A method for predicting production of a block well, a storage medium, and a program
Patent Information
- Application Number
- CN202510491540.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-18
- Publication Date
- 2026-10-23
AI Technical Summary
但是，该用于确定完井方式及其设置参数的模拟试验的限制较大，首先，模拟试验时无法对地下真实环境进行准确模拟，或者，模拟试验时的参数设置受到人为主观判断的影响较大，导致最终得到的产量输出结果极易脱离实际情况，其次，单一种类的模拟试验的应用范围较小，无法适应不同的石油天然气藏种类的变化，导致最终得到的产量输出结果存在偏差
[0034]与现有技术相比，本发明的区块井的产量预测方法至少具有以下优点：基于已有的实际生产数据和影响因素数据，从而将地下真实环境纳入考虑范围，对待预测时间节点的产量进行预测，提高预测的准确性。同时，通过对多个回归模型进行对比，从而能够减少人为的主观判断对建立的第二产量预测模型的影响。并且，通过建立的第二产量预测模型在已有的实际生产数据和影响因素数据与最终需要预测的待预测时间节点的产量之间建立联系，从而能够扩大应用范围，便于适应不同的石油天然气藏种类的变化，保证块井的产量预测的准确性。
Smart Images
Figure CN122965728A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of oil and gas development technology, and specifically relates to a method, storage medium and program for predicting the production of block wells. Background Technology
[0002] Well completion method refers to the way in which the wellbore connects to the oil and gas reservoir, as well as the well structure, wellhead equipment, and related technical measures adopted to achieve a specific connection. The choice of well completion method directly affects the production capacity and economic benefits of oil and gas wells; therefore, accurately determining the well completion method and its related settings is crucial.
[0003] Current well completion methods generally rely on simulation tests, comparing the production output results of multiple simulation tests to determine the relevant well completion methods and their settings. However, these simulation tests for determining well completion methods and their settings have significant limitations. First, simulation tests cannot accurately simulate the real underground environment, or the parameter settings during simulation tests are greatly influenced by subjective human judgment, causing the final production output results to easily deviate from reality. Second, the application scope of a single type of simulation test is limited, unable to adapt to the variations in different types of oil and gas reservoirs, resulting in deviations in the final production output results. Summary of the Invention
[0004] To address the above problems, this invention proposes a method for predicting the production of block wells, comprising the following steps:
[0005] Obtain production data and influencing factor data for each well within the block;
[0006] Based on the production data and the influencing factor data, a training set, a test set, and a validation set are obtained.
[0007] Based on the training set, corresponding regression models are built according to multiple ensemble learning algorithms;
[0008] Based on the test set, the accuracy of each regression model is determined, and the first production prediction model for the block is identified.
[0009] Based on the first production forecasting model, a second production forecasting model is optimized and established, and the production at the time node to be predicted is determined through the second production forecasting model.
[0010] In some specific embodiments, based on the production data and the influencing factor data, the training set, test set, and validation set are divided and obtained, including the following steps:
[0011] The production data and the influencing factor data are sorted according to multiple existing time points;
[0012] The production data and the influencing factor data are divided according to a preset ratio to obtain the training set, the test set, and the validation set.
[0013] In some specific embodiments, after dividing and obtaining the training set, test set, and validation set based on the production data and the influencing factor data, the following steps are set:
[0014] The data in the training set, the test set, and the validation set are compressed to a preset range.
[0015] In some specific embodiments, based on the test set, the accuracy of each regression model is determined to identify the first output prediction model for the block, including the following steps:
[0016] Based on the test set, the output value of each regression model is obtained and compared with the production data to determine the accuracy of each regression model;
[0017] Based on the accuracy of each regression model, the first production prediction model for the block is determined.
[0018] In some specific embodiments, a second production prediction model is optimized and established based on the first production prediction model, and the production at the time node to be predicted is determined through the second production prediction model, including the following steps:
[0019] Based on the first production prediction model, the controllable engineering parameters and well completion method parameters to be optimized in the influencing factor data are determined;
[0020] The controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model are optimized, and a second production prediction model is established based on the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model.
