Core permeability retention rate prediction method, device and equipment based on gradient algorithm

Abstract

The embodiment of the invention provides a core permeability retention rate prediction method, device and equipment based on a gradient algorithm. The core permeability retention rate prediction method comprises the following steps: obtaining core permeability retention rate data of a target block; preprocessing the core permeability retention rate data to obtain a feature set of the target block; based on the feature set, constructing an initial batch gradient descent model; parameters of the initial batch gradient descent model are randomly initialized, and a loss function is defined; a preset gradient descent algorithm is adopted, parameters of the initial batch gradient descent model are updated, a loss function is minimized, the initial batch gradient descent model is trained, and a target batch gradient descent model is obtained; substituting weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; and based on the permeability retention rate prediction equation, predicting the core permeability retention rate of the target block to obtain a prediction result.

Description

Technical Field

[0001] The invention relates to the technical field of core permeability retention rate prediction, and relates to but is not limited to a core permeability retention rate prediction method, device and equipment based on a gradient algorithm. Background Art

[0002] During the exploration and development of unconventional oil and gas reservoirs, core samples face a higher risk of fluid intrusion due to the heterogeneity of reservoir physical properties, which not only increases the difficulty of effective development of oil and gas resources, but also poses a challenge to the long-term stability of the reservoir. At present, the evaluation of core damage mainly follows industry standards, but these standards have obvious limitations when dealing with unconventional reservoir core samples with low porosity and low permeability. Core permeability, as a key parameter to characterize the permeability of formation rocks, has a profound impact on the development strategy and production efficiency of oil and gas reservoirs. Given the core position of core permeability in oil and gas reservoir development, how to use existing data resources and advanced algorithm models to accurately predict core permeability retention has become a hot issue in the field of oil and gas exploration.

[0003] In the field of oil and gas exploration, core analysis is a key step in evaluating the physical properties of reservoir rocks. Permeability, as an indicator of the ability of rock fluids to pass through, is crucial for predicting the production performance of oil and gas reservoirs. Understanding the damage mechanism of cores under the action of fluids is of decisive significance for predicting the production performance of oil and gas reservoirs. Therefore, developing effective core damage evaluation methods and permeability prediction models has important practical application value for optimizing the development strategy of oil and gas reservoirs, improving production efficiency, and extending the service life of oil and gas reservoirs. Although traditional core analysis methods, such as laboratory measurements, can provide permeability retention information to a certain extent, they have the following limitations: Time consumption: Laboratory measurements usually require a long time to prepare samples, conduct experiments, and analyze results, which may not be feasible in the process of rapid exploration and development; Cost issues: Laboratory testing is expensive, especially when a large number of samples need to be analyzed; Data incompleteness: Due to the complexity of geological conditions, traditional measurement methods may not be able to fully capture all the physical properties of the core; Human errors: Human errors may be introduced during laboratory testing and data analysis, affecting the reliability of the final results.

[0004] Given the limitations of current core damage assessment methods, there is an urgent need to develop a more efficient, cost-effective, and accurate method to predict core permeability retention. Summary of the invention

[0005] Based on the problems in the related art, the embodiments of the present invention provide a method, device and equipment for predicting core permeability retention rate based on a gradient algorithm.

[0006] The technical solution of the embodiment of the present invention is achieved as follows:

[0007] In a first aspect, an embodiment of the present invention provides a method for predicting core permeability retention rate based on a gradient algorithm, the method comprising:

[0008] Obtain core permeability retention rate data for target blocks;

[0009] Preprocessing the core permeability retention rate data to obtain a feature set of the target block;

[0010] Based on the feature set, construct an initial batch gradient descent model; randomly initialize the parameters of the initial batch gradient descent model, and define a loss function;

[0011] Using a preset gradient descent algorithm, updating the parameters of the initial batch gradient descent model and minimizing the loss function, training the initial batch gradient descent model, and obtaining a target batch gradient descent model;

[0012] Substituting the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation;

[0013] Based on the permeability retention rate prediction equation, the core permeability retention rate of the target block is predicted to obtain a prediction result.

