Method and device for predicting single-layer productivity of multi-layer gas reservoir

Abstract

The invention provides a method and a device for predicting the single-layer productivity of a multi-layer gas reservoir. Obtaining target features of a target layer related to the target well; wherein the target features comprise geological features and fracturing features; obtaining an initial prediction capacity corresponding to the target layer based on the target features of the target layer by using a preset capacity prediction model; wherein the preset productivity model is obtained by combining contribution degree analysis and feature weight adjustment in advance and utilizing sample data training; and according to the initial prediction capacity corresponding to the target layer, an exploitation strategy for the target layer of the target well is determined. Based on the method, the geological features and the fracturing features of the target layer are deeply analyzed by presetting the productivity prediction model and combining contribution degree analysis with feature weight adjustment, so that fine decoupling between the initial prediction productivity and each key factor is realized, the precision and the interpretation ability of single-layer gas production prediction are improved, and the prediction efficiency is improved. And powerful support is provided for formulating a more accurate and scientific mining strategy.

Description

Technical Field

[0001] This specification belongs to the technical field of oil extraction, and particularly relates to a method and device for predicting the single-layer productivity of a multi-layer gas reservoir. Background Art

[0002] At present, tight sandstone gas reservoirs have complex geological characteristics and development difficulties, with characteristics such as the development of multiple thin layers, large differences in reservoir characteristics, and difficulties in production increase. This makes the prediction of gas production per well and per layer particularly complex, and there are problems of low accuracy and poor interpretability in the prediction of gas production per well and per layer.

[0003] In response to the above problems, no effective solution has been proposed yet. Summary of the Invention

[0004] This specification provides a method and device for predicting the single-layer productivity of a multi-layer gas reservoir. By presetting a productivity prediction model and combining contribution degree analysis with feature weight adjustment, the geological characteristics and fracturing characteristics of the target layer are deeply analyzed, realizing a fine decoupling between the initial predicted productivity and various key factors, improving the accuracy and interpretability of the single-layer gas production prediction, and providing strong support for formulating more accurate and scientific exploitation strategies.

[0005] This specification provides a method for predicting the single-layer productivity of a multi-layer gas reservoir, including:

[0006] Obtain the target features of the target layer of the target well; wherein, the target features include geological features and fracturing features;

[0007] Using a preset productivity prediction model, based on the target features of the target layer, obtain the initial predicted productivity corresponding to the target layer; wherein, the preset productivity model is obtained by pre-combining contribution degree analysis and feature weight adjustment and training with sample data.

[0008] According to the initial predicted productivity corresponding to the target layer, determine the exploitation strategy for the target layer of the target well.

[0009] In one embodiment, before using the preset productivity prediction model to obtain the initial predicted productivity corresponding to the target layer based on the target features of the target layer, it further includes:

[0010] Obtain and use the target features of the sample layer and the corresponding first initial productivity to construct sample data;

[0011] Construct an initial productivity prediction model according to the XGBOOST algorithm;

[0012] Use the sample data to perform the first training on the initial productivity prediction model to obtain a first intermediate model;

[0013] Detect whether the first intermediate model meets the requirements according to the sample data;

[0014] In the case of determining that the first intermediate model does not meet the requirements, determine the first weight parameter for the target feature by performing SHAP value analysis according to the sample data;

[0015] Adjust the first intermediate model according to the first weight parameter of the target feature to obtain an adjusted first intermediate model;

[0016] Perform second training on the adjusted first intermediate model using the sample data to obtain a preset production capacity prediction model that meets the requirements.

[0017] In one embodiment, when constructing the initial production capacity prediction model, it further includes:

[0018] Determine the initial weight parameter of the target feature according to the sample data through a preset analytic hierarchy process;

[0019] Construct an initial production capacity prediction model based on the XGBOOST algorithm according to the initial weight parameter.

