on a random forest, recursive elimination and a BP neural network algorithm.

Description

Technical Field

[0001] This invention relates to the field of acidification technology, and more specifically, to a method and system for analyzing the main control parameters of acid fracturing effect on carbonate rocks. Background Technology

[0002] Carbonate reservoirs are characterized by low permeability, poor physical properties, well-developed fractures, strong heterogeneity, and low initial production. Acid fracturing is the most important technical means to improve gas well productivity. With the widespread use of acid fracturing technology, the difficulty of acid fracturing construction has increased, the success rate has decreased, and the construction cost has also risen. Therefore, the evaluation of acid fracturing effect has become particularly important.

[0003] For acid fracturing effect assessment, numerous parameters influence the effect. If the main controlling parameters of acid fracturing effect can be selected, the number of parameters to be analyzed can be reduced through dimensionality reduction while maintaining the accuracy of the acid fracturing effect assessment. However, currently, in the process of assessing the acid fracturing effect of carbonate oil and gas reservoirs, it is difficult to accurately identify and analyze the main controlling parameters affecting the acid fracturing effect.

[0004] Therefore, this application provides a method and system for analyzing the main control parameters of acid fracturing effect in carbonate rocks, thereby solving the above-mentioned problems. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for analyzing the main control parameters of acid fracturing effect in carbonate rocks, which solves the problem that it is difficult to accurately identify and analyze the main control parameters affecting the acid fracturing effect in the existing process of acid fracturing effect assessment of carbonate oil and gas reservoirs. This application analyzes the main control parameters of acid fracturing effect in carbonate rocks based on random forest algorithm, recursive elimination algorithm and BP neural network algorithm. The selected main control parameters can provide more comprehensive and accurate guidance for acid fracturing effect assessment.

[0006] This application first provides a method for analyzing the main control parameters of carbonate rock acid fracturing effect, including: S1, acquiring parameter data affecting carbonate rock acid fracturing effect, preprocessing the parameter data, calculating the correlation of each parameter based on the parameter data, and filtering out parameters with extremely strong correlation; S2, executing a random forest algorithm and a recursive elimination algorithm based on the filtered parameters to obtain the first weight and second weight of each parameter; S3, performing a weighted average of the first weight and second weight of each parameter to obtain the comprehensive weight of each parameter; S4, selecting parameters in descending order of comprehensive weight to construct multiple schemes, inputting the schemes into a BP neural network to simulate the yield and obtain the correlation between the actual yield and the predicted yield, and selecting the parameters contained in the scheme with the highest correlation as the main control parameters of acid fracturing effect.

[0007] In one possible implementation, S1 includes: selecting parameters that affect the acid fracturing effect of carbonate rocks from three aspects: engineering parameters, geological parameters, and well information parameters; collecting parameter data, deleting parameters with unique values, supplementing missing parameter data, and identifying and correcting abnormal parameter data; calculating the Pearson correlation coefficient between parameters based on the parameter data, and filtering out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

[0008] In one possible implementation, S2 includes: training a random forest model with parameters as features, assigning importance weights to each feature in the trained random forest model as the first weights of the corresponding parameters; eliminating the least important features based on the importance weights of the features, retraining the random forest model based on the remaining features until the stopping condition is met, and assigning importance weights to each feature in the trained random forest model as the second weights of the corresponding parameters.

[0009] In one possible implementation, S3 includes: normalizing the first weight and the second weight of the parameter; and performing a weighted average based on the normalized first weight and the second weight to obtain the comprehensive weight of the parameter.

[0010] In one possible implementation, S4 includes: selecting parameters to construct schemes in descending order of comprehensive weight, gradually increasing the number of selected parameters until all parameters are selected, resulting in multiple schemes; inputting the parameter data of the schemes into a BP neural network to obtain predicted yield, calculating the correlation between predicted yield and actual yield; and selecting the parameters contained in the scheme with the highest correlation as the master control parameters for acid stress effect.