[0014] In a second aspect, an embodiment of the present invention provides a core permeability retention rate prediction device based on a gradient algorithm, the device comprising:

[0015] An acquisition module is used to obtain the core permeability retention rate data of the target block;

[0016] A preprocessing module, used for preprocessing the core permeability retention rate data to obtain a feature set of the target block;

[0017] A construction module is used to construct an initial batch gradient descent model based on the feature set; randomly initialize the parameters of the initial batch gradient descent model, and define a loss function;

[0018] A training module, used to adopt a preset gradient descent algorithm to update the parameters of the initial batch gradient descent model and minimize the loss function, train the initial batch gradient descent model, and obtain a target batch gradient descent model;

[0019] A substitution module, used to substitute the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation;

[0020] The prediction module is used to predict the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result.

[0021] In some embodiments, the acquisition module is also used to collect production water samples from the target block; extract characteristic parameters with a high correlation with the core permeability retention rate from the production water samples; and perform core displacement experiments based on the characteristic parameters to obtain the core permeability retention rate data.

[0022] In some embodiments, the characteristic parameters include iron content, sulfur ion concentration, calcium ion concentration, bicarbonate ion concentration, suspended matter concentration, median particle size of suspended particles, oil content and mineralization.

[0023] In some embodiments, the preprocessing module is further used to sequentially clean, normalize, and process outliers on the core permeability retention rate data to obtain the feature set.

[0024] In some embodiments, the training module is also used to calculate the gradient of the loss function for each of the parameters under the current model parameters, and adjust the parameters based on the direction of the gradient and the learning rate until the number of iterations of the loss function reaches a preset iteration threshold; use a validation set to verify the initial batch gradient descent model to obtain a verified batch gradient descent model; use a test set to test the verified batch gradient descent model to obtain the target batch gradient descent model.

[0025] In a third aspect, an embodiment of the present invention provides a core permeability retention rate prediction device based on a gradient algorithm, comprising: a memory for storing executable instructions; and a processor for implementing the above-mentioned core permeability retention rate prediction method based on a gradient algorithm when executing the executable instructions stored in the memory.

[0026] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing executable instructions for causing a processor to execute the executable instructions to implement the above-mentioned core permeability retention rate prediction method based on the gradient algorithm.

[0027] The core permeability retention rate prediction method based on the gradient algorithm provided in the embodiment of the present invention first obtains the core permeability retention rate data of the target block; pre-processes the core permeability retention rate data to obtain the feature set of the target block; constructs an initial batch gradient descent model based on the feature set; randomly initializes the parameters of the initial batch gradient descent model, and defines a loss function; uses a preset gradient descent algorithm to update the parameters of the initial batch gradient descent model and minimize the loss function, trains the initial batch gradient descent model, and obtains a target batch gradient descent model; substitutes the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; predicts the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result. In this way, by applying machine learning to the prediction and calculation of core damage rate, the dependence on physical experiments can be reduced, and prediction can be made through historical data and acquired data, which can save prediction costs and time to a certain extent. On the other hand, based on the selection of optimal feature data, the effectiveness of the prediction is ensured. At the same time, the impact of multiple factors on core damage during oil field production is also taken into account, which can effectively process multi-dimensional data and improve the comprehensiveness and reliability of core permeability retention rate prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A schematic flow chart of a method for predicting core permeability retention rate based on a gradient algorithm provided in an embodiment of the present invention;

[0029] Figure 2 A schematic diagram of a process for predicting core permeability retention rate based on a gradient descent algorithm provided in an embodiment of the present invention;

[0030] Figure 3 The actual value and predicted value of the test set and R when selecting the learning rate for the gradient descent algorithm provided in the embodiment of the present invention 2 Comparison chart;

[0031] Figure 4 A visualization diagram of the loss function of the gradient descent algorithm model provided in an embodiment of the present invention;

[0032] Figure 5 A schematic diagram of the composition structure of a core permeability retention rate prediction device based on a gradient algorithm provided in an embodiment of the present invention;

[0033] Figure 6 A schematic diagram of the composition structure of a core permeability retention rate prediction device based on a gradient algorithm provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0035] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. Unless otherwise defined, all technical and scientific terms used in the embodiments of the present invention have the same meaning as those commonly understood by those skilled in the art to which the embodiments of the present invention pertain. The terms used in the embodiments of the present invention are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0036] Given the limitations of current core damage assessment methods, there is an urgent need to develop a more efficient, cost-effective and accurate method to predict core permeability retention. This demand has driven the widespread application of machine learning technology in the field of geological science, especially in seismic data processing, reservoir description and oil and gas prediction. By analyzing a large amount of historical data, machine learning algorithms can extract implicit rules in the data and predict unknown parameters. The application of this technology provides geologists with a new tool to improve the efficiency and accuracy of oil and gas exploration and development.