[0020] In one embodiment, the step of performing second training on the adjusted first intermediate model using the sample data to obtain a preset production capacity prediction model that meets the requirements includes:

[0021] Perform second training on the adjusted first intermediate model using the sample data to obtain a second intermediate model;

[0022] Detect whether the second intermediate model meets the requirements according to the sample data;

[0023] In the case of determining that the second intermediate model does not meet the requirements, determine the second weight parameter for the target feature by performing SHAP value analysis according to the sample data;

[0024] Adjust the second intermediate model according to the second weight parameter of the target feature to obtain an adjusted second intermediate model;

[0025] Perform third training on the adjusted second intermediate model using the sample data to obtain a preset production capacity prediction model that meets the requirements.

[0026] In one embodiment, the step of determining the first weight parameter for the target feature by performing SHAP value analysis according to the sample data includes:

[0027] By performing SHAP value analysis based on the sample data, the marginal contribution degree of each target feature to production capacity prediction is normalized, and a first weight parameter for the target feature is determined according to the marginal contribution degree obtained after normalization.

[0028] In one embodiment, the obtaining of the target feature corresponding to the sample layer includes:

[0029] Obtain the first features corresponding to the sample layer and construct a heat matrix between the first features;

[0030] According to the heat matrix, determine the correlation degree between the first features and the correlation degree between each first feature and the first initial production capacity;

[0031] According to the correlation degree between the first features and the correlation degree between each first feature and the first initial production capacity, perform a screening process on the first features to obtain the target features corresponding to the sample layer, and the target features corresponding to the sample layer include horizon identification, sand addition amount per unit thickness, sand-carrying fluid per unit thickness, net pressure, effective thickness, formation coefficient, and energy storage coefficient.

[0032] In one embodiment, the obtaining of the target feature of the target layer of the target well includes:

[0033] Obtain the initial target features corresponding to the target layer;

[0034] Obtain the average value and standard deviation corresponding to the initial target features;

[0035] Obtain the difference between each initial target feature and the average value, and perform an elimination process on the initial target features whose difference is greater than a preset multiple of the standard deviation to obtain the target features of the target layer of the target well.

[0036] This specification provides a single-layer production capacity prediction device for a multi-layer gas reservoir, including:

[0037] A feature acquisition module, configured to acquire the target features of the target layer of the target well; wherein, the target features include geological features and fracturing features;

[0038] A production capacity prediction module, configured to use a preset production capacity prediction model to obtain the initial predicted production capacity corresponding to the target layer based on the target features of the target layer; wherein, the preset production capacity model is trained in advance by combining contribution degree analysis and feature weight adjustment using sample data;

[0039] A strategy determination module, configured to determine an exploitation strategy for the target layer of the target well according to the initial predicted production capacity corresponding to the target layer.

[0040] The present specification also provides an electronic device, including a processor and a memory for storing instructions executable by the processor. When the processor executes the instructions, a method for predicting the single-layer productivity of a multi-layer gas reservoir is implemented.

[0041] The present specification also provides a computer-readable storage medium, on which computer instructions are stored. When the instructions are executed, a method for predicting the single-layer productivity of a multi-layer gas reservoir is implemented.

[0042] Based on a method for predicting the single-layer productivity of a multi-layer gas reservoir provided in the present specification, target features of a target layer of a target well are obtained. Among them, the target features include geological features and fracturing features. Using a preset productivity prediction model, based on the target features of the target layer, an initial predicted productivity corresponding to the target layer is obtained. Among them, the preset productivity model is obtained by training using sample data in combination with contribution degree analysis and feature weight adjustment. According to the initial predicted productivity corresponding to the target layer, a mining strategy for the target layer of the target well is determined. In this way, through the preset productivity prediction model, by combining contribution degree analysis and feature weight adjustment, the geological features and fracturing features of the target layer are deeply analyzed, realizing the fine decoupling between the initial predicted productivity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and also providing strong support for formulating a more accurate and scientific mining strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present specification, the drawings required for use in the embodiments will be briefly introduced below. The drawings described below are only some embodiments recorded in the present specification. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 is a schematic flowchart of a method for predicting the single-layer productivity of a multi-layer gas reservoir provided by an embodiment of the present specification;

[0045] Figure 2 is a schematic diagram of the structural composition of an electronic device provided by an embodiment of the present specification;

[0046] Figure 3 is a schematic diagram of the structural composition of a device for predicting the single-layer productivity of a multi-layer gas reservoir provided by an embodiment of the present specification;