[0021] Based on the second production forecasting model, and combined with the time node to be predicted, the production at the time node to be predicted is obtained.
[0022] In some specific embodiments, based on the first production prediction model, determining the controllable engineering parameters and well completion method parameters to be optimized from the influencing factor data includes the following steps:
[0023] Based on the first production prediction model, obtain the SHAP value of each of the influencing factor data and sort them;
[0024] Based on the SHAP value of each of the influencing factors, obtain the controllable engineering parameters and well completion method parameters that meet the preset requirements for SHAP value optimization.
[0025] In some specific embodiments, the controllable engineering parameters to be optimized, the well completion method parameters, and the hyperparameters of the first production prediction model are optimized. Based on the optimized controllable engineering parameters, the well completion method parameters, and the hyperparameters of the first production prediction model, a second production prediction model is established, including the following steps:
[0026] Based on the controllable engineering parameters to be optimized, the well completion method parameters, and the hyperparameters of the first production prediction model, multiple optimizations are performed using the Bayesian method to obtain the optimized controllable engineering parameters, the well completion method parameters, and the hyperparameters of the first production prediction model.
[0027] Based on the validation set, the optimized combination of controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model is determined.
[0028] Based on the parameter combination, the second production prediction model is established.
[0029] In some specific embodiments, before obtaining the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model by performing multiple optimizations using Bayesian methods, the following steps are set:
[0030] Randomly select the controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model from the preset group to be optimized for trial calculation, and obtain the corresponding output values respectively.
[0031] A computer storage medium based on the same concept, the computer storage medium storing one or more instructions;
[0032] When executed by one or more processors, the instructions cause one or more processors to implement the block well production prediction method described in any of the above specific embodiments.
[0033] A computer program product based on the same concept, wherein the computer program product stores at least one computer program, which is loaded and executed by a processor to enable the processor to execute the block well production prediction method described in any of the above specific embodiments.
[0034] Compared with existing technologies, the block well production prediction method of the present invention has at least the following advantages: Based on existing actual production data and influencing factor data, it incorporates the real underground environment into the prediction, thereby improving the accuracy of the prediction. Simultaneously, by comparing multiple regression models, the influence of subjective human judgment on the established second production prediction model can be reduced. Furthermore, by establishing a connection between the existing actual production data and influencing factor data and the production at the final predicted time point through the established second production prediction model, the application scope can be expanded, making it easier to adapt to changes in different types of oil and gas reservoirs and ensuring the accuracy of block well production prediction.
[0035] The computer storage medium of the present invention, being used to implement the production prediction method for block wells described above, has the same beneficial effects as the production prediction method for block wells described above, and therefore will not be repeated here.
[0036] The computer program product of the present invention is used to implement the production prediction method for block wells described above, and therefore has the same beneficial effects as the production prediction method for block wells described above, so it will not be described again here.
[0037] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description and the drawings. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 A schematic diagram of a block well production prediction method according to an embodiment of the present invention is shown;
[0040] Figure 2 This diagram illustrates how each output-affecting parameter in an embodiment of the present invention is sorted from largest to smallest according to its SHAP value. Detailed Implementation
[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0042] Reference Figure 1 This invention provides a method for predicting the production of block wells, comprising the following steps:
[0043] Acquire production data and influencing factor data for each well within the block. Based on the production data and influencing factor data, divide the block into training, test, and validation sets. Based on the training set, build corresponding regression models using multiple ensemble learning algorithms. Based on the test set, assess the accuracy of each regression model to determine the first production prediction model for the block. Based on the first production prediction model, optimize and build a second production prediction model, and use the second production prediction model to determine the production at the time node to be predicted.