[0037] Batch gradient descent is a widely used optimization algorithm that plays a core role in machine learning model training. The basic idea is to gradually approach the optimal solution by calculating the gradient of the loss function with respect to the model parameters and updating the parameters in the opposite direction of the gradient. This method is particularly important when training complex models because it can efficiently adjust the model parameters to achieve better prediction performance.

[0038] The following describes an exemplary application of a core permeability retention rate prediction device based on a gradient algorithm according to an embodiment of the present invention. The core permeability retention rate prediction device based on a gradient algorithm provided by an embodiment of the present invention can be implemented as a terminal or a server. In one implementation, the core permeability retention rate prediction device based on a gradient algorithm provided by an embodiment of the present invention can be implemented as various types of terminals such as a laptop computer, a tablet computer, a desktop computer, a mobile device, etc.; in another implementation, the core permeability retention rate prediction device based on a gradient algorithm provided by an embodiment of the present invention can also be implemented as a server, wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks (CDN, Content Delivery Network), and big data and artificial intelligence platforms. The terminal and the server can be directly or indirectly connected by wired or wireless communication, which is not limited in the embodiment of the present invention. Below, an exemplary application of the core permeability retention rate prediction device based on a gradient algorithm when it is implemented as a server will be described.

[0039] The embodiment of the present invention provides a method for predicting the core permeability retention rate based on a gradient algorithm, see Figure 1 , Figure 1 is a flow chart of a method for predicting core permeability retention rate based on a gradient algorithm provided by an embodiment of the present invention, which is combined with Figure 1 The steps shown are explained.

[0040] Step S110, obtaining core permeability retention rate data of the target block.

[0041] In some embodiments, permeability refers to the ability of rock to allow fluid to pass under a certain pressure difference, which is a parameter that characterizes the ability of soil or rock itself to conduct liquid. The core permeability retention rate refers to the ratio of the permeability of the core to the initial permeability after a certain pressure is applied.

[0042] In some embodiments, the core permeability retention rate data can be obtained through laboratory testing, geological exploration, historical data records, etc. Specifically, the core samples of the target block can be drilled and their permeability retention rate can be measured under laboratory conditions to obtain accurate experimental data. In addition, the permeability retention rate of the target block can be comprehensively analyzed in combination with geological exploration data and historical data to ensure the comprehensiveness and accuracy of the data.

[0043] Step S120, preprocessing the core permeability retention rate data to obtain a feature set of the target block.

[0044] In some embodiments, after obtaining the core permeability retention rate data, the data may be further preprocessed to extract a feature set of the target block.

[0045] In some embodiments, the preprocessing process includes but is not limited to steps such as data cleaning, data conversion, and feature selection. The quality and availability of data can be improved by removing duplicate data, processing missing values, and standardizing or normalizing data. At the same time, according to the geological characteristics of the target block and the characteristics of the permeability retention rate, features with strong correlation and large amount of information are selected to construct a feature set, which provides a basis for subsequent model construction.

[0046] Step S130, constructing an initial batch gradient descent model based on the feature set; randomly initializing the parameters of the initial batch gradient descent model, and defining a loss function.

[0047] In the present invention, an initial batch gradient descent model is constructed based on the preprocessed feature set. Batch gradient descent is a method for processing large-scale data sets. By constructing the initial model, the basic framework and parameters of the model can be set to provide a basis for subsequent training and optimization. When constructing the model, it is necessary to consider the geological characteristics of the target block and the characteristics of the permeability retention rate, and select appropriate model structures and parameters to ensure the accuracy and stability of the model.

[0048] Step S140, using a preset gradient descent algorithm, updating the parameters of the initial batch gradient descent model and minimizing the loss function, training the initial batch gradient descent model, and obtaining a target batch gradient descent model.

[0049] In the present invention, after the initial model is constructed, the parameters of the model are randomly initialized. Random initialization can prevent the model from falling into a local optimal solution and improve the generalization ability of the model. At the same time, in order to evaluate the performance of the model, the present invention defines a loss function. The loss function is used to measure the difference between the model prediction results and the actual results, and is an important indicator for model training and optimization. According to the permeability retention rate characteristics of the target block, a suitable loss function is selected, such as mean square error, cross entropy, etc., to ensure the accuracy and effectiveness of model training.

