Method and device for predicting fracture pressure and production classification of tight oil reservoirs

Abstract

This invention proposes a method and apparatus for predicting fracture pressure and production classification in tight oil reservoirs, relating to the field of reservoir exploration technology. The method includes: acquiring tight oil horizontal well data; extracting factors affecting fracture pressure from the tight oil horizontal well data, inputting these factors into a tight oil horizontal well fracture pressure prediction model to obtain the fracture pressure; extracting factors affecting tight oil production from the tight oil horizontal well data, inputting these factors into a tight oil production prediction model to obtain the tight oil production; classifying the tight oil horizontal wells according to the tight oil production to obtain tight oil production classification results; and generating a tight oil horizontal well development strategy based on the tight oil horizontal well fracture pressure and tight oil production classification results.

Description

Technical Field

[0001] This invention relates to the field of reservoir exploration technology, and more particularly to a method and apparatus for predicting the fracture pressure and production classification of tight oil reservoirs. Background Technology

[0002] This section is intended to provide background or context for embodiments of the present invention. The description herein is not intended to imply that it is prior art simply because it is included in this section.

[0003] When designing hydraulic fracturing for tight oil wells, a challenge arises from abnormal fracture pressure values. These abnormalities often lead to abnormal operating pressures, resulting in operation failure or unexpected incidents. Therefore, fracture pressure is a crucial parameter in hydraulic fracturing design and a vital reference factor for optimizing operating parameters. Due to the complex geological structures and other conditions in different blocks, traditional fracture pressure calculation formulas are no longer sufficient. Although regional fitting formulas have been derived from these formulas, their accuracy and rationality are significantly flawed. Furthermore, inaccurate fracture pressure calculations cannot provide accurate analytical results for tight oil production.

[0004] In summary, there is an urgent need for a technical solution that can overcome the shortcomings of existing technologies, accurately predict the fracture pressure and production classification of tight oil reservoirs, and provide strong technical support for the development of tight oil wells. Summary of the Invention

[0005] To address the problems existing in current technologies, this invention proposes a method and apparatus for predicting the fracture pressure and production classification of tight oil reservoirs. This invention establishes a mathematical model for predicting the fracture pressure of tight oil wells by combining actual geological and production data of the target block with theoretical and practical analysis. This model can be used to predict fracture pressure, thereby guiding hydraulic fracturing design, determining whether the fracture pressure can break up each cluster, and establishing a machine learning-based method for predicting tight oil production classification based on the ability to break up clusters, thus verifying the exploitability. The overall scheme establishes a complete predictive model to provide guidance for subsequent tight oil well development, which is of great significance for improving tight oil well production, reducing production costs, and promoting technological innovation in actual production.

[0006] In a first aspect of the present invention, a predictive method for classifying fracture pressure and production in tight oil reservoirs is proposed, the method comprising:

[0007] Collect data from tight oil horizontal wells;

[0008] Factors influencing the fracture pressure of tight oil horizontal wells are extracted from the tight oil horizontal well data. These factors are then input into a tight oil horizontal well fracture pressure prediction model to obtain the fracture pressure. The tight oil horizontal well fracture pressure prediction model is trained using the following method: multiple initial models are constructed using various machine learning algorithms; a first training set and a first test set are constructed using sample data of factors influencing the fracture pressure of tight oil horizontal wells; the multiple initial models are trained using the first training set and tested using the first test set; and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model.

[0009] Factors affecting tight oil production are extracted from the tight oil horizontal well data, and these factors are input into a tight oil production prediction model to obtain the tight oil production. The tight oil production prediction model is trained using the following method: multiple initial models are constructed using various prediction algorithms; a second training set and a second test set are constructed using sample data of factors affecting tight oil production; the second training set is used to train the multiple initial models, and the second test set is used to test them; the model is selected based on the test results to obtain a tight oil horizontal well fracture pressure prediction model.

[0010] Tight oil horizontal wells are classified according to their tight oil production, and a tight oil production classification result is obtained. A tight oil horizontal well development strategy is generated based on the fracture pressure and tight oil production classification result of the tight oil horizontal well.

[0011] In a second aspect of the present invention, a predictive device for classifying fracture pressure and production in tight oil reservoirs is provided, the device comprising:

[0012] The data acquisition module is used to acquire data from tight oil horizontal wells;

[0013] A tight oil horizontal well fracture pressure prediction module is used to extract factors affecting the fracture pressure of the tight oil horizontal well from the tight oil horizontal well data, input the factors affecting the fracture pressure of the tight oil horizontal well into the tight oil horizontal well fracture pressure prediction model, and obtain the fracture pressure of the tight oil horizontal well; wherein, the tight oil horizontal well fracture pressure prediction model is trained by the following method: multiple initial models are constructed using various machine learning algorithms, a first training set and a first test set are constructed using sample data of factors affecting the fracture pressure of the tight oil horizontal well, multiple initial models are trained using the first training set, and tested using the first test set, and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model;

[0014] The tight oil production prediction module is used to extract factors affecting tight oil production from the tight oil horizontal well data, input the factors affecting tight oil production into the tight oil production prediction model, and obtain the tight oil production. The tight oil production prediction model is trained using the following method: multiple initial models are constructed using various prediction algorithms; a second training set and a second test set are constructed using sample data of factors affecting tight oil production; the multiple initial models are trained using the second training set, and tested using the second test set; the model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model.

[0015] The development strategy generation module is used to classify tight oil horizontal wells according to the tight oil production, obtain tight oil production classification results, and generate a tight oil horizontal well development strategy based on the tight oil horizontal well fracture pressure and tight oil production classification results.

[0016] In a third aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for predicting the classification of fracture pressure and production in tight oil reservoirs.

[0017] In a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for predicting the classification of fracture pressure and production in tight oil reservoirs.

[0018] In a fifth aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements a method for predicting the classification of fracture pressure and production in tight oil reservoirs.

[0019] The proposed method and apparatus for predicting fracture pressure and production classification in tight oil reservoirs, through screening the main controlling factors affecting fracture pressure and production classification, constructs a fracture pressure prediction model based on multiple machine learning algorithms, selects the model most suitable for the complex geological conditions of different blocks, and further employs multiple prediction algorithms to construct a production prediction model to achieve production classification prediction. This can effectively distinguish between high-yield wells and low-yield wells, providing a reasonable development strategy for horizontal wells in tight oil, and providing precise technical support for fracturing design and benefit evaluation in tight oil development. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic flowchart of a method for predicting the fracture pressure and production classification of tight oil reservoirs according to an embodiment of the present invention.

[0022] Figure 2 This is a schematic diagram of a data preprocessing process according to an embodiment of the present invention.

[0023] Figure 3 This is a schematic flowchart illustrating an embodiment of the present invention for extracting factors affecting the fracture pressure of tight oil horizontal wells.

[0024] Figure 4 This is a schematic diagram of the random forest importance ranking results when processing factors affecting the fracture pressure of tight oil horizontal wells according to an embodiment of the present invention.

[0025] Figure 5 This is a schematic diagram of the correlation analysis when processing factors affecting tight oil production according to an embodiment of the present invention.

[0026] Figure 6 This is a schematic diagram of the training process for a tight oil horizontal well fracture pressure prediction model according to an embodiment of the present invention.

[0027] Figure 7 This is a schematic diagram comparing the actual value and the predicted value of GA-BP in one embodiment of the present invention.

[0028] Figure 8 This is a schematic diagram comparing the actual value and the predicted value of SVR in one embodiment of the present invention.

[0029] Figure 9 This is a schematic diagram comparing the actual value and the predicted value of XGBoost in an embodiment of the present invention.

[0030] Figure 10 This is a schematic diagram comparing the prediction results of an embodiment of the present invention.

[0031] Figure 11 This is a schematic diagram of a process for extracting factors affecting tight oil production from tight oil horizontal well data according to an embodiment of the present invention.

[0032] Figure 12 This is a schematic diagram of the random forest importance ranking results when processing factors affecting tight oil production according to an embodiment of the present invention.