[0011] This application also provides a system for analyzing the master control parameters of carbonate rock acid fracturing effect, comprising: a data processing unit for acquiring parameter data affecting the carbonate rock acid fracturing effect, preprocessing the parameter data, calculating the correlation of each parameter based on the parameter data, and filtering out parameters with extremely strong correlation; a weight analysis unit for executing a random forest algorithm and a recursive elimination algorithm based on the filtered parameters to obtain the first weight and second weight of each parameter; a comprehensive weight analysis unit for weighted averaging of the first weight and second weight of each parameter to obtain the comprehensive weight of each parameter; and a master control parameter analysis unit for selecting parameters in descending order of comprehensive weight to construct multiple schemes, inputting the schemes into a BP neural network to simulate the yield to obtain the correlation between the actual yield and the predicted yield, and selecting the parameters contained in the scheme with the highest correlation as the master control parameters of the acid fracturing effect.

[0012] In one possible implementation, the data processing unit specifically includes: a parameter selection unit, used to select parameters affecting the acid fracturing effect of carbonate rocks from three aspects: engineering parameters, geological parameters, and well information parameters; a data preprocessing unit, used to collect parameter data, delete parameters with unique values, supplement missing parameter data, and identify and correct abnormal parameter data; and a data filtering unit, used to calculate the Pearson correlation coefficient between parameters based on the parameter data, and filter out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

[0013] In one possible implementation, the weight analysis unit specifically includes: a first weight analysis unit, used to train a random forest model with parameters as features, and to assign importance weights to each feature in the trained random forest model as the first weights of the corresponding parameters; and a second weight analysis unit, used to eliminate the least important features according to the importance weights of the features, retrain the random forest model based on the remaining features until the stopping condition is met, and to assign importance weights to each feature in the trained random forest model as the second weights of the corresponding parameters.

[0014] In one possible implementation, the comprehensive weight analysis unit specifically includes: a weight normalization unit for normalizing the first weight and the second weight of the parameter; and a comprehensive weight calculation unit for performing a weighted average based on the normalized first weight and the second weight to obtain the comprehensive weight of the parameter.

[0015] In one possible implementation, the master control parameter analysis unit specifically includes: an equation construction unit, used to select parameter construction schemes in descending order of comprehensive weight, with the number of selected parameters gradually increasing until all parameters are selected, resulting in multiple schemes; a scheme prediction unit, used to input the parameter data of the schemes into a BP neural network to obtain predicted yield, and calculate the correlation between predicted yield and actual yield; and a parameter selection unit, used to select the parameters contained in the scheme with the highest correlation as the master control parameters for acid fracturing effect.

[0016] Compared with existing technologies, this application has the following advantages: The method for analyzing the main control parameters of carbonate rock acid fracturing effect provided in this application, by comprehensively applying random forest algorithm, recursive elimination algorithm and BP neural network, quantifies and analyzes the main control parameters of carbonate rock acid fracturing effect. The selected main control parameters can optimize the accuracy and reliability of acid fracturing effect assessment. Specifically, this application can select main control parameters from numerous influencing parameters, such as construction section length, reservoir coefficient, number of segments, acid fracturing process, pressure drop rate, and reservoir thickness. These parameters have been proven to have a decisive impact on acid fracturing effect. Furthermore, this application has verified through BP neural network simulation that when the number of parameters reaches 6, a correlation of up to 0.999 between predicted and actual production is achieved, and the mean square error is reduced to 0.002, which is significantly better than traditional methods. In addition, the main control parameters selected in this application can not only optimize the design of acid fracturing schemes, but also improve the production capacity and construction success rate of oil and gas wells, and reduce construction costs. It has important practical application value for acid fracturing stimulation of carbonate oil and gas reservoirs. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0018] Figure 1 A flowchart illustrating the method for analyzing the main control parameters of acid fracturing effect on carbonate rocks provided in this application embodiment;

[0019] Figure 2 This is a schematic diagram of a four-parameter Spearman correlation analysis for a lamp.

[0020] Figure 3 A scatter plot showing the fit between parameters with extremely strong correlation;

[0021] Figure 4 The graph is sorted by first weight for all parameters;

[0022] Figure 5 This is a ranking chart of geological parameters by weight.

[0023] Figure 6 This is a ranking chart of engineering parameters by their first weight.