[0050] Step S150, substituting the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation.

[0051] In the present invention, after the loss function is defined, the initial model is trained using a preset gradient descent algorithm. The gradient descent algorithm updates the parameters of the model through continuous iterations, so that the loss function is gradually reduced, thereby optimizing the performance of the model. During the training process, it is necessary to select appropriate parameters such as the number of iterations and the learning rate to ensure the stability and efficiency of the model training. Through continuous iteration and optimization, the target batch gradient descent model is finally obtained, which has high prediction accuracy and generalization ability. After obtaining the target model, the present invention substitutes the weight parameters in the model into the initial prediction equation to obtain the permeability retention rate prediction equation. The prediction equation is the core part of the model, which is used to predict the permeability retention rate of the target block based on the input feature values. By substituting the weight parameters, the prediction ability of the model can be converted into a specific mathematical expression, providing a basis for subsequent predictions.

[0052] Step S160: predicting the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result.

[0053] The present invention uses the permeability retention rate prediction equation to predict the core permeability retention rate of the target block. By inputting the characteristic value of the target block, the corresponding permeability retention rate prediction result can be obtained. The prediction result can be used to guide the decision-making process of geological exploration, oil and gas production, etc., to improve resource utilization efficiency and economic benefits. At the same time, the prediction results can also be verified and evaluated to further optimize the model and improve the prediction accuracy.

[0054] The core permeability retention rate prediction method based on the gradient algorithm provided in the embodiment of the present invention first obtains the core permeability retention rate data of the target block; pre-processes the core permeability retention rate data to obtain the feature set of the target block; constructs an initial batch gradient descent model based on the feature set; randomly initializes the parameters of the initial batch gradient descent model, and defines a loss function; uses a preset gradient descent algorithm to update the parameters of the initial batch gradient descent model and minimize the loss function, trains the initial batch gradient descent model, and obtains a target batch gradient descent model; substitutes the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; predicts the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result. In this way, by applying machine learning to the prediction and calculation of core damage rate, the dependence on physical experiments can be reduced, and prediction can be made through historical data and acquired data, which can save prediction costs and time to a certain extent. On the other hand, based on the selection of optimal feature data, the effectiveness of the prediction is ensured. At the same time, the impact of multiple factors on core damage during oil field production is also taken into account, which can effectively process multi-dimensional data and improve the comprehensiveness and reliability of core permeability retention rate prediction.

[0055] In some embodiments, the above step S110 can be implemented by the following steps S111 to S113:

[0056] Step S111, collecting production water samples from the target block.

[0057] It should be noted that production water samples are important samples that directly reflect the characteristics of reservoir fluids. Their collection and analysis are the basis for evaluating reservoir status, predicting production performance, and formulating subsequent treatment measures.

[0058] In some embodiments, it is necessary to first determine the specific location, depth and geological characteristics of the target block and select representative production wells for water sampling.

[0059] Step S112, extracting characteristic parameters with a high correlation with the core permeability retention rate from the production water sample.

[0060] It should be noted that the core permeability retention rate is a key indicator for evaluating the long-term production capacity of the reservoir and the effect of water injection and oil recovery. However, directly measuring the core permeability retention rate is not only costly but also complicated to operate. Therefore, the present invention proposes to indirectly predict the core permeability retention rate by extracting key characteristic parameters from the production water sample to improve the evaluation efficiency and accuracy.

[0061] In some embodiments, the collected production water samples need to be pre-processed, such as filtering, centrifuging, etc., to remove suspended matter and impurities to ensure the accuracy of the analysis results. Through comparative experiments and data analysis, characteristic parameters with a high correlation with the core permeability retention rate are screened out.

[0062] Step S113, performing a core displacement experiment based on the characteristic parameters to obtain the core permeability retention rate data.

[0063] It should be noted that after determining the key characteristic parameters in the production water samples, it is necessary to verify the correlation between these parameters and the core permeability retention rate through experiments.

[0064] In some embodiments, core samples with similar geological characteristics to the target block are first selected, and corresponding displacement fluids are configured according to the characteristic parameters screened in the early stage. In the displacement experiment, the fluid pressure, flow rate and other conditions need to be accurately controlled to simulate the fluid flow under real reservoir conditions. The permeability retention rate data is calculated by monitoring the permeability changes of the core before and after the displacement.