[0033] Figure 13This is a schematic diagram of the correlation analysis when processing factors affecting tight oil production according to an embodiment of the present invention.

[0034] Figure 14 This is a schematic diagram of the training process of a tight oil production prediction model according to an embodiment of the present invention.

[0035] Figure 15 This is a schematic diagram of SVC parameter optimization according to an embodiment of the present invention.

[0036] Figure 16 This is a schematic diagram of RF parameter optimization according to an embodiment of the present invention.

[0037] Figure 17 This is a schematic diagram of AdaBoost parameter optimization according to an embodiment of the present invention.

[0038] Figure 18 This is a schematic diagram of XGBoost parameter optimization according to an embodiment of the present invention.

[0039] Figure 19 This is a schematic diagram of an SVC confusion matrix according to an embodiment of the present invention.

[0040] Figure 20 This is a schematic diagram of an RF confusion matrix according to an embodiment of the present invention.

[0041] Figure 21 This is a schematic diagram of an AdaBoost confusion matrix according to an embodiment of the present invention.

[0042] Figure 22 This is a schematic diagram of an XGBoost confusion matrix according to an embodiment of the present invention.

[0043] Figure 23 This is a schematic diagram of the technical route of an embodiment of the present invention.

[0044] Figure 24 This is a schematic diagram of the architecture of a predictive device for classifying fracture pressure and production in tight oil reservoirs according to an embodiment of the present invention.

[0045] Figure 25 This is a schematic diagram of a computer device structure according to an embodiment of the present invention. Detailed Implementation

[0046] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0047] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, this disclosure can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0048] According to an embodiment of the present invention, a method and apparatus for predicting the fracture pressure and production classification of tight oil reservoirs are proposed, which relates to the field of oil reservoir exploration technology.

[0049] The principles and spirit of the present invention will be explained in detail below with reference to several representative embodiments.

[0050] Figure 1 This is a schematic flowchart of a method for predicting fracture pressure and production classification in tight oil reservoirs according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0051] S101, collecting data from tight oil horizontal wells;

[0052] S102, extract the factors affecting the fracture pressure of the tight oil horizontal well from the tight oil horizontal well data, input the factors affecting the fracture pressure of the tight oil horizontal well into the tight oil horizontal well fracture pressure prediction model, and obtain the fracture pressure of the tight oil horizontal well; wherein, the tight oil horizontal well fracture pressure prediction model is trained according to the following method: construct multiple initial models through various machine learning algorithms, construct a first training set and a first test set using sample data of factors affecting the fracture pressure of the tight oil horizontal well, train multiple initial models using the first training set, test using the first test set, select a model based on the test results, and obtain the tight oil horizontal well fracture pressure prediction model;

[0053] S103, extract the factors affecting tight oil production from the tight oil horizontal well data, input the factors affecting tight oil production into the tight oil production prediction model, and obtain the tight oil production; wherein, the tight oil production prediction model is trained according to the following method: construct multiple initial models through various prediction algorithms, construct a second training set and a second test set using sample data of factors affecting tight oil production, train multiple initial models using the second training set, test using the second test set, select a model based on the test results, and obtain the tight oil horizontal well fracture pressure prediction model;

[0054] S104. Tight oil horizontal wells are classified according to the tight oil production to obtain tight oil production classification results. Tight oil horizontal well development strategies are generated based on the fracture pressure and tight oil production classification results of the tight oil horizontal wells.

[0055] The proposed method for predicting fracture pressure and production classification in tight oil reservoirs involves screening the main controlling factors affecting fracture pressure and production classification, constructing a fracture pressure prediction model based on multiple machine learning algorithms, selecting the model best suited to the complex geological conditions of different blocks, and further employing multiple prediction algorithms to construct a production prediction model to achieve production classification prediction. This method can effectively distinguish between high-yield and low-yield wells, providing a reasonable development strategy for horizontal wells in tight oil reservoirs, and offering precise technical support for fracturing design and benefit assessment in tight oil development.

[0056] To provide a clearer explanation of the prediction method for fracture pressure and production classification in tight oil reservoirs, each step will be explained in detail below.

[0057] In one embodiment, S101, tight oil horizontal well data is acquired.

[0058] Due to anomalies and missing data, data preprocessing is necessary. For details, please refer to [link / reference needed]. Figure 2 After acquiring data from tight oil horizontal wells, the method also includes:

[0059] S201, preprocess the tight oil horizontal well data, fill in missing data, remove outliers and fill in missing data accordingly;

[0060] S202. Based on the pre-processed tight oil horizontal well data, the factors affecting the fracture pressure and production of tight oil horizontal wells are initially selected. The initial selection method is as follows: the factors are initially selected through manual investigation, analysis and summarization; the logging, fracturing and drainage data of different blocks of tight oil horizontal wells are analyzed and sorted out to eliminate factors that are not involved in the blocks and factors with abnormal data.

[0061] In practical applications, fracture pressure prediction and production classification forecasting are based on a large amount of feature data. However, due to incompleteness, inconsistency, and significant missing data, direct data mining is impossible, or the mining results are poor. To improve the quality of data mining, data preprocessing is necessary. Research on fracture pressure prediction is conducted at the cluster level, while research on production classification is conducted at the well level. Taking a target area as an example, by collecting and organizing field data on tight oil in the target area, 102 clusters of logging data samples and 295 well production data samples can be obtained.

[0062] Missing values ​​can arise from a lack of measurement or failure to record measurements during construction. The main methods for handling missing values ​​are imputation and reasonable summarization. Anomalies in the drainage data are caused by human factors, equipment malfunctions, and environmental factors, such as weather, mechanical failures, or mismanagement. These factors are unpredictable and have a significant impact on the pressure of tight gas rupture. Based on drainage pattern analysis, anomalies caused by uncertain factors during the drainage process are not used; instead, they are removed and imputed as missing data.

[0063] Before feature extraction, a preliminary selection of factors influencing the fracture pressure and production classification of tight oil horizontal wells was conducted, primarily through the following two approaches: 1. Literature review: Extensive literature review and analysis of numerous scholars' research were conducted to initially select some parameters. 2. Analysis and organization of logging, fracturing, and drainage data from different blocks of tight oil horizontal wells. Factors not present in some blocks and those with data anomalies were eliminated.

[0064] In one embodiment, S102, factors affecting the fracture pressure of the tight oil horizontal well are extracted from the tight oil horizontal well data, and the factors affecting the fracture pressure of the tight oil horizontal well are input into the tight oil horizontal well fracture pressure prediction model to obtain the fracture pressure of the tight oil horizontal well; wherein, the tight oil horizontal well fracture pressure prediction model is trained according to the following method: multiple initial models are constructed using various machine learning algorithms, a first training set and a first test set are constructed using sample data of factors affecting the fracture pressure of the tight oil horizontal well, multiple initial models are trained using the first training set, and tested using the first test set, and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model.

[0065] refer to Figure 3 The specific process for extracting factors affecting fracture pressure in tight oil horizontal wells from the tight oil horizontal well data is as follows:

[0066] S301, For the tight oil horizontal well data, all factors affecting the fracture pressure of tight oil horizontal wells are obtained through random forest importance algorithm and grey relational analysis algorithm, and their influence is comprehensively ranked.

[0067] S302, by stepwise regression analysis to eliminate factors that did not meet the correlation with fracture pressure in tight oil horizontal wells, the main controlling factors affecting fracture pressure in tight oil horizontal wells were obtained.

[0068] The main controlling factors affecting the fracture pressure of tight oil horizontal wells include at least: sonic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, and maximum principal stress.

[0069] Random Forest is an ensemble learning algorithm based on a classification and regression tree (CRAT) decision tree. The base predictor is suitable for processing high-dimensional datasets with highly correlated features because it is insensitive to collinearity of features. Random Forest determines the importance of different variables by randomly replacing variables in the sample data and then calculating the decrease in prediction accuracy. The mean squared error (MSE) is used to describe the importance of variables, representing the magnitude of the increase in model prediction accuracy.