[0047] Figure 4 is a schematic diagram of the technical idea for predicting the initial productivity of a single well and single layer of a multi-layer tight sandstone gas reservoir provided by an embodiment of the present specification;

[0048] Figure 5It is a heat map of the correlation between target features and between target features and initial production capacity provided by an embodiment of this specification;

[0049] Figure 6 It is a feature importance map of visual SHAP values of features provided by an embodiment of this specification;

[0050] Figure 7 It is a line chart comparing the actual values and predicted values of the training set of the production capacity prediction model provided by an embodiment of this specification;

[0051] Figure 8 It is a line dot chart comparing the actual values and predicted values of the training set of the production capacity prediction model provided by an embodiment of this specification. Detailed implementation manners

[0052] In order to enable those skilled in the art of this technology to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification with reference to the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of this specification.

[0053] Tight sandstone gas reservoirs have complex geological characteristics and development difficulties. Among them, characteristics such as the development of multiple thin layers, large differences in reservoir characteristics, and difficulties in production increase make the development of such gas reservoirs face great challenges. Among these challenges, hydraulic fracturing technology, as a common mining method, promotes the release of gas flow by increasing the permeability of the reservoir. However, since different layers in tight sandstone gas reservoirs often have different physical and chemical properties, the prediction of gas production per well and per layer is particularly complex. The quantitative prediction of the initial gas production per well and per layer, as an important technical link in the development of tight sandstone gas reservoirs, plays an extremely important role in aspects such as reasonably allocating mining resources, improving development efficiency, and reducing development costs. Through accurate prediction, not only can the technical scheme of multi-layer combined production be effectively formulated, but also a scientific basis can be provided for further engineering optimization and long-term development, promoting the progress and maturity of tight sandstone gas reservoir development technology. Analyzing the main controlling factors affecting the initial production capacity is of great significance for further optimizing the development policy of tight sandstone gas reservoirs.

[0054] In recent years, machine learning algorithms have achieved many results in the oil and gas field. Various oil and gas production-related studies use machine learning algorithms as tools for classification and regression problems. Currently, the main machine learning methods applied to oil include algorithms such as support vector machines, fully connected neural networks (FCNNs), convolutional neural networks (CNNs), long short-term memory neural networks (LSTMs), random forests (RFs), and gradient boosting decision trees (GBDTs). XGBOOST is an efficient machine learning algorithm that implements gradient boosting. By integrating multiple decision trees and using the gradient boosting method to gradually correct the model error, the prediction accuracy of the model is improved.

[0055] There has been little research on the initial production capacity prediction of a single well and single layer in a multi-layered tight sandstone gas reservoir by current prediction methods. Moreover, the accuracy and interpretability of existing prediction models are poor, and the weights of various factors in the prediction process are unknown. Therefore, the existing technology lacks a method that can predict the initial production capacity of a single well and single layer in a multi-layered tight sandstone gas reservoir and has strong interpretability.

[0056] Regarding the root cause of the above problems, this specification considers presetting a production capacity prediction model, combining contribution degree analysis with feature weight adjustment, deeply analyzing the geological features and fracturing features of the target layer, realizing a fine decoupling between the initially predicted production capacity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and providing strong support for formulating a more accurate and scientific production strategy.

[0057] Refer to Figure 1 As shown, the embodiments of this specification provide a method for predicting the production capacity of a single layer in a multi-layered gas reservoir, which is specifically applied to the server side. Specifically implemented, the method may include the following:

[0058] S101: Obtain the target features of the target layer of the target well; wherein, the target features include geological features and fracturing features;

[0059] S102: Use the preset production capacity prediction model to obtain the initially predicted production capacity corresponding to the target layer based on the target features of the target layer; wherein, the preset production capacity model is obtained by training using sample data in advance by combining contribution degree analysis and feature weight adjustment;

[0060] S103: Determine the production strategy for the target layer of the target well according to the initially predicted production capacity corresponding to the target layer.

[0061] Among them, the above geological features include horizon code, effective thickness, total porosity, permeability, gas saturation, shale content, formation coefficient, and energy storage coefficient; the above fracturing features include sand addition per unit thickness, sand-carrying fluid per unit thickness, fracture pressure, and net pressure.