[0065] In some embodiments, the characteristic parameters include iron content, sulfur ion concentration, calcium ion concentration, bicarbonate ion concentration, suspended matter concentration, median particle size of suspended particles, oil content, and mineralization.

[0066] In some embodiments, the above step S120 can be implemented by:

[0067] The core permeability retention rate data is cleaned, normalized, and processed for outliers in sequence to obtain the feature set.

[0068] In some embodiments, the above step S140 can be implemented by the following steps S141 to S143:

[0069] Step S141, calculate the gradient of the loss function for each of the parameters under the current model parameters, and adjust the parameters based on the direction of the gradient and the learning rate until the number of iterations of the loss function reaches a preset iteration threshold.

[0070] Step S142, using a validation set to validate the initial batch gradient descent model to obtain a validated batch gradient descent model.

[0071] Step S143, using a test set to test the verified batch gradient descent model to obtain the target batch gradient descent model.

[0072] The following describes an exemplary application of an embodiment of the present invention in a practical application scenario.

[0073] The present invention provides a method for predicting the core permeability retention rate, which is mainly studied by combining the core displacement experimental data with batch gradient descent, such as Figure 2 As shown, the following steps are included:

[0074] 1) Collect production water samples from the target block, and test the ion composition, water quality control indicators and auxiliary control indicators of the extracted water samples. In the feature selection process, the characteristic parameters that are highly correlated with the prediction of permeability retention rate are preferentially selected, including total iron content, sulfur ion concentration, calcium ion concentration, bicarbonate ion concentration, suspended matter concentration, median particle size of suspended particles, oil content and mineralization, etc. Carry out core displacement experiments according to the characteristics that meet the requirements, such as the characteristic parameters are: 1 / 0.5 / 841 / 463 / 2 / 1.5 / 6 / 34688mg / L, so as to obtain the core permeability retention rate data, and perform feature processing on the acquired data, such as cleaning, normalization, processing of outliers, etc., to obtain the feature set described in the target block.

[0075] 2) Build a batch gradient descent model: Convert feature data into a form suitable for model input, such as standardization or normalization, to ensure that different features have the same dimension and range.

[0076] 3) Randomly initialize the model parameters, including weights and bias terms. Define a loss function, such as mean squared error, to quantify the deviation between the model's predicted value and the true value.

[0077] 4) Use the gradient descent algorithm to gradually update the model parameters to minimize the loss function. Specifically, calculate the gradient of the loss function for each parameter under the current parameters, and then adjust the parameters according to the direction of the gradient and the learning rate (stepsize). Repeat this process until the loss function converges or reaches the preset number of iterations.

[0078] 5) Validate the batch gradient descent model and use the validation set (data not involved in training) to evaluate the model performance, such as using appropriate evaluation metrics (mean square error and R square), adjust hyperparameters to improve the generalization ability of the model, and further evaluate the model stability and performance through cross-validation methods.

[0079] 6) Test the batch gradient descent model and use an independent test set to evaluate the final performance of the model to ensure the performance of the model on real data. Deploy the trained model in practical applications to predict core damage rate.

[0080] 7) The batch gradient descent algorithm can also obtain the functional relationship between each influencing factor and the core damage rate. By analyzing the prediction results of the functional relationship, the prediction ability of the model can be explained.

[0081] Furthermore, the data prediction process includes the following steps:

[0082] (1) Data standardization: standardize the data (i.e. the data of each influencing factor).

[0083] (2) Parameter initialization: Before model training, the weights (i.e., the parameters of each feature) are randomly initialized. These weights represent the importance of each influencing factor in predicting the permeability retention rate.

[0084] (3) Simulation calculation: The batch gradient descent algorithm is used for training, and the gradient of the loss function to the model parameters is calculated using the entire data set. The model weights are then updated using the gradient to reduce the loss. This process is iterated until the loss function converges or the preset number of iterations is reached. (In order to improve generalization ability, a validation set test will be performed after training.) After training, the model will substitute the new input data into the model again and use the trained parameters for prediction to obtain a predicted value, which is the permeability retention rate predicted by the batch gradient descent algorithm model.

[0085] (4) Obtaining equations: After training is completed, the model obtains stable weight parameters. Substituting these weights into the prediction equation can form a specific permeability retention rate prediction equation.