[0070] In grey relational analysis, "grey" represents that the information obtained about the objects or systems may not be entirely usable or entirely clear, while "relationship" indicates that there is a certain connection between the objects or systems. Since grey relational analysis quantifies comparable factors, the degree of correlation can be calculated. The specific steps are as follows:

[0071] 1. Determine the analysis sequence. The reference sequence can be used to obtain the characteristics of the target subject, while the comparison sequence can be used to obtain the characteristics of the reference subject. Therefore, it is necessary to first determine the reference sequence and the comparison sequence. The reference sequence refers to the part of the data sequence that reflects the behavioral characteristics of the system.

[0072] 2. Dimensionless Variable Reduction. To eliminate the possibility of drawing incorrect conclusions due to differing meanings and units among various factors in an indicator system, it is necessary to perform dimensionless data reduction to maintain consistency. Methods for eliminating dimensions include initialization and mean normalization.

[0073] 3. Calculate the correlation coefficient and correlation degree. The correlation coefficient represents the degree of correlation between the reference series and the comparison series at different times (each point in the curve). The correlation degree is calculated based on the correlation coefficient.

[0074] 4. Sort the correlation scores by magnitude.

[0075] Taking a specific target block as an example, 16 factors affecting rupture pressure were summarized based on 102 clusters of data from the target block. Then, random forest was used to rank the importance of these 16 factors. The results of the random forest importance ranking of all factors are shown in Table 1.

[0076] Table 1. Ranking of factors affecting fracture pressure in tight oil horizontal wells

[0077]

[0078] Figure 4This is a schematic diagram illustrating the random forest importance ranking results when processing factors affecting fracture pressure in tight oil horizontal wells according to an embodiment of the present invention. Combined with... Figure 4 The ranking of importance using random forests shows that the factors with the highest scores for tight oil fracture pressure, obtained through random forest analysis, are, in descending order: sonic transit time, Poisson's ratio, density, depth, brittleness index, natural gamma, minimum principal stress, Young's modulus, and wellhead pressure. The remaining eight factors—Young's modulus, wellhead pressure, displacement, porosity, wellbore diameter 2, wellbore diameter 3, wellbore diameter 1, and resistivity—have lower importance scores.

[0079] The data is then dimensionless and standardized using the minimization method. Finally, grey relational analysis is used to analyze the correlation between these 16 factors (e.g., ...). Figure 5 As shown in the diagram, this is a schematic diagram of the correlation analysis. The correlation values ​​for fracture pressure are as follows: sonic transit time, density, brittleness index, Poisson's ratio, depth, minimum principal stress, porosity, natural gamma, maximum principal stress, displacement, Young's modulus, wellbore 2, resistivity, wellbore 1, wellhead pressure, and wellbore 3.

[0080] Combining the importance ranking results of random forest and the correlation ranking results of grey relational analysis, a comprehensive ranking of the two methods was obtained through statistical processing, and the main controlling factors affecting the fracture pressure of tight oil were obtained. The specific ranking results are shown in Table 1.

[0081] As shown in Table 1, the top 10 factors are: acoustic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, maximum principal stress, porosity, and Young's modulus. Figure 5 This is a schematic diagram illustrating the correlation analysis of factors affecting fracture pressure in tight oil horizontal wells according to an embodiment of the present invention. It also integrates the relative degree trend line analysis of two models (such as...). Figure 4 , Figure 5 The six factors of displacement, well diameter 2, wellhead pressure, well diameter 1, resistivity, and well diameter 3 had relatively low scores and rankings, so they were removed.

[0082] Analysis revealed that the remaining 10 geological factors significantly impacted the fracture pressure of tight oil. Stepwise regression analysis was used for secondary screening. Since the regression model contained 10 factors, some might not be significantly correlated with fracture pressure. Therefore, a backward stepwise regression method was employed to optimize the controlling factors, construct a full variable model, and progressively eliminate insignificant factors until all factors were significantly correlated with fracture pressure. The full variable model is as follows:

[0083] Y = a1X1 + a2X2 + ... + a 10 X 10 +X0

[0084] Where Y represents the rupture pressure, X iSee Table 2 for specific meanings, a i Let i represent the coefficients, i = 1, 2, ..., 10.

[0085] Table 2 Definitions of relevant variables

[0086]

[0087] After three backward regressions, porosity and Young's modulus did not meet the criteria and were removed from the model. Through final screening, eight factors that significantly affect the fracture pressure of tight oil were identified: sonic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, and maximum principal stress.

[0088] refer to Figure 6 The training method for the fracture pressure prediction model of the tight oil horizontal well includes:

[0089] S601 uses the XGBoost algorithm, GA-BP neural network, support vector regression algorithm (SVR) and partition fitting algorithm (conventional method) to construct the corresponding initial models respectively;

[0090] S602, after training and testing multiple initial models, obtained the corresponding prediction accuracy, and selected the model with the highest prediction accuracy as the prediction model for fracture pressure in tight oil horizontal wells.

[0091] Specifically, the partitioned fitting algorithm, assuming the rock is impermeable, applies Terzaghi effective stress and proposes the first formula for calculating the rock fracture pressure when vertical fractures occur in the formation under open-hole completion conditions:

[0092] p b =3σ h -σ H +σ f -p o ;

[0093] p b Indicates rock fracture pressure; σ h σ represents the minimum horizontal principal stress of the formation; H σ represents the maximum horizontal principal stress of the formation; f p represents the uniaxial tensile stress strength of the rock formation; o This represents the pore pressure of the formation rocks. If the rocks are non-permeable, the pore pressure p is... o It is 0.

[0094] In practical applications, traditional fracturing pressure calculation formulas can no longer meet the needs. For complex geological structures and other conditions, polynomial fitting formulas are developed based on traditional fracturing calculation formulas to obtain fitting formulas for different blocks.

[0095] Support Vector Regression (SVR) uses a regression-type support vector machine, which is an important model primarily used to handle regression problems. SVR differs from some traditional regression models in the following ways, as shown in Table 3.

[0096] Table 3. Differences between SVR model and traditional regression model

[0097]

[0098] The GA-BP neural network is a backpropagation (BP) neural network algorithm based on genetic algorithms. This algorithm is a random search method derived from the genetic and evolutionary mechanisms of organisms in nature. The algorithm first encodes and assigns values ​​to individuals and calculates their fitness values. Then, based on the fitness function, it selects the best individual. All selected individuals undergo a series of operations such as selection, crossover, and mutation, keeping the individuals with good fitness and discarding those with poor fitness. The new individual inherits the good genes from the previous generation. This process is repeated until the requirements are met or the required number of iterations is reached.

[0099] The difference between GA-BP neural networks and BP neural networks lies in the optimization of the initial weights and thresholds of the GA-BP neural network using genetic operations such as selection, crossover, and mutation, thereby obtaining the optimal weights and thresholds. The most crucial aspect of using GA-BP neural network models to predict coalbed methane production is optimizing the network's weights and thresholds.

[0100] XGBoost, a member of the Boosting algorithm family, fundamentally aims to build a highly accurate strong classifier using multiple simple weak classifiers. XGBoost is primarily an improvement on Gradient Boosting Decision Tree (GBDT). Aside from some differences in engineering implementation and problem-solving, the biggest difference lies in the objective function. Boosting combines individual learners to generate dependencies, effectively building boosting trees and running them in parallel. XGBoost boasts advantages such as fast computation, high efficiency and accuracy, high stability, and strong generalization ability. Its main idea is to learn a new function by adding trees, fitting the residuals of the final prediction, obtaining sample scores, and then summing the scores of each tree to get the final predicted score for the sample. During training, a key issue in building a tree is finding the optimal split point for the leaf nodes. XGBoost supports two methods for splitting nodes: a greedy algorithm and an approximate algorithm.