[0062] In some embodiments, the initial predicted production capacity corresponding to the target layer is obtained by using a preset production capacity prediction model based on the target features of the target layer; wherein, the preset production capacity model is obtained by training with sample data in advance by combining contribution analysis and feature weight adjustment. Specifically, in implementation, it may include:

[0063] In the preset production capacity prediction model, first, data preprocessing and standardization are performed on the target features (including geological features and fracturing features) of the target layer to ensure data quality and consistency. Then, a strategy of combining contribution analysis and feature weight adjustment in advance is used to train the model with sample data, and the SHAP algorithm is adopted for contribution analysis. Specifically, based on the Shapley value theory, the SHAP algorithm calculates the marginal contribution degree of each feature to the predicted production capacity under each combination by traversing different feature combinations, assigns reasonable weights to each combination, and finally obtains the normalized SHAP values, which accurately reflect the relative influence of each feature in the model prediction. Then, the model uses these SHAP values as feedback to dynamically adjust the feature weights and related parameters, so that the key influencing factors are fully amplified and utilized during the training process, forming a closed-loop feedback mechanism. Finally, based on the target features of the target layer and the model parameters with optimized feature weights, the preset production capacity prediction model can output the initial predicted production capacity corresponding to the target layer.

[0064] In some embodiments, the exploitation strategy for the target layer of the target well is determined according to the initial predicted production capacity corresponding to the target layer. Specifically, in implementation, it may include:

[0065] Based on the initial predicted production capacity of the target layer, factors such as geological features, reservoir physical properties, and fluid properties are combined to evaluate the development potential and economic benefits of the target layer. According to the evaluation results, corresponding exploitation plans are formulated, including determining a reasonable production pressure difference, optimizing fracturing parameters, selecting an appropriate oil production method, etc. At the same time, considering the interlayer interference effect, the production sequence and production ratio of each layer are reasonably arranged to achieve the overall optimal development effect. During the exploitation process, production data is monitored in real time, and the exploitation strategy is dynamically adjusted to ensure the efficient and stable production of the target well.

[0066] In some embodiments, the target features of the target layer of the target well are obtained. Specifically, in implementation, it may include:

[0067] By collecting downhole logging data, seismic exploration data, and fracturing construction records, the geological structure and reservoir characteristics of the target well area are analyzed in detail to determine basic information such as the horizons, thicknesses, and fracture distributions of each target layer. Secondly, the collected data is preprocessed, including denoising, standardization, and feature extraction, to extract key geological features (such as horizon identification, effective thickness, formation coefficient, energy storage coefficient, etc.) and fracturing features (such as sand addition per unit thickness, sand-carrying fluid per unit thickness, net pressure, etc.).

[0068] Based on the above embodiments, through a preset production capacity prediction model, by combining contribution degree analysis and feature weight adjustment, the geological features and fracturing features of the target layer are deeply analyzed, realizing a fine decoupling between the initial predicted production capacity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and providing strong support for formulating more accurate and scientific exploitation strategies.

[0069] In some embodiments, before obtaining the initial predicted production capacity corresponding to the target layer by using the preset production capacity prediction model based on the target features of the target layer, when the method is specifically implemented, the following content may further be included:

[0070] S1: Obtain and utilize the target features of the sample layer and the corresponding first initial production capacity to construct sample data;

[0071] S2: Construct an initial production capacity prediction model according to the XGBOOST algorithm;

[0072] S3: Use the sample data to perform the first training on the initial production capacity prediction model to obtain a first intermediate model;

[0073] S4: Detect whether the first intermediate model meets the requirements according to the sample data;

[0074] S5: In the case of determining that the first intermediate model does not meet the requirements, determine the first weight parameter regarding the target features by performing SHAP value analysis according to the sample data;

[0075] S6: Adjust the first intermediate model according to the first weight parameter of the target features to obtain an adjusted first intermediate model;

[0076] S7: Use the sample data to perform the second training on the adjusted first intermediate model to obtain a preset production capacity prediction model that meets the requirements.