[0086] (5) Substituting data: The actual values ​​of the experimental data obtained by simulating water displacement are compared with the predicted values ​​of the various values ​​of simulated water substituting into the equation to verify the accuracy of the equation.

[0087] The present invention provides a method for predicting core permeability retention rate using a gradient descent algorithm, comprising the following steps:

[0088] 1) Collect core flooding experimental data:

[0089] Water samples from the target area were collected, and the ion composition, water quality control index and auxiliary control index of the extracted production water samples were tested. When selecting features, features that are helpful for prediction were selected, such as total iron content, sulfur ion content, calcium ion content, bicarbonate ion content, suspended matter content, suspended matter particle size median, and oil content. Core displacement experiments were carried out according to the features that met the requirements, and 70 sets of core permeability retention rate data were obtained (the core permeability level was 10-50m D). After the data collection was completed, it was imported into Excel software for storage to facilitate subsequent data management and analysis.

[0090] 2) Construct a batch gradient descent model, perform feature processing on the acquired data, and obtain the feature set of the target block: convert the Excel of the reservoir feature data into a CSV file, read the CSV file, divide it into a 30% test set and a 70% training set, that is, perform data preprocessing. Before batch gradient descent, due to the large data differences between multiple core features and permeability, for example, there are data with large differences between 300 and 1.5 in the independent variables, which may lead to slow convergence and unclear correlation display during training. Therefore, in order to make the data at the same order of magnitude, the data is standardized according to the formula to be in [-1, 1] before batch gradient descent; normalization formula: (data-mean) / standard deviation, for example, the first independent variable suspended matter content is normalized, that is, the mean and standard deviation of all suspended matter contents in 70 groups of data are first calculated, and then a certain suspended matter content data is substituted into the formula to obtain the standardized value.

[0091] 3) Define functions to measure prediction error, such as mean square error (MSE), coefficient of determination (R 2 ); adjust the learning rate and number of iterations, for example, set the learning rate to 0.01, 0.001, 0.0001 to compare R 2 The value of Figure 3 As shown; set the number of iterations to 100000, and the learning rate to 0.01. 2 maximum, so the learning rate is set to 0.01.

[0092] 4) Use the gradient descent algorithm to update the model parameters to minimize the loss function, then validate the batch gradient descent model and use the validation set (data not involved in training) to evaluate the model performance.

[0093] 5) Start prediction. According to the trained model, we can get the comparison chart of the real value and the predicted value, as shown in Table 1. At the same time, in order to obtain the relationship between each feature and the relationship between the feature and the permeability, we continuously optimize the correlation between each feature and the permeability. The specific optimization process is shown in Figure 4 , it can be seen in the figure that as the number of iterations increases, different coefficients are assigned to each feature so that it can approximate the given permeability to the greatest extent. The functional relationship between the independent variables (total iron content, sulfur ion content, calcium ion content, bicarbonate ion content, suspended matter content, median suspended matter particle size, oil content, mineralization) and the dependent variable (core permeability retention rate) is obtained:

[0094] Y=-3.47762223e -16 -4.52181852e -01 *X1+2.79678038e -01 *X2

[0095] -5.85827924e -01 *X3-6.49768151e -01 *X4-8.06227285e -03 *X5

[0096] +8.38393340e -02 *X6-1.04274750e -01 *X7-6.47316896e -01 *X8

[0097] Table 1 Comparison of actual value and predicted value

[0098]

[0099] Example

[0100] The core permeability and grade are selected as 33 (10-50mD). Through indoor experiments, the permeability retention rate of the core is measured to be 80%. The values ​​of total iron content, sulfur ion content, calcium ion content, bicarbonate ion content, suspended matter content, median suspended matter particle size, oil content, and mineralization are substituted into the function relationship. The specific values ​​are 0.2 / 5 / 40 / 200 / 2 / 1.5 / 8 / 35000, respectively. The predicted core permeability retention rate is 79.3977%, with an error of only 0.6033%. Under the same conditions, two groups of simulated water were used to carry out core displacement experiments. At the same time, the simulated water data: 1 / 5 / 500 / 600 / 2 / 1.2 / 6 / 30000 and 2 / 0.5 / 650 / 550 / 2 / 1.5 / 6 / 32000 were substituted into the prediction equation for comparison. The results showed that the true values ​​of the two groups of data were 80% and 65.4%, respectively, and the predicted values ​​were 80.9142% and 64.1907%, respectively. The errors were 0.9142% and 1.2093%, respectively. The average error of the three groups was only 0.9089%, indicating that the model has strong applicability within this range, and the prediction results are more stable and reliable, which is very important for decision support in practical applications.