[0024] Figure 7 The first weighted sorting chart of well information;

[0025] Figure 8 The second weighted sorting graph for all parameters;

[0026] Figure 9 This is a ranking chart of geological parameters by second weight.

[0027] Figure 10 This is a ranking chart of the second weight of engineering parameters;

[0028] Figure 11 The second weighted sorting chart for well information;

[0029] Figure 12 A structural diagram of the main control parameter analysis system for the acid fracturing effect of carbonate rocks provided in the embodiments of this application. Detailed Implementation

[0030] In the following, the terms “comprising” or “may include” as used in the various embodiments of this application indicate the presence of the claimed function, operation, or element, and do not limit the addition of one or more functions, operations, or elements. Furthermore, as used in the various embodiments of this application, the terms “comprising,” “having,” and their cognates are intended only to indicate a specific feature, number, step, operation, element, component, or combination of the foregoing, and should not be construed as primarily excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more combinations of the foregoing.

[0031] In various embodiments of this application, the expression "or" or "at least one of B and / or C" includes any combination or all combinations of the words listed simultaneously. For example, the expression "B or C" or "at least one of B and / or C" may include B, may include C, or may include both B and C.

[0032] The terms used in the various embodiments of this application (such as "first," "second," etc.) may modify various constituent elements in the various embodiments, but do not limit the corresponding constituent elements. For example, the above terms do not limit the order and / or importance of the elements. The above terms are only used for the purpose of distinguishing one element from other elements. For example, a first user device and a second user device refer to different user devices, although both are user devices. For example, without departing from the scope of the various embodiments of this application, a first element may be referred to as a second element, and similarly, a second element may also be referred to as a first element.

[0033] The terminology used in the various embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the various embodiments of this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in a generally used dictionary) are to be interpreted as having the same meaning as in the context of the relevant technical field and are not to be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this application are only for explaining this application and are not intended to limit this application.

[0035] Please see Figure 1 As shown, Figure 1 This is a flowchart illustrating the method for analyzing the main control parameters of carbonate rock acid fracturing effect provided in this application embodiment. The method includes: S1, acquiring parameter data affecting the carbonate rock acid fracturing effect, preprocessing the parameter data, calculating the correlation of each parameter based on the parameter data, and filtering out parameters with extremely strong correlation; S2, executing a random forest algorithm and a recursive elimination algorithm based on the filtered parameters to obtain the first weight and second weight of each parameter; S3, performing a weighted average of the first weight and second weight of each parameter to obtain the comprehensive weight of each parameter; S4, selecting parameters in descending order of comprehensive weight to construct multiple schemes, inputting the schemes into a BP neural network to simulate the yield and obtain the correlation between the actual yield and the predicted yield, and selecting the parameters contained in the scheme with the highest correlation as the main control parameters of the acid fracturing effect.

[0036] Specifically, to evaluate the acid fracturing effect of carbonate rocks, the parameters affecting the fracturing effect are first analyzed, and parameter data are collected and processed to ensure data accuracy and usability. Next, the correlation between parameters is calculated, and parameters with extremely strong correlations are eliminated to avoid affecting the prediction results. Then, the random forest algorithm and recursive elimination algorithm are used to perform weight (importance) analysis on the remaining parameters, obtaining the first and second weights for each parameter. These weights are then weighted and averaged to obtain the comprehensive weight of each parameter. Finally, multiple schemes are constructed based on the comprehensive weights of the parameters. For example, the first N parameters form Scheme 1, the first N+1 parameters form Scheme 2, and so on until all parameters are selected. A backpropagation neural network is used to simulate the predicted yields of these schemes, and the correlation between actual and predicted yields is calculated. The scheme with the strongest correlation is selected, as the parameters contained in this scheme are most closely related to the acid fracturing effect and can be used as the main control parameters affecting the acid fracturing effect of carbonate rocks.