[0077] Among them, the above SHAP value analysis can be a model post - hoc interpretation method (SHapley Additive exPlanation, SHAP). Using the Shapley value idea in cooperative game theory, by calculating the marginal contribution of each feature in all possible feature combinations and then taking a weighted average with the combination weights, a normalized contribution degree (i.e., the SHAP value) is obtained. The sum of these SHAP values is equal to the difference between the model prediction result and the baseline value, thereby realizing the precise decomposition and interpretation of the prediction result of a single sample and intuitively showing the influence degree of each feature on the prediction output.

[0078] In some embodiments, when constructing the initial production capacity prediction model and specifically implementing the method, the following contents may further be included:

[0079] S1: According to the sample data, determine the initial weight parameters of the target feature through a preset analytic hierarchy process.

[0080] S2: Based on the XGBOOST algorithm and according to the initial weight parameters, construct an initial production capacity prediction model.

[0081] In some embodiments, when using the sample data to perform a second training on the adjusted first intermediate model to obtain a preset production capacity prediction model that meets the requirements, when specifically implementing the method, the following contents may further be included:

[0082] S1: Use the sample data to perform a second training on the adjusted first intermediate model to obtain a second intermediate model.

[0083] S2: Detect whether the second intermediate model meets the requirements according to the sample data.

[0084] S3: In the case of determining that the second intermediate model does not meet the requirements, determine the second weight parameters of the target feature through SHAP value analysis according to the sample data.

[0085] S4: Adjust the second intermediate model according to the second weight parameters of the target feature to obtain an adjusted second intermediate model.

[0086] S5: Use the sample data to perform a third training on the adjusted second intermediate model to obtain a preset production capacity prediction model that meets the requirements.

[0087] In some embodiments, when determining the first weight parameters of the target feature through SHAP value analysis according to the sample data, when specifically implementing the method, the following contents may further be included:

[0088] By performing SHAP value analysis on the sample data, the marginal contribution degree of each target feature to production capacity prediction is normalized, and based on the normalized marginal contribution degree, the first weight parameter regarding the target feature is determined.

[0089] Among them, the above-mentioned marginal contribution degree can refer to the incremental change in the overall output or result of the model when introducing a target feature, which measures the additional contribution of the target feature to the final result. For example, in a machine learning model, by comparing the differences in the predicted outputs of the model with and without this feature, the marginal impact of this feature on the prediction result under different combinations can be obtained.

[0090] Specifically, normalizing the marginal contribution degree of each target feature to production capacity prediction includes calculating the proportion of the marginal contribution degree of each target feature in the overall contribution degree, and using a standardization method (such as min-max normalization or Z-score normalization) to map it to a unified numerical range to eliminate the influence of feature dimension differences on weight allocation. The normalized marginal contribution degree can more objectively reflect the relative importance of different target features to production capacity prediction, thereby improving the stability and generalization ability of the model.

[0091] After completing the normalization process, the normalized marginal contribution degree is used as the weight value corresponding to each target feature, and based on these weight values, the initial production capacity prediction model is adjusted so that the model can more accurately measure the role of each target feature during the calculation process. In this way, not only the rationality of feature weight allocation is improved, but also the adaptability of the production capacity prediction model to complex geological features is enhanced.

[0092] In some embodiments, when specifically implementing the method of obtaining the target features corresponding to the sample layer, the following content may further be included:

[0093] S1: Obtain the first features corresponding to the sample layer and construct a heat matrix between the first features;

[0094] S2: According to the heat matrix, determine the correlation degree between the first features and the correlation degree between each first feature and the first initial production capacity;

[0095] S3: According to the correlation degree between the first features and the correlation degree between each first feature and the first initial production capacity, perform screening processing on the first features to obtain the target features corresponding to the sample layer, and the target features corresponding to the sample layer include horizon identification, sand addition amount per unit thickness, carrying fluid per unit thickness, net pressure, effective thickness, formation factor, storage coefficient.