[0101] To accurately and intuitively evaluate the performance of the three models, the following three evaluation metrics are used: 1. Coefficient of determination R 21. The closer the value is to 1, the better the prediction accuracy of the model; 2. Mean relative error (MAPE), its value ranges from [0, +∞). The closer it is to 0, the better the performance of the model. If the value exceeds 100%, it means that the model is a poor model; 3. Mean squared error (RMSE), the smaller the value of the RMSE index, the better the prediction performance of the model. It can also indicate the degree of dispersion of a dataset.

[0102] The fracturing pressure data comes from three blocks—Block 2, Block 18, and Block 131—in a certain region, totaling 102 clusters (as shown in Table 4). Due to differences in geological conditions and other factors among the blocks, directly using the traditional fracturing pressure calculation formula would produce significant errors. Therefore, based on the classic fracturing calculation formula, 80% of the sample data from each block was selected and multiple linear regression was used to obtain the fitting formula for these three blocks (as shown in Table 5). The remaining 20% ​​of the sample data was used to validate its accuracy. The validation rate and accuracy of Blocks 2, 18, and 131 are shown in Table 6.

[0103] Table 4. Well data for some blocks

[0104]

[0105]

[0106] Table 5 Polynomial Fitting Formulas

[0107] Block Fitting formula Block 2 <![CDATA[P b =-74.68σ h +72.90s H +0.80P0-666.30]]> Block 18 <![CDATA[P b =20.79σ h -19.47s H +0.80P0+201.45]]> Block 131 <![CDATA[P b =0.20σ h +0.23σ H -3.17P0+193.87]]>

[0108] Table 6 Block Prediction Results

[0109] Block 2 Block 18 Block 131 average Fit accuracy 92.71％ 90.16％ 94.98％ 92.62％ Prediction accuracy 90.50％ 82.87％ 87.72％ 87.03％

[0110] As shown in Table 5, the fitting formulas for blocks 2, 18, and 131 are different, reflecting the differences in geological and other factors among each block. Furthermore, the fitting formulas were used to validate the clusters within the three blocks, with block 2 achieving the best validation accuracy of 90.50%. The average validation accuracy for the three blocks was 87.03%.

[0111] By studying the main controlling factors of fracture pressure in tight gas wells, eight parameters—sonic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, and maximum principal stress—were used as input samples. The fracture pressure was used as the output sample. Three models (XGBoost, GA-BP, and SVR) were established for all samples, and their prediction performance was compared and analyzed. In the selected 102 complete sample clusters, 80% of the samples (as shown in Table 7) were used as the training set, and the remaining 20% ​​were used as the test set (as shown in Table 8) to verify the accuracy of the prediction models. The three methods were then applied respectively.

[0112] Table 7 Training Dataset

[0113]

[0114] Table 8 Test Dataset

[0115]

[0116]

[0117] The model was trained and simulated using deep learning software. In the GA-BP neural network, the number of generations was set to 50, the population size to 10, the crossover probability to 0.2, the mutation probability to 0.1, the number of hidden layer nodes to 15, the maximum number of iterations to 1000, and the error target to 0.001. The Logsigmod activation function was selected for both the input layer to the hidden layer and the hidden layer to the output layer.

[0118] The GA-BP neural network model was trained on 81 clusters of training samples and used to predict the rupture pressure of 21 clusters of test samples. The actual values ​​and predicted values ​​were compared as follows: Figure 7 As shown, the GA-BP prediction accuracy is 84.35%. The prediction error RMSE is 18.31 MPa, the MAPE is 19.78%, and the R... 2 The value is 0.84.

[0119] The model was trained and simulated using programming software. The RBF function was selected as the kernel function in the SVR algorithm, with the kernel coefficients set to "auto" by default. After five-fold cross-validation, the optimal penalty coefficient C for the model was found to be 5.

[0120] The SVR was used to train on 81 clusters of training samples, and the fracture pressure of 21 clusters in the test set was predicted. The actual values ​​and predicted values ​​were compared as follows: Figure 8 As shown, the SVR prediction accuracy is 79.89%. The prediction error RMSE is 20.51 MPa, the MAPE is 25.39%, and the R² value is 0.74.

[0121] The model was trained and simulated using programming software. In the XGBoost algorithm, the learning rate was set to 0.1, the maximum tree depth was set to 6, the sample sampling rate was set to 0.5, and the minimum weight sum was 1.

[0122] The XGBoost algorithm was used to train on 81 clusters of training samples, and the fracture pressure of 21 clusters in the test set was predicted. The comparison between the actual and predicted values ​​is shown below. Figure 9As shown, XGBoost's prediction accuracy is 92.05%. The prediction error metric RMSE is 8.79 MPa, the prediction error metric MAPE is 8.65%, and the R² value is 0.88.

[0123] The following discussion and comparative analysis of different methods, different sample sizes, and different blocks are used to verify the universality of the tight oil fracture pressure prediction method based on the XGBoost algorithm.

[0124] To verify the superiority of the tight oil fracture pressure prediction method established in this invention and the rationality of the main controlling factors, the prediction results of different methods were compared using the main controlling factors as input, respectively, through XGBoost algorithm, GA-BP neural network, SVR, and fitting formula. Figure 10 The image shown is a diagram illustrating the comparison of prediction results.

[0125] The comparison results show that the XGBoost algorithm outperforms the GA-BP neural network, SVR, and fitting formula.

[0126] Because different blocks have different geological conditions, different fitting formulas are used for different blocks, which lack universality. In contrast, machine learning algorithms such as XGBoost learn from the data of all blocks, and have strong learning ability, universality, and higher prediction accuracy when facing various complex geological conditions.

[0127] The number of training samples affects the prediction accuracy, and generally, more training samples are better. However, in practical applications, the optimal number of training samples needs to be determined based on the specific model and method. Therefore, this study examines the impact of different training set sample numbers on the accuracy of the XGBoost algorithm in predicting the fracture pressure of tight oil to select the optimal number of samples. The comparison results for different sample numbers are shown in Table 9.

[0128] Table 9. Impact of different training samples on prediction accuracy

[0129] Number of training samples Accuracy / % 66 87.25 75 85.63 81 92.05 87 89.58

[0130] By analyzing Table 9 in conjunction with the research results above, it can be seen that the XGBoost algorithm achieves the highest accuracy in predicting the fracture pressure of tight oil when the number of training samples is selected as 81.

[0131] Block 2 is located on the northwestern edge of a basin, bordered by Block 131 to the north and Block 18 to the southwest. Wells in Block 131 are approximately 3100m deep, those in Block 2 are approximately 3300m deep, and those in Block 18 are approximately 3800m deep. Formation parameters vary significantly with increasing depth. The entire block has a depth of 2812–4230m, with oil saturation of 45%–65%, porosity of 7.7%–11.8%, overburden permeability of 0.02–0.45mD, gravel size of 2–35mm, poorly developed natural fractures, a biaxial stress difference of 10–22MPa, Young's modulus of 19–25GPa, and Poisson's ratio of 0.2–0.25.

[0132] After studying the fitting formulas for blocks 2, 18, and 131 respectively, a prediction method for tight oil fracture pressure was established based on the XGBoost algorithm by combining the data from the three blocks. The prediction results of the four methods are shown in Table 10.

[0133] Table 10 Prediction results for blocks 2, 18, and 131

[0134] Block Block 2 Block 18 Block 131 XGBoost (Combined Three Blocks) Prediction accuracy 90.50％ 82.87％ 87.72％ 92.05％

[0135] Analysis of Table 10 reveals that the XGBoost algorithm performs well in predicting the fracture pressure of the overall data for these three blocks, significantly reducing related workload. Therefore, the tight oil fracture pressure prediction method proposed in this invention is not only applicable to these three blocks simultaneously, but also paves the way for further application of machine learning methods to predict fracture pressure.

[0136] Factors influencing the fracture pressure of tight oil were ranked by importance using random forest and by correlation degree using grey relational analysis. A comprehensive ranking was then performed, and stepwise regression was used to further select the optimal controlling factors. Based on these controlling factors, a machine learning prediction model for the fracture pressure of tight oil was established and compared with a partitioned fitting formula.