[0044] Specifically, the existing actual production data and influencing factor data for each well within the block are acquired. Based on this data, the existing actual production data and influencing factor data are sorted and divided into training, testing, and validation sets according to a certain ratio. First, based on the existing actual production data and influencing factor data in the training set, regression models are built one-to-one with each of the multiple ensemble learning algorithms. Second, by substituting the existing actual production data and influencing factor data in the testing set into each regression model, the accuracy of each regression model is obtained and judged. Based on the accuracy of each regression model, one model is selected as the first production prediction model. Then, the controllable parameters of the first production prediction model are optimized to establish an optimized second production prediction model. Finally, the production output corresponding to the predicted time point is obtained by substituting the predicted time point into the second production prediction model.
[0045] This approach incorporates existing actual production data and influencing factor data to take into account the real underground environment, thereby improving the accuracy of production forecasts for the desired timeframe. Furthermore, by comparing multiple regression models and optimizing the established first production forecasting model, the influence of subjective human judgment on the established second production forecasting model can be reduced. Finally, by establishing a link between the existing actual production data and influencing factor data and the production at the desired timeframe in the final production forecast, the application scope can be broadened, making it easier to adapt to changes in different types of oil and gas reservoirs and ensuring the accuracy of production forecasts for block wells.
[0046] It should be noted that the existing actual production data and influencing factor data for each well within the block include the geological, engineering, and production data of each well to be evaluated within the same block. Specifically, this includes production output data after a determined production period since well commencement, as well as the corresponding production influencing factor data. The production influencing factor data mainly consists of geological parameters, engineering parameters, and well data. The determined production period can be 100 days, 350 days, 700 days, 1000 days, etc., but this determined production period must be a time point in time when corresponding production output data can be collected for each well to be evaluated.
[0047] Examples of data can be found in Tables 1-4 below.
[0048] Table 1. Well Parameters
[0049]
[0050]
[0051]
[0052] Table 2. Geological parameters
[0053]
[0054] Table 3. Engineering Parameters
[0055]
[0056]
[0057] Table 4. Production output data after production duration is determined
[0058] Serial Number 100 days 350 days 700 days 1100 days 1 154672 84898 66490 59868 2 191823 199815 220561 239013 3 525871 469722 440838 409385 4 223212 239421 266987 265411 5 227069 208934 181116 163679
[0059] In some specific embodiments, based on production data and influencing factor data, the acquisition of training, testing, and validation sets is divided, including the following steps:
[0060] The production data and influencing factor data are sorted according to multiple existing time points. The production data and influencing factor data are then divided according to a preset ratio to obtain training, testing, and validation sets.
[0061] Specifically, after obtaining the production output data after a certain production duration and the corresponding output influencing factor data, all the acquired data are sorted according to the chronological order of the production duration and divided into training, testing, and validation sets according to a preset ratio for subsequent processing. The preset ratio is determined based on the amount of data; for example, when the amount of data is small, the ratio of the training set to the testing set is 7:3 or 6:4.
[0062] In some specific embodiments, after dividing and obtaining training, testing, and validation sets based on production data and influencing factor data, the following steps are set:
[0063] The training set data is compressed to a preset range. Based on the peak and trough values of the training set data, the test and validation sets are also compressed to a preset range.
[0064] Specifically, the training set data is subjected to min-max normalization, which compresses all existing actual production data and influencing factor data contained in the training set into a preset range. This ensures that the magnitude of parameter data of different features in the training set is consistent, preventing some features from becoming dominant due to excessively large values. If the numerical differences between features are too large, the subsequent model may prioritize features with larger values and ignore features with smaller values, thus affecting the final prediction effect and ensuring the accuracy of the final prediction result.
[0065] At the same time, after the training set data is compressed, the test set and validation set are also compressed in the same way, so that the parameter data of different features in the test set and validation set are of the same magnitude.
[0066] It should be noted that maximum-minimum normalization is an existing data processing method. By obtaining the peak and valley values of the corresponding data, data compression can be completed.
[0067] In some specific embodiments, based on the training set, corresponding regression models are built according to multiple ensemble learning algorithms, including the following steps:
[0068] Determine the specific category of each ensemble learning algorithm to be selected, and build a regression model for each ensemble learning algorithm.