[0037] The improvement of this application lies in providing a clear mechanism for identifying key control parameters, quantitatively analyzing the impact of each parameter on post-fracturing production, and thus screening out the key control parameters that play a decisive role. Specifically, by comprehensively applying random forest algorithm, recursive elimination algorithm and BP neural network, the parameters affecting the acid fracturing effect of carbonate rocks are analyzed in depth and their weights are evaluated, accurately identifying the key parameters affecting the acid fracturing effect as key control parameters. Based on this, the acid fracturing scheme can be optimized, the success rate of acid fracturing measures and the production capacity of oil and gas wells can be improved, thereby providing scientific and effective decision support for acid fracturing stimulation of carbonate oil and gas reservoirs.

[0038] In one possible implementation, S1 includes: selecting parameters that affect the acid fracturing effect of carbonate rocks from three aspects: engineering parameters, geological parameters, and well information parameters; collecting parameter data, deleting parameters with unique values, supplementing missing parameter data, and identifying and correcting abnormal parameter data; calculating the Pearson correlation coefficient between parameters based on the parameter data, and filtering out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

[0039] Specifically, step S1 involves acquiring parameter data affecting the acid fracturing effect of carbonate rocks, preprocessing the parameter data, calculating the correlation of each parameter based on the parameter data, and filtering out parameters with extremely strong correlations. First, parameters affecting the acid fracturing effect of carbonate rocks are selected from three aspects: engineering parameters, geological parameters, and well information parameters. Geological parameters include: reservoir coefficient, reservoir thickness, Young's modulus, porosity, brittleness index, gas saturation, total hydrocarbon content, Poisson's ratio, minimum horizontal principal stress, permeability, and reservoir space type. Engineering parameters include: acid strength, pressure drop rate, acid fracturing process, acid volume, number of stages, discharge rate, pump shutdown pressure, and acid type. Well information parameters include: construction section length, well depth, cementing quality, and well area.

[0040] Secondly, parameter data is collected, parameters with unique values ​​are removed, missing parameter data is supplemented, and abnormal parameter data is identified and corrected. Details are as follows: A. Unique Value Analysis: If a parameter has only one value, it is useless for the analysis of the master control parameters. In other words, parameters with unique values ​​are useless and redundant for data mining, so such parameters are removed. B. Invalid Value Analysis: When parameters have missing data and the missing rate is low, data can be supplemented using methods such as median or mean. When the missing rate is high, random forest multiple imputation can be used to supplement data. Specifically, each missing parameter is used as a function of other parameters to construct a random forest model, and the missing data is estimated based on the other parameters and this function. C. Multivariate Analysis: Spearman correlation coefficient is used to evaluate the correlation between parameters. Spearman correlation coefficient is not sensitive to outliers and can identify parameters with extremely strong correlation (correlation > 0.9) and strongly correlated parameters (correlation 0.8–0.9). These parameters can be used to identify outliers in the data through scatter plot fitting trends, and then the outliers can be corrected.

[0041] Finally, the Pearson correlation coefficient between the parameters is calculated based on the parameter data. For multiple parameters with Pearson correlation coefficients in the extremely strong correlation range, only one is retained. Strongly correlated parameters can affect the model's interpretation results; therefore, the Pearson correlation coefficient is used to measure parameter correlation. The formula for calculating the Pearson correlation coefficient is:

[0042]

[0043] In the formula, ρ > 0 indicates a positive correlation, ρ < 0 indicates a negative correlation, and ρ = 0 indicates zero correlation. Parameters with Pearson correlation coefficients in the extremely strong correlation range (|ρ| > 0.9) are filtered out.

[0044] In one possible implementation, S2 includes: training a random forest model with parameters as features, assigning importance weights to each feature in the trained random forest model as the first weights of the corresponding parameters; eliminating the least important features based on the importance weights of the features, retraining the random forest model based on the remaining features until the stopping condition is met, and assigning importance weights to each feature in the trained random forest model as the second weights of the corresponding parameters.