[0101] The above embodiments provide an innovative method. The innovative method proposed in this study uses a large amount of historical data for training through the gradient descent algorithm of the machine learning model, so that the model can automatically identify the potential patterns and characteristics of core damage, and improve the accuracy of prediction through continuous parameter optimization. This method not only improves the accuracy and efficiency of prediction, but also reduces the error in the prediction process by reducing the intervention of human factors. Through application verification in actual cases, it has demonstrated superior performance and reliability. The performance of the model in predicting the permeability retention rate of 10-50m D cores proves its practical value in oil and gas exploration and development. In addition, the efficiency and accuracy of this method provide strong technical support for the effective development of oil and gas resources.

[0102] Figure 5 is a schematic diagram of the composition structure of a core permeability retention rate prediction device based on a gradient algorithm provided in an embodiment of the present invention, such as Figure 5As shown, the core permeability retention rate prediction device 500 based on the gradient algorithm includes: an acquisition module 501, used to obtain the core permeability retention rate data of the target block; a preprocessing module 502, used to preprocess the core permeability retention rate data to obtain the feature set of the target block; a construction module 503, used to construct an initial batch gradient descent model based on the feature set; randomly initialize the parameters of the initial batch gradient descent model, and define a loss function; a training module 504, used to adopt a preset gradient descent algorithm, update the parameters of the initial batch gradient descent model and minimize the loss function, train the initial batch gradient descent model, and obtain a target batch gradient descent model; a substitution module 505, used to substitute the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; a prediction module 506, used to predict the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result.

[0103] In some embodiments, the acquisition module 501 is also used to collect production water samples from the target block; extract characteristic parameters with a high correlation with the core permeability retention rate from the production water samples; and perform core displacement experiments based on the characteristic parameters to obtain the core permeability retention rate data.

[0104] In some embodiments, the characteristic parameters include iron content, sulfur ion concentration, calcium ion concentration, bicarbonate ion concentration, suspended matter concentration, median particle size of suspended particles, oil content and mineralization.

[0105] In some embodiments, the preprocessing module 502 is further used to sequentially clean, normalize, and process outliers on the core permeability retention rate data to obtain the feature set.

[0106] In some embodiments, the training module 504 is also used to calculate the gradient of the loss function for each of the parameters under the current model parameters, and adjust the parameters based on the direction of the gradient and the learning rate until the number of iterations of the loss function reaches a preset iteration threshold; use a validation set to verify the initial batch gradient descent model to obtain a verified batch gradient descent model; use a test set to test the verified batch gradient descent model to obtain the target batch gradient descent model.

[0107] It should be noted that the description of the device of the embodiment of the present invention is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment, so it is not repeated. For technical details not disclosed in the device embodiment, please refer to the description of the method embodiment of the present invention for understanding.

[0108] It should be noted that, in the embodiment of the present invention, if the above-mentioned core permeability retention rate prediction method based on the gradient algorithm is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a terminal to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), disk or optical disk and other media that can store program codes. In this way, the embodiment of the present invention is not limited to any specific combination of hardware and software.

[0109] Correspondingly, an embodiment of the present invention provides a core permeability retention rate prediction device based on a gradient algorithm. Figure 6 is a schematic diagram of the composition structure of a core permeability retention rate prediction device based on a gradient algorithm provided in an embodiment of the present invention, such as Figure 6 As shown, the core permeability retention rate prediction device 600 based on the gradient algorithm includes at least: a processor 601 and a computer-readable storage medium 602 configured to store executable instructions, wherein the processor 601 generally controls the overall operation of the core permeability retention rate prediction device 600 based on the gradient algorithm. The computer-readable storage medium 602 is configured to store instructions and applications executable by the processor 601, and can also cache data to be processed or processed by each module in the processor 601 and the core permeability retention rate prediction device 600 based on the gradient algorithm, which can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0110] An embodiment of the present invention provides a storage medium storing executable instructions, wherein the executable instructions are stored. When the executable instructions are executed by a processor, the processor will be caused to execute the method provided by the embodiment of the present invention, for example, Figure 1 The method shown.