[0069] Specifically, each ensemble learning algorithm processes the existing actual production data and influencing factor data contained in the training set, thus creating a one-to-one correspondence between the ensemble learning algorithm and the corresponding regression model. Generally, the specific categories of ensemble learning algorithms include random forest, gradient boosting decision tree, extreme gradient boosting, LightGBM, and extreme random tree algorithms. By selecting multiple different ensemble learning algorithms, it is easier to compare the output results after building the corresponding regression models, thereby improving the accuracy of the final prediction results.
[0070] In some specific embodiments, based on the test set, the accuracy of each regression model is determined to identify the first yield prediction model for the block, including the following steps:
[0071] Based on the test set, the output value of each regression model is obtained and compared with the production data to determine the accuracy of each regression model. Based on the accuracy of each regression model, the first production prediction model for the block is determined.
[0072] Specifically, after obtaining the regression model corresponding to each ensemble learning algorithm, the existing actual influencing factor data contained in the test set are substituted into each model to obtain the output value of each regression model. Then, the output value of each regression model is compared with the existing actual production data contained in the test set to determine the accuracy of each regression model, thereby determining the first production prediction model for the block. After calculating the accuracy of each regression model, the accuracy of each regression model is ranked, and the regression model with the highest accuracy is selected as the first production prediction model for the block.
[0073] It should be noted that during the calculation, it is necessary to calculate the accuracy of each regression model at each time point with a certain production duration. For example, the accuracy of each regression model when the production duration is 100 days needs to be calculated separately, and the accuracy of each regression model when the production duration is 350 days also needs to be calculated separately. Specifically, the accuracy of each regression model at different production durations needs to be compared separately. Referring to Table 5, when the production duration is 100 days, the accuracy of each regression model is ranked, and the regression model with the highest accuracy is the first production prediction model for the block when the production duration is 100 days. Alternatively, when the production duration is 350 days, the accuracy of each regression model is ranked again, and the regression model with the highest accuracy in this case is the first production prediction model for the block when the production duration is 350 days. Applying different regression models as the first production prediction model for the block at different time points can further improve the accuracy of the prediction results.
[0074] Table 5. Ranking of Regression Model Accuracy
[0075]
[0076]
[0077] In some specific embodiments, a second production forecasting model is optimized and established based on the first production forecasting model, and the production at the time node to be predicted is determined through the second production forecasting model, including the following steps:
[0078] Based on the first production prediction model, the controllable engineering parameters and well completion method parameters to be optimized in the influencing factor data are determined. The controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, are optimized. Based on the optimized controllable engineering parameters and well completion method parameters, and the hyperparameters of the first production prediction model, a second production prediction model is established. Based on the second production prediction model, and combined with the time node to be predicted, the production at the determined time node is obtained.
[0079] Specifically, after determining the first production prediction model, the controllable engineering parameters among the production-influencing parameters in the influencing factor data that need optimization are identified as the optimization objects. Simultaneously, the existing well completion method parameters and the hyperparameters of the first production prediction model are also optimized. Based on the optimized controllable engineering parameters, well completion method parameters, and the hyperparameters of the first production prediction model, a corresponding second production prediction model is established. Once the second production prediction model is established, the production corresponding to the predicted time node can be obtained by substituting the predicted time node into the second production prediction model.
[0080] In some specific embodiments, based on the first production prediction model, the controllable engineering parameters and well completion method parameters to be optimized in the influencing factor data are determined, including the following steps:
[0081] Based on the first production prediction model, the SHAP value of each influencing factor is obtained and sorted. Based on the SHAP value of each influencing factor, controllable engineering parameters and well completion method parameters that meet preset requirements are obtained for optimization.
[0082] Specifically, after determining the first output forecasting model, the SHAP value of the output impact parameter for each influencing factor data in the first output forecasting model is calculated, referring to... Figure 2Each production-affecting parameter is sorted from largest to smallest according to its SHAP value. Then, the top ten controllable engineering parameters with the highest SHAP values among the production-affecting parameters are selected as optimization objects. At the same time, the existing well completion method parameters and the hyperparameters of the first production prediction model are also selected as optimization objects, thereby determining the controllable engineering parameters and well completion method parameters to be optimized.