[0096] Among them, in the above-mentioned thermal matrix, each matrix element represents the correlation between two first features or between a first feature and a first initial production capacity, and these correlations are usually calculated by statistical methods (such as Pearson or Spearman correlation coefficients). The thermal matrix uses a color gradient to visually represent the magnitude and direction of the correlation, where the darker the color usually indicates a stronger correlation. By analyzing the numerical values and color distributions of the elements in the thermal matrix, the degree of mutual correlation between the first features can be accurately determined, and at the same time, the relationship between each first feature and the first initial production capacity can also be clearly revealed.

[0097] Based on the above embodiments, the correlation determination method based on the thermal matrix not only realizes the scientific nature of data dimensionality reduction and feature screening, but also significantly improves the accuracy and interpretability of the model input. By removing redundant or low-correlated features, the finally selected target features can more effectively reflect the key factors in production capacity prediction.

[0098] In some embodiments, there are multiple target layers, and for determining the production strategy for the target well according to the initial predicted production capacity corresponding to the target layer, when the method is specifically implemented, the following content may further be included:

[0099] Using a preset strategy generation model, according to the initial predicted production capacity corresponding to each target layer and the layer information of the target layer in the target well, determine the production strategy for the target well, and the preset strategy generation model is a model constructed based on a preset machine learning algorithm.

[0100] Based on the above embodiments, by accurately decoupling the initial predicted production capacity of the target layer and the layer information, the quantitative analysis and automated processing of each key influencing factor under complex reservoir conditions are realized, thereby significantly improving the scientific nature and accuracy of the production strategy.

[0101] In some embodiments, for obtaining the target features of the target layer of the target well, when the method is specifically implemented, the following content may further be included:

[0102] S1: Obtain the initial target features corresponding to the target layer;

[0103] S2: Obtain the average value and standard deviation corresponding to the initial target features;

[0104] S3: Obtain the difference between each initial target feature and the average value, and perform a removal process on the initial target features whose difference is greater than a preset multiple of the standard deviation to obtain the target features of the target layer of the target well.

[0105] For example, when screening the target features of a target layer in a target well, first obtain the initial target features of the target layer, such as permeability, porosity, formation pressure, sand addition amount per unit thickness, etc. Then, calculate the average value and standard deviation of these features to evaluate their distribution. Next, if a feature value (such as the permeability of a certain layer) deviates from the corresponding average value by more than a preset multiple (such as 3 times) of the standard deviation, it is considered that the data may be an outlier or have a large deviation, and it is removed to ensure that the data input into the model is more representative and stable.

[0106] Based on the above embodiments, by setting a reasonable outlier removal mechanism, the influence of extreme values on the production capacity prediction model is reduced, and the data quality and the robustness of the model are improved. After removing the abnormal features, the obtained target features can more accurately reflect the reservoir characteristics of the target well, making the subsequent production capacity prediction more accurate.

[0107] As can be seen from the above, a method for predicting the single-layer production capacity of a multi-layer gas reservoir provided by the embodiments of the present specification obtains the target features of the target layer of the target well; wherein, the target features include geological features and fracturing features; using a preset production capacity prediction model, based on the target features of the target layer, obtain the initial predicted production capacity corresponding to the target layer; wherein, the preset production capacity model is obtained by training using sample data in combination with contribution degree analysis and feature weight adjustment; according to the initial predicted production capacity corresponding to the target layer, determine the exploitation strategy for the target layer of the target well. In this way, through the preset production capacity prediction model, by combining contribution degree analysis with feature weight adjustment, the geological features and fracturing features of the target layer are deeply analyzed, realizing the fine decoupling between the initial predicted production capacity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and also providing strong support for formulating a more accurate and scientific exploitation strategy.

[0108] Refer to Figure 2 As shown, the embodiments of the present specification also provide a specific electronic device, wherein the electronic device includes a network communication port 201, a processor 202, and a memory 203, and the above structures are connected by internal cables so that each structure can perform specific data interaction.

[0109] Among them, the network communication port 201 can specifically be used to obtain the target features of the target layer of the target well; wherein, the target features include geological features and fracturing features.

[0110] The processor 202 can specifically be used to obtain the initial predicted production capacity corresponding to the target layer based on the target features of the target layer by using a preset production capacity prediction model; wherein, the preset production capacity model is obtained by training with sample data in advance by combining contribution degree analysis and feature weight adjustment; and a production strategy for the target layer of the target well is determined according to the initial predicted production capacity corresponding to the target layer.