[0137] In one embodiment, S103, factors affecting tight oil production are extracted from the tight oil horizontal well data, and these factors are input into a tight oil production prediction model to obtain the tight oil production. The tight oil production prediction model is trained using the following method: multiple initial models are constructed using various prediction algorithms; a second training set and a second test set are constructed using sample data of factors affecting tight oil production; the multiple initial models are trained using the second training set, and tested using the second test set; a model is selected based on the test results to obtain a tight oil horizontal well fracture pressure prediction model.

[0138] refer to Figure 11The specific process for extracting factors affecting tight oil production from the tight oil horizontal well data is as follows:

[0139] S1101, For the tight oil horizontal well data, all factors affecting tight oil production are obtained through random forest importance algorithm and grey relational analysis algorithm, and then ranked by importance and correlation.

[0140] S1102, by stepwise regression analysis to eliminate factors that did not meet the correlation with tight oil production, the main controlling factors affecting tight oil production were obtained.

[0141] Among them, the main controlling factors affecting tight oil production include at least: cluster spacing, total sand content, permeability, number of fractures, average construction discharge rate, total liquid volume, average pump stop pressure, pre-filled liquid ratio, porosity, and oil saturation.

[0142] Taking a specific block as an example, 15 factors affecting tight oil production were summarized based on data from 295 wells in that block. Then, random forest was used to rank the importance of these 15 factors. The results of the random forest importance ranking for all factors are as follows: Figure 12 As shown.

[0143] Figure 12 This is a schematic diagram illustrating the random forest importance ranking results when processing factors affecting tight oil production according to an embodiment of the present invention. (Reference) Figure 12 The factors with the highest importance scores for tight oil production, obtained through random forest analysis, are, in descending order: total sand content, permeability, cluster spacing, total liquid volume, number of fractures, average construction discharge rate, average pump stop pressure, pre-flush ratio, oil saturation, and porosity. The remaining five factors have lower importance scores. The data is then dimensionless and standardized using minimization and maximization methods. Finally, grey relational analysis is used to analyze the correlation between these 15 factors (e.g.,...). Figure 13 As shown, Figure 13 This is a schematic diagram illustrating the correlation analysis of factors affecting tight oil production according to an embodiment of the present invention. The correlation values ​​for tight oil production are, in descending order: cluster spacing, number of fractures, average drilling displacement, total sand volume, average shutdown pressure, permeability, total fluid volume, pre-flush ratio, porosity, oil saturation, oil layer encounter rate, horizontal section length, average sand ratio, oil layer thickness, and wellbore depth.

[0144] Combining the importance ranking results of random forest and the correlation ranking results of grey relational analysis, a comprehensive ranking of the two methods was obtained through statistical processing, and the main controlling factors affecting tight oil production were obtained. The specific ranking results are shown in Table 11.

[0145] Table 11 Ranking of Factors Affecting Tight Oil Production

[0146]

[0147] Referring to Table 11, the following 12 factors rank highest: cluster spacing, total sand content, permeability, number of fractures, average construction discharge rate, total fluid volume, average pump shutdown pressure, pre-fluid ratio, porosity, oil saturation, oil layer encounter rate, and average sand ratio. Simultaneously, a comprehensive analysis of the relative degree trend lines of the two models (such as...) Figure 12 , Figure 13 The three factors of oil layer thickness, horizontal section length, and wellbore depth had relatively low scores and rankings, so they were excluded.

[0148] Analysis revealed that the remaining 12 geological factors significantly impacted tight oil production. A second screening was conducted using stepwise regression analysis. Since the regression model contained 12 factors, some might not be significantly correlated with production. Therefore, a backward stepwise regression method was employed to optimize the controlling factors, construct a full variable model, and progressively eliminate insignificant factors until all factors were significantly correlated with fracture pressure. The full variable model is as follows:

[0149] Z = β1Y1 + β2Y2 + ... + β 11 Y 11 +β 12 Y 12 +Y0

[0150] Where Z represents the rupture pressure, Y i See Table 12 for specific meanings, β i Let i represent the coefficients, i = 1, 2, ..., 12.

[0151] Table 12 Definitions of Relevant Variables

[0152]

[0153] After three backward regressions, the oil layer encounter rate and average sand ratio did not meet the criteria and were removed from the model. Through final screening, 10 factors that significantly affect tight oil production were identified: cluster spacing, total sand content, permeability, number of fractures, average construction discharge rate, total fluid volume, average pump stop pressure, pre-fluid ratio, porosity, and oil saturation.

[0154] refer to Figure 14 The training method for the tight oil production prediction model includes:

[0155] S1401 uses the Support Vector Machine (SVM) classification algorithm, Random Forest classification algorithm, AdaBoos algorithm, and XGBoost algorithm to construct the corresponding initial models respectively;

[0156] S1402, after training and testing multiple initial models, the corresponding prediction accuracy, precision, and recall are obtained. The model with the highest average of the three is selected as the tight oil production prediction model. Wherein, the accuracy is the proportion of samples correctly predicted by the model out of the total samples; the precision is the proportion of samples predicted as positive by the model that are actually positive; and the recall is the proportion of samples that are actually positive that are correctly predicted as positive by the model.

[0157] Production forecasting is a hot research topic in oil and gas field development. Accurate prediction of high-yield and low-yield wells in tight oil fields can provide effective guidance for scientific decision-making, contributing to the efficient and sustainable development of tight oil. This invention employs a binary classification method to form a prediction method for high-yield and low-yield wells in tight oil fields, and simultaneously compares, analyzes, and summarizes the effects of various prediction methods.

[0158] Support Vector Machines (SVMs) are a commonly used machine learning algorithm in data mining. Their good generalization ability avoids overfitting and the local optima problem inherent in traditional machine learning. Furthermore, the algorithm's complexity is calculated based on the number of support vectors, avoiding the curse of dimensionality. SVMs can be used for classification (Support Vector Classification, SVC) and regression (Support Vector Regression, SVR). SVC (Support Vector Machine Classification Algorithm) performs particularly well in solving binary classification problems.

[0159] Random forest (RF) classification algorithm is an ensemble learning algorithm based on binary decision trees. Since the generation of decision trees is independent, random forests are easy to parallelize when processing large datasets. In classification problems with high-dimensional data, it has significant advantages such as high speed, high accuracy, and good stability, and its effectiveness has been verified in numerous applications. Furthermore, the differences in sample subsets reduce the correlation between decision trees, thus ensuring that the random forest algorithm has strong generalization ability. The random forest algorithm has very wide applications in classification, regression, and feature selection. The application of random forests in feature selection is described in the foregoing embodiments of this invention.

[0160] AdaBoost (Adaptive Boosting) is an ensemble learning algorithm that achieves high accuracy by integrating multiple weak classifiers with accuracies greater than random guessing (50%) to make joint decisions. Compared to various classifier training algorithms, AdaBoost also has feature selection capabilities. Furthermore, the classification model trained by AdaBoost has low complexity, good real-time computation, and strict convergence, making it less prone to overfitting. Both experimental and theoretical studies have demonstrated that AdaBoost possesses extremely strong generalization ability.

[0161] XGBoost (Extreme Gradient Boosting) is a boosting ensemble algorithm based on CART regression trees. XGBoost has no special requirements for the data; regardless of the size or distribution of the data samples, it demonstrates advantages such as fast computation speed, input data invariance, and high prediction accuracy. Therefore, it is widely used and highly regarded in machine learning and data mining. Unlike traditional decision tree-based ensemble algorithms, XGBoost incorporates regularization terms such as tree depth and leaf node weights into its cost function. This not only controls the complexity of the prediction model but also prevents overfitting. Furthermore, XGBoost uses a second-order Taylor expansion to approximate the cost function, making the objective function closer to the actual value, thus achieving higher prediction accuracy. The objective function can determine the quality of the tree structure. Theoretically, all possible tree structures can be enumerated to find the optimal spanning tree, but this is difficult to achieve in practice. Therefore, XGBoost optimizes one layer of the tree at a time, searching for the optimal split point using the above method, thereby gradually optimizing the optimal tree structure.