[0111] In some embodiments, the storage medium can be a computer-readable storage medium, for example, a ferroelectric random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface memory, an optical disk, or a compact disk read-only memory (CD-ROM), etc.; it can also be various devices including one or any combination of the above memories.

[0112] In some embodiments, executable instructions may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0113] As an example, executable instructions may, but need not necessarily, correspond to a file in a file system, may be stored as part of a file storing other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files storing one or more modules, subroutines, or code portions). As an example, executable instructions may be deployed to be executed on one electronic device, or on multiple electronic devices located at one location, or on multiple electronic devices distributed at multiple locations and interconnected by a communication network.

[0114] The above description is only an embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement and improvement made within the spirit and scope of the present invention are included in the protection scope of the present invention.

[0115] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the size of the serial number of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The serial numbers of the above-mentioned embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0116] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of further restrictions, an element defined by the statement "comprises one..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed.

[0117] The above is only an embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting core permeability retention rate based on a gradient algorithm, characterized in that: The method comprises: Obtain core permeability retention rate data for target blocks; Preprocessing the core permeability retention rate data to obtain a feature set of the target block; Based on the feature set, construct an initial batch gradient descent model; randomly initialize the parameters of the initial batch gradient descent model, and define a loss function; Using a preset gradient descent algorithm, updating the parameters of the initial batch gradient descent model and minimizing the loss function, training the initial batch gradient descent model, and obtaining a target batch gradient descent model; Substituting the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; Based on the permeability retention rate prediction equation, the core permeability retention rate of the target block is predicted to obtain a prediction result.

2. The method according to claim 1, characterized in that The obtaining of the core permeability retention rate data of the target block includes: Collecting production water samples from the target block; Extracting characteristic parameters with a high correlation with the core permeability retention rate from the produced water sample; A core displacement experiment is performed based on the characteristic parameters to obtain the core permeability retention rate data.

3. The method according to claim 2, characterized in that The characteristic parameters include iron content, sulfur ion concentration, calcium ion concentration, bicarbonate ion concentration, suspended matter concentration, median particle size of suspended particles, oil content and mineralization.

4. The method according to claim 1, characterized in that The preprocessing of the core permeability retention rate data to obtain the feature set of the target block includes: The core permeability retention rate data is cleaned, normalized, and processed for outliers in sequence to obtain the feature set.

5. The method according to claim 1, characterized in that The method adopts a preset gradient descent algorithm, updates the parameters of the initial batch gradient descent model and minimizes the loss function, trains the initial batch gradient descent model, and obtains a target batch gradient descent model, including: Calculate the gradient of the loss function for each of the parameters under the current model parameters, and adjust the parameters based on the direction of the gradient and the learning rate until the number of iterations of the loss function reaches a preset iteration number threshold; Using a validation set to validate the initial batch gradient descent model to obtain a validated batch gradient descent model; The verified batch gradient descent model is tested using a test set to obtain the target batch gradient descent model.

6. A core permeability retention rate prediction device based on a gradient algorithm, characterized in that: The device comprises: An acquisition module is used to obtain the core permeability retention rate data of the target block; A preprocessing module, used for preprocessing the core permeability retention rate data to obtain a feature set of the target block; A construction module is used to construct an initial batch gradient descent model based on the feature set; randomly initialize the parameters of the initial batch gradient descent model, and define a loss function; A training module, used to adopt a preset gradient descent algorithm to update the parameters of the initial batch gradient descent model and minimize the loss function, train the initial batch gradient descent model, and obtain a target batch gradient descent model; A substitution module, used to substitute the weight parameters in the target batch gradient descent model into the initial prediction equation to obtain a permeability retention rate prediction equation; The prediction module is used to predict the core permeability retention rate of the target block based on the permeability retention rate prediction equation to obtain a prediction result.

7. An electronic device, characterized in that: include: A memory for storing executable instructions; The processor is used to implement the core permeability retention rate prediction method based on the gradient algorithm as described in any one of claims 1 to 5 when executing the executable instructions stored in the memory.

8. A computer-readable storage medium, characterized in that: Executable instructions are stored, which are used to cause a processor to execute the executable instructions to implement the core permeability retention rate prediction method based on the gradient algorithm as described in any one of claims 1 to 5.

Images