[0045] Specifically, step S2 executes the Random Forest algorithm and the Recursive Elimination algorithm based on the filtered parameters to obtain the first and second weights of each parameter. First, the Random Forest algorithm is executed, constructing multiple decision trees to evaluate the importance of features. Each tree randomly selects a subset of features during training, and then aggregates the feature importance scores based on the training results of these trees, thus obtaining the first weight of each parameter. Next, the Recursive Elimination algorithm is executed, progressively removing the least important (or most important) features and retraining the model until all features have been filtered, thus determining the second weight of each parameter. This process can be viewed as constructing an inverted tree, where the path from the root to the leaf node represents the decision rules used to classify and predict new data. Based on these decision rules, new data is classified and predicted, selecting the best (or worst) features. The model is then repeatedly constructed based on the remaining features until all features have been filtered, ultimately obtaining the importance score of each feature, representing the feature's contribution to the model, thus determining the second weight of each parameter.

[0046] In one possible implementation, S3 includes: normalizing the first weight and the second weight of the parameter; and performing a weighted average based on the normalized first weight and the second weight to obtain the comprehensive weight of the parameter.

[0047] Specifically, step S3 involves taking a weighted average of the first and second weights of each parameter to obtain the comprehensive weight of each parameter. To ensure uniformity of magnitude, the weights obtained from the random forest algorithm and the recursive elimination algorithm are normalized, and then the comprehensive weight of the parameters is calculated using a weighted average method. Finally, all parameters are sorted according to their comprehensive weights.

[0048] In one possible implementation, S4 includes: selecting parameters to construct schemes in descending order of comprehensive weight, gradually increasing the number of selected parameters until all parameters are selected, resulting in multiple schemes; inputting the parameter data of the schemes into a BP neural network to obtain predicted yield, calculating the correlation between predicted yield and actual yield; and selecting the parameters contained in the scheme with the highest correlation as the master control parameters for acid stress effect.

[0049] Specifically, step S4 involves selecting parameters to construct multiple schemes according to their comprehensive weights from high to low. These schemes are then input into a BP neural network to simulate yield, obtaining the correlation between actual and predicted yields. The parameters contained in the scheme with the highest correlation are selected as the master control parameters for acid pressing effectiveness. To avoid post-pressurization yield data affecting the parameters and causing the model to fail to converge to the preset accuracy, a minimum of four parameters is specified during scheme construction; that is, each scheme must contain at least four parameters. In this way, starting with the scheme containing the fewest parameters, the number of parameters is gradually increased for modeling and simulation. Simultaneously, reserved sample data is used to verify the correlation between the model's predicted and actual yields. Finally, the parameters contained in the scheme with the highest correlation are selected as the master control parameters affecting acid pressing effectiveness.

[0050] It is understood that the method for analyzing the main control parameters of carbonate rock acid fracturing effect provided in this application, by comprehensively applying the random forest algorithm, recursive elimination algorithm, and BP neural network, quantifies and analyzes the main control parameters affecting the carbonate rock acid fracturing effect. These main control parameters can optimize the accuracy and reliability of acid fracturing effect assessment.

[0051] The following uses reservoir data as a verification example, combined with... Figures 2-11 Verify the effectiveness of the method in this application.

[0052] Step 1: Select 26 parameters, including engineering parameters, geological parameters, and well information parameters of the fourth section of the Penglai gas field reservoir, and preprocess the parameter data.

[0053] Specifically, no parameter data has a unique value. For parameters such as minimum horizontal principal stress and total hydrocarbon content, the missing rate is less than 15%, and the median value of these parameters can be used for imputation. For parameters such as cementing quality, Young's modulus, Poisson's ratio, brittleness index, and reservoir type, the missing rate is higher, and random forest multiple imputation can be used for imputation. The Spearman correlation coefficient is used to evaluate the correlation between parameters, such as... Figure 2 As shown. Figure 2 The horizontal and vertical axes represent parameters, and colors indicate the magnitude of the correlation: green represents a positive correlation, red represents a negative correlation, and darker colors indicate a stronger correlation. Based on Figure 2We selected parameters with extremely strong correlation (correlation > 0.9) and parameters with strong correlation (correlation 0.8–0.9), as shown in Table 1. These parameter combinations can be used to identify outliers in the data by fitting the trend of scatter plots.

[0054] Table 1. Parametric Correlation Analysis Based on Spearman Correlation Coefficient

[0055] Parameter 1 Parameter 2 Correlation coefficient Total liquid volume Acid dosage 0.99

[0056] Figure 3 The scatter plot shows a strong linear correlation between the two parameters. Data points far from the fitted line can be considered outliers, possibly due to input errors or other reasons. After identification, these outliers are corrected to the correct values ​​based on the fitted line.