[0162] Based on the field data of tight oil wells in a certain region, in order to facilitate differentiation and subsequent classification and optimization studies, 295 tight oil well samples were divided into two types: high-yield wells and low-yield wells according to their average daily oil production. The specific classification criteria and the number of samples in each category are shown in Table 13.

[0163] Table 13 Production Classification Criteria

[0164] category Sample type Production range (t / d) Label Sample size Average output (t / d) 1 low-yield wells (0,15] 0 131 8.25 2 High-yield wells (15,61] 1 164 24.92

[0165] For 295 tight oil well data samples, 20 samples were randomly selected from 163 samples belonging to the high-yield well category, and 15 samples were randomly selected from 131 samples belonging to the low-yield well category. These 35 data samples constituted the test dataset for classification and production prediction studies. The remaining 260 samples were used as the training dataset for building various models. Partial data of the training and test sets before normalization are shown in Tables 14 and 15. Before the formal calculation, all samples underwent maximum / minimum value normalization.

[0166] Table 14 Training Dataset

[0167]

[0168] Table 15 Test Dataset

[0169]

[0170] The SVC (Support Vector Machine) algorithm uses the RBF kernel function, therefore the kernel parameter sigma and penalty factor C need to be determined. The RF algorithm needs to determine the number of classification trees and leaves. The AdaBoost algorithm needs to determine the number of iterations. The XGBoost algorithm needs to determine the tree depth and gamma parameter, etc. To ensure each algorithm reaches its optimal state, optimization of key parameters is performed, referring to... Figures 15 to 18 The diagrams illustrate parameter optimization for SVC, RF, AdaBoost, and XGBoost, respectively. The optimization results are shown for one or two of the most important parameters in each of the four algorithms. The SVC, RF, and AdaBoost algorithms are implemented using built-in functions in the deep learning software; the XGBoost algorithm is implemented using the third-party library xgboost, with "subsample" set to 0.7, "eta" set to 0.1, and the number of iterations set to 500. Other non-optimizable parameters are left as default. Based on the optimization results, the optimal parameter values ​​are shown in Table 16.

[0171] Table 16 Optimal values ​​for parameters of each classification algorithm

[0172] Serial Number algorithm software method Optimal value of parameter 1 SVC Deep learning software svmtrain() Sigma is set to 20, and C is set to 5. 2 RF Deep learning software TreeBagger() The number of leaves is 100, and the number of trees is 28. 3 AdaBoost Deep learning software fitensemble() The number of iterations is 670. 4 XGBoost Programming software xgboost.train() Gamma is set to 1, and the tree depth is set to 6.

[0173] For the binary classification problem of high-yield and low-yield wells in tight oil, let TP be the number of correctly classified high-yield wells in the test set, FP be the number of incorrectly classified high-yield wells, FN be the number of incorrectly classified low-yield wells, and TN be the number of correctly classified low-yield wells. These form a second-order confusion matrix arranged row by row. The effectiveness of the high-yield and low-yield well prediction method for tight oil is evaluated using accuracy, precision, recall, and F1 score. The calculation formulas are as follows:

[0174]

[0175] For 260 tight oil well samples in the training set, prediction methods for high-yield and low-yield tight oil wells based on SVC, RF, AdaBoost, and XGBoost algorithms were established according to the optimal parameter values. Predictions were then performed on 35 samples in the test set, and the results are as follows: Figures 19 to 22The diagrams shown illustrate the confusion matrices for SVC, RF, AdaBoost, and XGBoost, respectively. Classification task 1 and class 0 represent the two classes in the classification task, while model prediction result 1 and output result 0 represent the predicted class. By comparing the number of misclassifications (the numbers indicating discrepancies between predicted and actual classes) in the four confusion matrices, it is clear that XGBoost performs best in this classification task, with the fewest misclassified samples; RF is second best, while SVC and AdaBoost have relatively more misclassifications. The confusion matrices can be used to determine which algorithm is more suitable for tight oil production classification, providing a quantitative basis for model selection; Table 17 shows the evaluation results of the prediction methods.

[0176] Table 17 Evaluation Results of Prediction Methods

[0177]

[0178]

[0179] In Table 17, precision represents the proportion of all correctly predicted wells out of the total number of samples in the test set; accuracy represents the proportion of correctly predicted high-yield wells out of all wells predicted as high-yield wells; recall represents the proportion of correctly predicted high-yield wells out of the total number of high-yield wells in the test set; and the F1 score is the harmonic mean of precision and recall. All of the above evaluation metrics have passed... Figures 19 to 22 The confusion matrix is ​​calculated, and the larger the value, the better the prediction method.

[0180] As shown in Table 17, among the four methods for predicting high-yield and low-yield tight oil wells, the accuracy rates of the RF algorithm and the XGBoost algorithm are both the highest at 88.57%. The RF algorithm has the highest precision rate of 94.44%, but the XGBoost method also achieves a precision rate of 90.00%, with the highest recall rate and F1 score of 0.90. Overall, considering all evaluation indicators, the prediction method based on the SVC algorithm performs the worst. The prediction method based on the AdaBoost algorithm has too few adjustable parameters, making it difficult to effectively improve accuracy through parameter tuning; therefore, its prediction performance is not outstanding. Both the RF algorithm-based and XGBoost algorithm-based methods have their advantages. Comparatively, the XGBoost algorithm-based method performs better overall, and its more parameter settings and more comprehensive theoretical support allow for further improvement in prediction performance through more intensive parameter tuning.

[0181] For predicting high-yield and low-yield wells in tight oil, this invention describes the SVC, RF, AdaBoost, and XGBoost classification algorithms respectively. Combining the results of the main control factors, a prediction method for high-yield and low-yield wells in tight oil is formed, and implemented using deep learning software and programming software. The effectiveness of each prediction method is analyzed and evaluated in detail, considering the evaluation indicators of multiple classification problems and the advantages and disadvantages of the algorithms themselves. The XGBoost-based prediction method for high-yield and low-yield wells in tight oil can meet current accuracy requirements and has broader application prospects.

[0182] The present invention will now be described with reference to a specific embodiment. (See reference...) Figure 23 This is a schematic diagram of the technical route of a specific embodiment of the present invention. Figure 23 As shown, through analysis of the main controlling factors of tight oil fracture pressure and production classification, prediction of tight oil fracture pressure, and investigation and summary of production classification methods, the main research content was determined. The specific process is as follows:

[0183] S2301, collect field data on tight oil.

[0184] S2302, Random Forest Method, Gray-Level Relational Analysis.

[0185] Based on the differences in field data of tight oil and extensive surveys, a preliminary selection of factors was made. On this basis, the factors affecting the fracture pressure and production classification of tight oil were comprehensively studied by combining the importance ranking analysis of random forest and grey relational analysis.

[0186] S2303, gradual regression, the main controlling factor.

[0187] Stepwise regression analysis was used to select the main controlling factors affecting the fracture pressure and production classification of tight oil.

[0188] Specifically, by studying and analyzing geological and fracturing data from tight oil wells in the block, and reviewing relevant literature, the factors influencing the fracture pressure and production of tight oil wells were initially selected. Based on this, random forest and grey relational analysis were used to conduct importance and correlation analyses, respectively. The results were then comprehensively ranked, and stepwise regression analysis was used to optimize the factors. The final selected controlling factors are more reasonable and lay the foundation for predicting the fracture pressure and classifying the production of tight oil wells.

[0189] S2304, Rupture Pressure Prediction Model.