[0111] The memory 203 can specifically be used to store corresponding instruction programs.

[0112] Based on the above method, the relevant structural performance of the electronic device can be effectively utilized, the data processing speed of the electronic device can be improved, and the single-layer production capacity prediction method for multi-layer gas reservoirs can be efficiently realized.

[0113] In this embodiment, the network communication port 201 can be bound to different communication protocols, so as to send or receive different data virtual ports. For example, the network communication port can be a port responsible for web data communication, or a port responsible for FTP data communication, or a port responsible for email data communication. In addition, the network communication port can also be a physical communication interface or communication chip. For example, it can be a wireless mobile network communication chip, such as GSM, CDMA, etc.; it can also be a Wifi chip; it can also be a Bluetooth chip.

[0114] In this embodiment, the processor 202 can be implemented in any suitable manner. For example, the processor can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, etc. This specification does not make a limitation.

[0115] In this embodiment, the memory 203 can include multiple levels. In a digital system, as long as it can store binary data, it can be a memory; in an integrated circuit, a circuit with a storage function without a physical form is also called a memory, such as RAM, FIFO, etc.; in a system, a storage device with a physical form is also called a memory, such as a memory stick, a TF card, etc.

[0116] The embodiments of this specification also provide a computer-readable storage medium based on the above-mentioned single-layer productivity prediction method for multi-layer gas reservoirs. The computer-readable storage medium stores computer program instructions, which, when executed, implement the following: obtaining target features of a target layer of a target well; wherein the target features include geological features and fracturing features; using a preset productivity prediction model, based on the target features of the target layer, obtaining the initial predicted productivity corresponding to the target layer; wherein the preset productivity model is obtained by training with sample data in advance by combining contribution degree analysis and feature weight adjustment; and determining a production strategy for the target layer of the target well according to the initial predicted productivity corresponding to the target layer.

[0117] In this embodiment, the above storage medium includes, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a cache, a hard disk drive (HDD), or a memory card. The memory can be used to store computer program instructions. The network communication unit can be set according to the standards specified by the communication protocol and is used as an interface for network connection communication.

[0118] In this embodiment, the functions and effects specifically implemented by the program instructions stored in the computer-readable storage medium can be explained by comparison with other embodiments and will not be elaborated here.

[0119] Refer to Figure 3 In terms of software, the embodiments of this specification also provide a single-layer productivity prediction device for multi-layer gas reservoirs. The device may specifically include the following structural modules:

[0120] A feature acquisition module 301, configured to obtain target features of a target layer of a target well; wherein the target features include geological features and fracturing features;

[0121] A productivity prediction module 302, configured to use a preset productivity prediction model, based on the target features of the target layer, to obtain the initial predicted productivity corresponding to the target layer; wherein the preset productivity model is obtained by training with sample data in advance by combining contribution degree analysis and feature weight adjustment;

[0122] A strategy determination module 303, configured to determine a production strategy for the target layer of the target well according to the initial predicted productivity corresponding to the target layer.

[0123] In some embodiments, before the above-mentioned production capacity prediction module 302, during specific implementation, the target features of the sample layer and the corresponding first initial production capacity are obtained and utilized to construct sample data; an initial production capacity prediction module, which is used to construct an initial production capacity prediction model according to the XGBOOST algorithm; the sample data is used to perform the first training on the initial production capacity prediction model to obtain a first intermediate model; the first intermediate model is detected according to the sample data to determine whether it meets the requirements; in the case where it is determined that the first intermediate model does not meet the requirements, a weight parameter determination module is used to perform SHAP value analysis according to the sample data to determine the first weight parameter regarding the target features; according to the first weight parameter of the target features, the first intermediate model is adjusted to obtain an adjusted first intermediate model; a preset production capacity prediction module is used to perform the second training on the adjusted first intermediate model by using the sample data to obtain a preset production capacity prediction model that meets the requirements.