[0083] It's important to note that SHAP (Shapley Additive Explanations) is a method for interpreting model predictions, measuring the contribution of each feature to the model's output. By assigning a corresponding "contribution value" based on the importance of each feature, the prediction results of the model can be interpreted.
[0084] In some specific embodiments, the controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, are optimized. Based on the optimized controllable engineering parameters and well completion method parameters, as well as the hyperparameters of the first production prediction model, a second production prediction model is established, including the following steps:
[0085] Based on the controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, multiple optimizations are performed using the Bayesian method to obtain the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model. Based on the validation set, the parameter combinations of the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model are determined. Based on these parameter combinations, a second production prediction model is established.
[0086] Specifically, using Bayesian optimization methods, the determined controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model are optimized. Through iterative optimization, the results are validated using a validation set, thus determining the final and unique combination of optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model. This determined parameter combination is then substituted into the first production prediction model to update it, thereby establishing the second production prediction model.
[0087] It should be noted that the Bayesian optimization method is an existing optimization method. By selecting sampling points, it continuously tries each set of controllable engineering parameters, well completion parameters, and hyperparameters of the first production prediction model, ultimately obtaining a unique optimized combination of controllable engineering parameters, well completion parameters, and hyperparameters of the first production prediction model. The validation set serves a validation purpose; that is, the final unique optimized combination of controllable engineering parameters, well completion parameters, and hyperparameters of the first production prediction model should be the combination that makes the output result closest to the validation set data.
[0088] In some specific embodiments, based on the controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, multiple optimizations are performed using the Bayesian method. Before obtaining the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model, the following steps are set:
[0089] Randomly select the preset controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, and perform trial calculations to obtain the corresponding output values.
[0090] Specifically, before continuously trying each set of controllable engineering parameters, well completion parameters, and hyperparameters of the first production prediction model using Bayesian optimization, a pre-selected set of controllable engineering parameters, well completion parameters, and hyperparameters of the first production prediction model are randomly selected for trial calculations. The corresponding representative production output values are then obtained. This process of calculating production using random initial parameter combinations serves as the first step in the Bayesian optimization method. This approach covers different regions of the parameter space, preventing parameter selection results from being too concentrated in a localized area. This facilitates further comprehensive exploration and helps find the global optimum rather than a local optimum. The specific number of pre-selected sets is manually set.
[0091] This invention also provides a computer storage medium storing one or more instructions. When executed by one or more processors, these instructions cause the processors to implement the block well production prediction method described in any of the above embodiments. This computer storage medium is used to implement the block well production prediction method. By referencing existing actual production data and influencing factor data, it incorporates the real underground environment into the calculation, thereby improving the accuracy of production prediction at the desired time point. Simultaneously, by comparing multiple regression models and optimizing the determined first production prediction model, the influence of subjective human judgment on the established second production prediction model can be reduced. Furthermore, by establishing a link between the existing actual production data and influencing factor data and the production at the desired time point through the finally established second production prediction model, the application scope can be expanded, making it easier to adapt to changes in different types of oil and gas reservoirs and ensuring the accuracy of block well production prediction.
[0092] This invention also provides a computer program product storing at least one computer program, which is loaded and executed by a processor to enable the processor to execute the block well production prediction method described in any of the above specific embodiments. This computer program product is used to execute the block well production prediction method. By referencing existing actual production data and influencing factor data, it incorporates the real underground environment into the calculation, improving the accuracy of production prediction at the desired time point. Simultaneously, by comparing multiple regression models and optimizing the determined first production prediction model, the influence of subjective human judgment on the established second production prediction model can be reduced. Furthermore, by establishing a connection between the existing actual production data and influencing factor data and the production at the desired time point through the finally established second production prediction model, the application scope can be expanded, facilitating adaptation to changes in different types of oil and gas reservoirs and ensuring the accuracy of block well production prediction.
[0093] It should be noted that the terms "first," "second," etc., used in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein.