[0190] Based on field data of tight oil in the target area, this study investigates fracture pressure prediction at the cluster level to address the question of whether fracture can occur. Building upon traditional fracture pressure formulas, a polynomial formula suitable for different blocks is fitted. Furthermore, based on the main controlling factors of tight oil fracture pressure, various classic machine learning algorithms are incorporated to establish a machine learning-based tight oil fracture pressure prediction model. The prediction performance is compared and analyzed to select the optimal model.

[0191] Specifically, based on field data of tight oil wells in the target area, fracturing prediction was studied on a cluster basis to address the question of whether fracturing is feasible. A machine learning model for predicting fracture pressure in tight oil wells was established using eight selected key control factors. The XGBoost algorithm achieved the highest prediction accuracy of 92.05%, while the GA-BP neural network and SVR algorithm achieved 84.35% and 79.89%, respectively. The prediction accuracy of these three methods is higher than the average accuracy of 87.03% for different blocks using a polynomial fitting formula based on traditional fracture pressure methods. The advantage of machine learning algorithms such as XGBoost over polynomial fitting formulas lies not only in their higher prediction accuracy but also in their strong learning ability and applicability to variations arising from complex geological environments.

[0192] S2305, tight oil production prediction model.

[0193] This study classifies tight oil production based on field data from the target region, using wells as the unit. Building upon existing well development, it analyzes the development benefits of tight oil wells. Using the main controlling factors of tight oil production and combining various classic classification algorithms with ensemble learning classification algorithms, a method for predicting high-yield and low-yield tight oil wells was developed. The application results were compared and analyzed, and the optimal model was selected.

[0194] Specifically, based on field data of tight oil in the target area, a well-by-well classification study was conducted to analyze the development benefits of tight oil wells after they had been opened. A method for predicting high-yield and low-yield tight oil wells was established based on 10 selected empirical factors for tight oil production classification. The highest accuracy rate of each method in the test set of 35 wells was 94.44%.

[0195] S2306, Tight Oil Well Development Strategy.

[0196] Based on the prediction results, reasonable development strategies are provided for tight oil wells in the target block. For example, wells with more suitable fracture pressure and production rates are prioritized for development.

[0197] It should be noted that although the operation of the method of the present invention has been described in a specific order in the above embodiments and figures, this does not require or imply that the 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.

[0198] After introducing the method of exemplary embodiments of the present invention, the following references are made. Figure 24 An exemplary embodiment of the present invention will be described, which is a predictive device for classifying fracture pressure and production in tight oil reservoirs.

[0199] The implementation of the predictive device for fracture pressure and production classification in tight oil reservoirs can refer to the implementation of the above-described method, and will not be repeated here. The term "module" or "unit" used below can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0200] Based on the same inventive concept, this invention also proposes a predictive device for classifying fracture pressure and production in tight oil reservoirs, such as... Figure 24 As shown, the device includes:

[0201] Data acquisition module 2410 is used to acquire data from tight oil horizontal wells;

[0202] The tight oil horizontal well fracture pressure prediction module 2420 is used to extract factors affecting the fracture pressure of the tight oil horizontal well from the tight oil horizontal well data, input the factors affecting the fracture pressure of the tight oil horizontal well into the tight oil horizontal well fracture pressure prediction model, and obtain the fracture pressure of the tight oil horizontal well; wherein, the tight oil horizontal well fracture pressure prediction model is trained by the following method: multiple initial models are constructed by various machine learning algorithms, a first training set and a first test set are constructed using sample data of factors affecting the fracture pressure of the tight oil horizontal well, multiple initial models are trained using the first training set, and tested using the first test set, and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model;

[0203] The tight oil production prediction module 2430 is used to extract factors affecting tight oil production from the tight oil horizontal well data, input the factors affecting tight oil production into the tight oil production prediction model, and obtain the tight oil production; wherein, the tight oil production prediction model is trained by the following method: multiple initial models are constructed by various prediction algorithms, a second training set and a second test set are constructed using sample data of factors affecting tight oil production, multiple initial models are trained using the second training set, and tested using the second test set, and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model;

[0204] The development strategy generation module 2440 is used to classify tight oil horizontal wells according to the tight oil production, obtain tight oil production classification results, and generate a tight oil horizontal well development strategy based on the tight oil horizontal well fracture pressure and tight oil production classification results.

[0205] In one embodiment, after acquiring data from a tight oil horizontal well, the data acquisition module 2410 is further configured to:

[0206] The tight oil horizontal well data is preprocessed, missing data is filled in, outliers are removed and filled in as missing data;

[0207] Based on the pre-processed tight oil horizontal well data, the factors affecting the fracture pressure and production of tight oil horizontal wells were initially selected. The initial selection method was as follows: the factors were initially selected through manual survey, analysis and summarization; the logging, fracturing and drainage data of different blocks of tight oil horizontal wells were analyzed and sorted to eliminate factors that were not involved in the blocks and factors with abnormal data.

[0208] In one embodiment, the tight oil horizontal well fracture pressure prediction module 2420 extracts factors affecting the fracture pressure of the tight oil horizontal well from the tight oil horizontal well data, including:

[0209] For the tight oil horizontal well data, all factors affecting the fracture pressure of tight oil horizontal wells are obtained through random forest importance algorithm and grey relational analysis algorithm, and their influence is comprehensively ranked.

[0210] By eliminating factors that did not meet the correlation with fracture pressure in tight oil horizontal wells through stepwise regression analysis, the main controlling factors affecting fracture pressure in tight oil horizontal wells were obtained.

[0211] In one embodiment, the main controlling factors affecting the fracture pressure of tight oil horizontal wells include at least: sonic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, and maximum principal stress.

[0212] In one embodiment, the training method for the fracture pressure prediction model of the tight oil horizontal well includes:

[0213] Initial models were constructed using the XGBoost algorithm, GA-BP neural network, support vector regression algorithm, and partition fitting algorithm, respectively.

[0214] After training and testing multiple initial models, the corresponding prediction accuracies were obtained. The model with the highest prediction accuracy was selected as the prediction model for fracture pressure in tight oil horizontal wells.

[0215] In one embodiment, the tight oil production prediction module 2430 extracts factors affecting tight oil production from the tight oil horizontal well data, including:

[0216] For the tight oil horizontal well data, all factors affecting tight oil production are obtained through random forest importance algorithm and grey relational analysis algorithm, and then ranked comprehensively by importance and correlation degree.

[0217] By eliminating factors that did not meet the correlation with tight oil production through stepwise regression analysis, the main controlling factors affecting tight oil production were obtained.

[0218] In one embodiment, the main controlling factors affecting tight oil production include at least: cluster spacing, total sand content, permeability, number of fractures, average construction discharge rate, total liquid volume, average pump stop pressure, pre-fluid ratio, porosity, and oil saturation.

[0219] In one embodiment, the training method for the tight oil production prediction model includes:

[0220] Initial models were constructed using the Support Vector Machine (SVM) classification algorithm, Random Forest (Random Forest) algorithm, AdaBoos algorithm, and XGBoost algorithm, respectively.

[0221] After training and testing multiple initial models, the corresponding prediction accuracy, precision, and recall were obtained. The model with the highest average of the three was selected as the tight oil production prediction model. The accuracy is the proportion of samples correctly predicted by the model out of the total samples; the precision is the proportion of samples predicted as positive by the model that are actually positive; and the recall is the proportion of samples that are actually positive that were correctly predicted as positive by the model.

[0222] It should be noted that although several modules of the predictive device for classifying fracture pressure and production in tight oil reservoirs have been mentioned in the detailed description above, this classification is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above can be embodied in a single module. Conversely, the features and functions of a single module described above can be further divided and embodied by multiple modules.

[0223] Based on the aforementioned inventive concept, such as Figure 25 As shown, the present invention also proposes a computer device 2500, including a memory 2510, a processor 2520, and a computer program 2530 stored in the memory 2510 and executable on the processor 2520. When the processor 2520 executes the computer program 2530, it implements the aforementioned method for predicting the classification of fracture pressure and production in tight oil reservoirs.

[0224] Based on the aforementioned inventive concept, this invention proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for predicting the classification of fracture pressure and production in tight oil reservoirs.