[0057] The Pearson correlation coefficient method was used to filter out parameter data with extremely strong correlations, as shown in Table 2.

[0058] Table 2. Parametric Correlation Analysis Based on Pearson Correlation Coefficient

[0059] Block Parameter 1 Parameter 2 Correlation coefficient Delete parameters Four sections of lamp Acid dosage Total liquid volume 0.99 Total liquid volume

[0060] Step 2: Using the random forest algorithm and the recursive feature elimination algorithm, determine the overall sorting of the four parameters of the lamp and the first weight sorting and second weight sorting of each type of parameter.

[0061] like Figure 4 , 5 As shown in Figures 6 and 7, among the parameters affecting the acid fracturing effect in Block 4, the geological parameters have the highest weight of 0.31, the construction parameters have the highest weight of 0.34, and the well information parameters have the highest weight of 0.35. The number of segments in the engineering parameters and the reservoir coefficient in the geological parameters have the greatest impact on the acid fracturing effect.

[0062] like Figure 8 , 9 As shown in Figures 10 and 11, after executing the recursive feature elimination algorithm, among the parameters affecting the acid fracturing effect in Block 4, the second weight of the geological parameter is 208, the second weight of the engineering parameter is 146, and the second weight of the well information parameter is 60. Among the geological parameters, the reservoir coefficient, and among the engineering parameters, the acid strength, have the greatest impact on the acid fracturing effect.

[0063] Step 3: The random forest algorithm and recursive feature elimination algorithm are used to analyze the parameters affecting acid pressure in the four lamp sections. A weighted average method is used to calculate the comprehensive weight of the parameters, and the parameters of the four lamp sections are ranked based on the comprehensive weight, as shown in Table 3.

[0064] Table 3. Ranking by Comprehensive Weight

[0065] Construction section length 0.20 Storage coefficient 0.09 Number of segments 0.08 acid frosting process 0.07 Pressure drop rate 0.06 reservoir thickness 0.05 gas saturation 0.05 Acid dosage 0.05 Acid strength 0.04 Brittleness Index 0.04 Penetration 0.03 Young's modulus 0.03 Porosity 0.03 Total hydrocarbon content 0.03 Displacement 0.03 Poisson's ratio 0.02 well deep 0.02 Minimum horizontal principal stress 0.02 Cementing quality 0.02 Pump stop pressure 0.01 Acid type 0.01 Storage space types 0.01 well area 0.01

[0066] Step 4: Train the schemes containing different data parameters using a BP neural network. The results are shown in Table 4 below.

[0067] Table 4 Output results of the BP neural network

[0068] Number of parameters Optimal number of hidden layers Mean square error Correlation coefficient 4 10 0.14 0.71 5 5 0.097 0.81 6 10 0.002 0.999 7 11 0.067 0.67 8 12 0.035 0.601 9 6 0.075 0.859 10 4 0.006 0.741 11 4 0.026 0.859 12 6 0.034 0.709 13 5 0.239 0.842 14 7 0.0264 0.918 15 13 0.0618 0.831 16 12 0.0224 0.860 17 8 0.0668 0.811 18 5 0.1239 0.737 19 6 0.0568 0.768 20 5 0.0641 0.657 21 8 0.145 0.415 22 7 0.0626 0.663 23 13 0.0344 0.597

[0069] Based on the production prediction model established by the BP neural network, Table 4 shows that the correlation is best when the number of parameters reaches 6. Therefore, the main control parameters of the four-stage lamp are: construction section length, reservoir coefficient, number of segments, acid fracturing process, pressure drop rate, and reservoir thickness.

[0070] As can be seen, this application can screen out the main controlling parameters from numerous influencing parameters, such as the construction section length, reservoir coefficient, number of segments, acid fracturing process, pressure drop rate, and reservoir thickness. These parameters have been proven to have a decisive impact on the acid fracturing effect. Furthermore, this application, through BP neural network simulation, has demonstrated that when the number of parameters reaches six, a correlation of up to 0.999 between predicted and actual production is achieved, with the mean square error reduced to 0.002, significantly outperforming traditional methods. In addition, this application can not only optimize the design of acid fracturing schemes but also improve the productivity and construction success rate of oil and gas wells while reducing construction costs, thus possessing significant practical application value for acid fracturing stimulation of carbonate oil and gas reservoirs.