[0094] In the description of this specification, the references to terms such as "embodiment," "example," and "specific example" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0095] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.
Claims
1. A method for predicting the production of block wells, characterized in that, Includes the following steps: Obtain production data and influencing factor data for each well within the block; Based on the production data and the influencing factor data, a training set, a test set, and a validation set are obtained. Based on the training set, corresponding regression models are built according to multiple ensemble learning algorithms; Based on the test set, the accuracy of each regression model is determined, and the first production prediction model for the block is identified. Based on the first production forecasting model, a second production forecasting model is optimized and established, and the production at the time node to be predicted is determined through the second production forecasting model.
2. The method for predicting production from block wells according to claim 1, characterized in that, Based on the production data and the influencing factor data, the training set, test set, and validation set are obtained, including the following steps: The production data and the influencing factor data are sorted according to multiple existing time points; The production data and the influencing factor data are divided according to a preset ratio to obtain the training set, the test set, and the validation set.
3. The method for predicting production from block wells according to claim 1, characterized in that, Based on the production data and the influencing factor data, after dividing and obtaining the training set, test set, and validation set, the following steps are set: The data in the training set, the test set, and the validation set are compressed to a preset range.
4. The method for predicting production from block wells according to claim 1, characterized in that, Based on the test set, the accuracy of each regression model is determined, and the first production prediction model for the block is identified, including the following steps: Based on the test set, the output value of each regression model is obtained and compared with the production data to determine the accuracy of each regression model; Based on the accuracy of each regression model, the first production prediction model for the block is determined.
5. The method for predicting production from block wells according to claim 1, characterized in that, Based on the first production forecasting model, a second production forecasting model is optimized and established. The production output at the time node to be forecasted is then determined using the second production forecasting model, including the following steps: Based on the first production prediction model, the controllable engineering parameters and well completion method parameters to be optimized in the influencing factor data are determined; The controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model are optimized, and a second production prediction model is established based on the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model. Based on the second production forecasting model, and combined with the time node to be predicted, the production at the time node to be predicted is obtained.
6. The method for predicting production from block wells according to claim 5, characterized in that, Based on the first production prediction model, the controllable engineering parameters and well completion method parameters to be optimized in the influencing factor data are determined, including the following steps: Based on the first production prediction model, obtain the SHAP value of each of the influencing factor data and sort them; Based on the SHAP value of each of the influencing factors, obtain the controllable engineering parameters and well completion method parameters that meet the preset requirements for SHAP value optimization.
7. The method for predicting production from block wells according to claim 5, characterized in that, The controllable engineering parameters and well completion method parameters to be optimized, as well as the hyperparameters of the first production prediction model, are optimized. Based on the optimized controllable engineering parameters and well completion method parameters, as well as the hyperparameters of the first production prediction model, a second production prediction model is established, including the following steps: Based on the controllable engineering parameters to be optimized, the well completion method parameters, and the hyperparameters of the first production prediction model, multiple optimizations are performed using the Bayesian method to obtain the optimized controllable engineering parameters, the well completion method parameters, and the hyperparameters of the first production prediction model. Based on the validation set, the optimized combination of controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model is determined. Based on the parameter combination, the second production prediction model is established.
8. The method for predicting production from block wells according to claim 7, characterized in that, Based on the controllable engineering parameters to be optimized, the well completion method parameters, and the hyperparameters of the first production prediction model, multiple optimizations are performed using the Bayesian method. Before obtaining the optimized controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model, the following steps are set: Randomly select the controllable engineering parameters, well completion method parameters, and hyperparameters of the first production prediction model from the preset group to be optimized for trial calculation, and obtain the corresponding output values respectively.
9. A computer storage medium, characterized in that, The computer storage medium stores one or more instructions; When executed by one or more processors, the instructions cause one or more processors to implement the block well production prediction method according to any one of claims 1 to 8.
10. A computer program product, characterized in that, The computer program product stores at least one computer program, which is loaded and executed by a processor to enable the processor to perform the block well production prediction method according to any one of claims 1 to 8.