[0225] Based on the aforementioned inventive concept, this invention proposes a computer program product, which includes a computer program that, when executed by a processor, implements a method for predicting the fracture pressure and production classification of tight oil reservoirs.

[0226] The proposed method and apparatus for predicting fracture pressure and production classification in tight oil reservoirs, through screening the main controlling factors affecting fracture pressure and production classification, constructs a fracture pressure prediction model based on multiple machine learning algorithms, selects the model most suitable for the complex geological conditions of different blocks, and further employs multiple prediction algorithms to construct a production prediction model to achieve production classification prediction. This can effectively distinguish between high-yield wells and low-yield wells, providing a reasonable development strategy for horizontal wells in tight oil, and providing precise technical support for fracturing design and benefit evaluation in tight oil development.

[0227] The acquisition, storage, use, and processing of data in this application all comply with relevant laws and regulations.

[0228] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0229] This invention is described with reference to flowchart illustrations and / or block diagrams of methods and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0230] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0231] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0232] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention 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 within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these 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 the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting the fracture pressure and production classification of a tight reservoir, characterized in that, The method includes: Collect data from tight oil horizontal wells; Factors influencing the fracture pressure of tight oil horizontal wells are extracted from the tight oil horizontal well data. These factors are then input into a tight oil horizontal well fracture pressure prediction model to obtain the fracture pressure. The tight oil horizontal well fracture pressure prediction model is trained using the following method: multiple initial models are constructed using various machine learning algorithms; a first training set and a first test set are constructed using sample data of factors influencing the fracture pressure of tight oil horizontal wells; the multiple initial models are trained using the first training set and tested using the first test set; and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model. Factors affecting tight oil production are extracted from the tight oil horizontal well data, and these factors are input into a tight oil production prediction model to obtain the tight oil production. The tight oil production prediction model is trained using the following method: multiple initial models are constructed using various prediction algorithms; a second training set and a second test set are constructed using sample data of factors affecting tight oil production; the second training set is used to train the multiple initial models, and the second test set is used to test them; the model is selected based on the test results to obtain a tight oil horizontal well fracture pressure prediction model. Tight oil horizontal wells are classified according to their tight oil production to obtain tight oil production classification results. Tight oil horizontal well development strategies are generated based on the fracture pressure and tight oil production classification results. Among them, the factors affecting the fracture pressure of tight oil horizontal wells extracted from the tight oil horizontal well data include: For the tight oil horizontal well data, all factors affecting the fracture pressure of tight oil horizontal wells are obtained through random forest importance algorithm and grey relational analysis algorithm, and their influence is comprehensively ranked. By eliminating factors that did not meet the correlation with fracture pressure in tight oil horizontal wells through stepwise regression analysis, the main controlling factors affecting fracture pressure in tight oil horizontal wells were obtained. Among them, the factors affecting tight oil production extracted from the tight oil horizontal well data include: For the tight oil horizontal well data, all factors affecting tight oil production are obtained through random forest importance algorithm and grey relational analysis algorithm, and then ranked comprehensively by importance and correlation degree. By eliminating factors that did not meet the correlation with tight oil production through stepwise regression analysis, the main controlling factors affecting tight oil production were obtained.

2. The method for predicting the fracture pressure and production classification of tight oil reservoirs according to claim 1, characterized in that, After acquiring data from tight oil horizontal wells, the method also includes: The tight oil horizontal well data is preprocessed, missing data is filled in, outliers are removed and filled in as missing data; Based on the pre-processed tight oil horizontal well data, the factors affecting the fracture pressure and production of tight oil horizontal wells were initially selected. The initial selection method was as follows: the factors were initially selected through manual survey, analysis and summarization; the logging, fracturing and drainage data of different blocks of tight oil horizontal wells were analyzed and sorted to eliminate factors that were not involved in the blocks and factors with abnormal data.

3. The method for predicting the fracture pressure and production classification of tight oil reservoirs according to claim 1, characterized in that, The main controlling factors affecting the fracture pressure of tight oil horizontal wells include at least: sonic transit time, density, Poisson's ratio, brittleness index, depth, minimum principal stress, natural gamma, and maximum principal stress.

4. The method for predicting the fracture pressure and production classification of tight oil reservoirs according to claim 1, characterized in that, The training method for the fracture pressure prediction model of tight oil horizontal wells includes: Initial models were constructed using the XGBoost algorithm, GA-BP neural network, support vector regression algorithm, and partition fitting algorithm, respectively. After training and testing multiple initial models, the corresponding prediction accuracies were obtained. The model with the highest prediction accuracy was selected as the prediction model for fracture pressure in tight oil horizontal wells.

5. The method for predicting the fracture pressure and production classification of tight oil reservoirs according to claim 1, characterized in that, The main controlling factors affecting tight oil production include at least: cluster spacing, total sand content, permeability, number of fractures, average construction discharge rate, total liquid volume, average pump stop pressure, pre-fluid ratio, porosity, and oil saturation.

6. The method for predicting the fracture pressure and production classification of tight oil reservoirs according to claim 1, characterized in that, The training method for the tight oil production prediction model includes: Initial models were constructed using the Support Vector Machine (SVM) classification algorithm, Random Forest (Random Forest) algorithm, AdaBoos algorithm, and XGBoost algorithm, respectively. After training and testing multiple initial models, the corresponding prediction accuracy, precision, and recall were obtained. The model with the highest average of the three was selected as the tight oil production prediction model. The accuracy is the proportion of samples correctly predicted by the model out of the total samples; the precision is the proportion of samples predicted as positive by the model that are actually positive; and the recall is the proportion of samples that are actually positive that were correctly predicted as positive by the model.

7. A predictive device for classifying fracture pressure and production in tight oil reservoirs, characterized in that, The device includes: The data acquisition module is used to acquire data from tight oil horizontal wells; A tight oil horizontal well fracture pressure prediction module is used to extract factors affecting the fracture pressure of the tight oil horizontal well from the tight oil horizontal well data, input the factors affecting the fracture pressure of the tight oil horizontal well into the tight oil horizontal well fracture pressure prediction model, and obtain the fracture pressure of the tight oil horizontal well; wherein, the tight oil horizontal well fracture pressure prediction model is trained by the following method: multiple initial models are constructed using various machine learning algorithms, a first training set and a first test set are constructed using sample data of factors affecting the fracture pressure of the tight oil horizontal well, multiple initial models are trained using the first training set, and tested using the first test set, and a model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model; The tight oil production prediction module is used to extract factors affecting tight oil production from the tight oil horizontal well data, input the factors affecting tight oil production into the tight oil production prediction model, and obtain the tight oil production. The tight oil production prediction model is trained using the following method: multiple initial models are constructed using various prediction algorithms; a second training set and a second test set are constructed using sample data of factors affecting tight oil production; the multiple initial models are trained using the second training set, and tested using the second test set; the model is selected based on the test results to obtain the tight oil horizontal well fracture pressure prediction model. The development strategy generation module is used to classify tight oil horizontal wells according to the tight oil production, obtain tight oil production classification results, and generate a tight oil horizontal well development strategy based on the fracture pressure and tight oil production classification results. The tight oil horizontal well fracture pressure prediction module is also used for: For the tight oil horizontal well data, all factors affecting the fracture pressure of tight oil horizontal wells are obtained through random forest importance algorithm and grey relational analysis algorithm, and their influence is comprehensively ranked. By eliminating factors that did not meet the correlation with fracture pressure in tight oil horizontal wells through stepwise regression analysis, the main controlling factors affecting fracture pressure in tight oil horizontal wells were obtained. The tight oil production prediction module is also used for: For the tight oil horizontal well data, all factors affecting tight oil production are obtained through random forest importance algorithm and grey relational analysis algorithm, and then ranked comprehensively by importance and correlation degree. By eliminating factors that did not meet the correlation with tight oil production through stepwise regression analysis, the main controlling factors affecting tight oil production were obtained.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.

Images