[0071] Please see Figure 12 As shown, Figure 12 This is a structural diagram of the main control parameter analysis system for carbonate rock acid fracturing effect provided in this application embodiment. It includes: a data processing unit for acquiring parameter data affecting the carbonate rock acid fracturing effect, preprocessing the parameter data, calculating the correlation of each parameter based on the parameter data, and filtering out parameters with extremely strong correlations; a weight analysis unit for executing a random forest algorithm and a recursive elimination algorithm based on the filtered parameters to obtain the first weight and second weight of each parameter; a comprehensive weight analysis unit for weighted averaging of the first weight and second weight of each parameter to obtain the comprehensive weight of each parameter; and a main control parameter analysis unit for selecting parameters in descending order of comprehensive weight to construct multiple schemes, inputting the schemes into a BP neural network to simulate yield and obtain the correlation between actual yield and predicted yield, and selecting the parameters contained in the scheme with the highest correlation as the main control parameters for the acid fracturing effect.

[0072] In one possible implementation, the data processing unit specifically includes: a parameter selection unit, used to select parameters affecting the acid fracturing effect of carbonate rocks from three aspects: engineering parameters, geological parameters, and well information parameters; a data preprocessing unit, used to collect parameter data, delete parameters with unique values, supplement missing parameter data, and identify and correct abnormal parameter data; and a data filtering unit, used to calculate the Pearson correlation coefficient between parameters based on the parameter data, and filter out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

[0073] In one possible implementation, the weight analysis unit specifically includes: a first weight analysis unit, used to train a random forest model with parameters as features, and to assign importance weights to each feature in the trained random forest model as the first weights of the corresponding parameters; and a second weight analysis unit, used to eliminate the least important features according to the importance weights of the features, retrain the random forest model based on the remaining features until the stopping condition is met, and to assign importance weights to each feature in the trained random forest model as the second weights of the corresponding parameters.

[0074] In one possible implementation, the comprehensive weight analysis unit specifically includes: a weight normalization unit for normalizing the first weight and the second weight of the parameter; and a comprehensive weight calculation unit for performing a weighted average based on the normalized first weight and the second weight to obtain the comprehensive weight of the parameter.

[0075] In one possible implementation, the master control parameter analysis unit specifically includes: an equation construction unit, used to select parameter construction schemes in descending order of comprehensive weight, with the number of selected parameters gradually increasing until all parameters are selected, resulting in multiple schemes; a scheme prediction unit, used to input the parameter data of the schemes into a BP neural network to obtain predicted yield, and calculate the correlation between predicted yield and actual yield; and a parameter selection unit, used to select the parameters contained in the scheme with the highest correlation as the master control parameters for acid fracturing effect.

[0076] It should be noted that the carbonate rock acid fracturing effect master control parameter analysis system provided in this application is used to implement the above-mentioned carbonate rock acid fracturing effect master control parameter analysis method, which corresponds one-to-one with the above method and has the corresponding technical effect, so it will not be elaborated further.

[0077] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing the main control parameters of acid fracturing effect on carbonate rocks, characterized in that, include: S1. Obtain parameter data that affect the acid fracturing effect of carbonate rocks, preprocess the parameter data, calculate the correlation of each parameter based on the parameter data, and filter out parameters with extremely strong correlation. S2. Based on the filtered parameters, execute the random forest algorithm and the recursive elimination algorithm to obtain the first weight and second weight of each parameter; S3. Take the weighted average of the first and second weights of each parameter to obtain the comprehensive weight of each parameter; S4. Select parameters in descending order of comprehensive weight to construct multiple schemes. Input the schemes into a BP neural network to simulate the yield and obtain the correlation between the actual yield and the predicted yield. Select the parameters contained in the scheme with the highest correlation as the main control parameters of the acid stress effect.

2. The method for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 1, characterized in that, S1 includes: The parameters affecting the acid fracturing effect of carbonate rocks are selected from three aspects: engineering parameters, geological parameters, and well information parameters. Collect parameter data, delete parameters with unique values, supplement missing parameter data, and identify and correct abnormal parameter data; Calculate the Pearson correlation coefficient between parameters based on the parameter data, and filter out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

3. The method for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 1, characterized in that, S2 include: The parameters are used as features to train a random forest model, and the importance weights assigned to each feature by the trained random forest model are used as the first weights of the corresponding parameters. Eliminate the least important features based on their importance weights, retrain the random forest model based on the remaining features until the stopping condition is met, and use the importance weights assigned to each feature in the trained random forest model as the second weights of the corresponding parameters.

4. The method for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 1, characterized in that, S3 includes: Normalize the first and second weights of the parameters; The comprehensive weight of the parameters is obtained by weighting the first and second weights after normalization.

5. The method for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 1, characterized in that, S4 includes: The parameters are selected in descending order of comprehensive weight to construct the scheme. The number of selected parameters is gradually increased until all parameters are selected, resulting in multiple schemes. The parameter data of the scheme is input into a BP neural network to obtain the predicted output, and the correlation between the predicted output and the actual output is calculated. The parameters included in the scheme with the highest correlation were selected as the main control parameters for acid fracturing effect.

6. A system for analyzing the main control parameters of acid fracturing effect in carbonate rocks, characterized in that, include: The data processing unit is used to acquire parameter data that affect the acid fracturing effect of carbonate rocks, preprocess the parameter data, calculate the correlation of each parameter based on the parameter data, and filter out parameters with extremely strong correlation. The weight analysis unit is used to execute the random forest algorithm and the recursive elimination algorithm based on the filtered parameters to obtain the first weight and the second weight of each parameter. The comprehensive weight analysis unit is used to calculate the comprehensive weight of each parameter by weighting the first and second weights of each parameter. The master control parameter analysis unit is used to select parameters in descending order of comprehensive weight to construct multiple schemes. The schemes are input into a BP neural network to simulate the yield and obtain the correlation between the actual yield and the predicted yield. The parameters contained in the scheme with the highest correlation are selected as the master control parameters for acid stress effect.

7. The system for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 6, characterized in that, The data processing unit specifically includes: The parameter selection unit is used to select parameters that affect the acid fracturing effect of carbonate rocks from three aspects: engineering parameters, geological parameters, and well information parameters. The data preprocessing unit is used to collect parameter data, delete parameters with unique values, supplement missing parameter data, and identify and correct abnormal parameter data. The data filtering unit is used to calculate the Pearson correlation coefficient between parameters based on the parameter data, and to filter out parameters whose Pearson correlation coefficient is in the range of extremely strong correlation.

8. The system for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 6, characterized in that, The weight analysis unit specifically includes: The first weight analysis unit is used to train a random forest model with parameters as features, and the importance weights assigned to each feature by the trained random forest model are used as the first weights of the corresponding parameters. The second weight analysis unit is used to eliminate the least important features based on the importance weights of the features, retrain the random forest model based on the remaining features until the stopping condition is met, and use the importance weights assigned to each feature by the trained random forest model as the second weights of the corresponding parameters.

9. The system for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 6, characterized in that, The comprehensive weight analysis unit specifically includes: The weight normalization unit is used to normalize the first and second weights of the parameters; The comprehensive weight calculation unit is used to perform a weighted average based on the normalized first and second weights to obtain the comprehensive weight of the parameters.

10. The system for analyzing the main control parameters of acid fracturing effect in carbonate rocks according to claim 6, characterized in that, The main control parameter analysis unit specifically includes: The equation building unit is used to select parameters to build schemes in descending order of comprehensive weight. The number of selected parameters gradually increases until all parameters are selected, resulting in multiple schemes. The program prediction unit is used to input the program's parameter data into a BP neural network to obtain the predicted output and calculate the correlation between the predicted output and the actual output. The parameter selection unit is used to select the parameters contained in the scheme with the highest correlation as the main control parameters for acid fracturing effect.

Images