[0124] In some embodiments, for the above-mentioned initial production capacity prediction module, during specific implementation, according to the sample data, the initial weight parameter of the target features is determined through a preset analytic hierarchy process; according to the initial weight parameter, an initial production capacity prediction model is constructed based on the XGBOOST algorithm.

[0125] In some embodiments, for the above-mentioned preset production capacity prediction module, during specific implementation, the sample data is used to perform the second training on the adjusted first intermediate model to obtain a second intermediate model; the second intermediate model is detected according to the sample data to determine whether it meets the requirements; in the case where it is determined that the second intermediate model does not meet the requirements, the second weight parameter regarding the target features is determined through SHAP value analysis according to the sample data; according to the second weight parameter of the target features, the second intermediate model is adjusted to obtain an adjusted second intermediate model; the sample data is used to perform the third training on the adjusted second intermediate model to obtain a preset production capacity prediction model that meets the requirements.

[0126] In some embodiments, for the above-mentioned weight parameter determination module, during specific implementation, according to the sample data, SHAP value analysis is performed to normalize the marginal contribution degree of each target feature to production capacity prediction, and according to the normalized marginal contribution degree, the first weight parameter regarding the target features is determined.

[0127] In some embodiments, when the above-mentioned sample layer feature determination module is specifically implemented, it obtains the first features corresponding to the sample layer and constructs a heat matrix between the first features; according to the heat matrix, it determines the correlation between the first features and the correlation between each first feature and the first initial production capacity; according to the correlation between the first features and the correlation between each first feature and the first initial production capacity, it performs screening processing on the first features to obtain the target features corresponding to the sample layer, and the target features corresponding to the sample layer include formation identification, sand addition amount per unit thickness, carrying fluid per unit thickness, net pressure, effective thickness, formation coefficient, and energy storage coefficient.

[0128] In some embodiments, there are multiple target layers. When the above-mentioned strategy determination module 403 is specifically implemented, it uses a preset strategy generation model to determine the exploitation strategy for the target well according to the initial predicted production capacity corresponding to each target layer and the formation information of the target layer in the target well. The preset strategy generation model is a model constructed based on a preset machine learning algorithm.

[0129] In some embodiments, when the above-mentioned feature acquisition module 301 is specifically implemented, it obtains the initial target features corresponding to the target layer; obtains the average value and standard deviation corresponding to the initial target features; obtains the difference between each initial target feature and the average value, and eliminates the initial target features whose difference is greater than a preset multiple of the standard deviation to obtain the target features of the target layer of the target well.

[0130] It should be noted that the units, devices, or modules described in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described by dividing them into various modules according to functions. Of course, when implementing this specification, the functions of each module can be implemented in the same or multiple software and / or hardware, or the modules implementing the same function can be realized by a combination of multiple sub-modules or sub-units, etc. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical, mechanical, or other form.

[0131] As can be seen from the above, based on a single-layer production capacity prediction device for multi-layer gas reservoirs provided by the embodiments of this specification, through a preset production capacity prediction model, the combination of contribution degree analysis and feature weight adjustment is used to deeply analyze the geological features and fracturing features of the target layer, realizing the fine decoupling between the initial predicted production capacity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and also providing strong support for formulating more accurate and scientific exploitation strategies.

[0132] In a specific scenario example, a method, device and equipment for predicting the single-layer production capacity of multi-layer gas reservoirs provided by this specification can be applied. Through a preset production capacity prediction model, the combination of contribution degree analysis and feature weight adjustment is used to deeply analyze the geological features and fracturing features of the target layer, realizing the fine decoupling between the initial predicted production capacity and each key factor, improving the accuracy and interpretability of the single-layer gas production prediction, and also providing strong support for formulating more accurate and scientific exploitation strategies. The specific implementation process may include the following content. Refer to Figure 4 as shown.

[0133] S1: Preprocess the training samples;

[0134] Obtain training samples. Among them, the training samples are the historical production data of each target production layer after the production of multi-layer tight sandstone gas wells has been batch-divided. The training samples are composed of first data and second data. The first data includes two aspects of production parameters, namely the fracturing process (i.e., fracturing features) and geological engineering (i.e., geological features) corresponding to each target production layer of the multi-layer tight sandstone gas wells that have been put into production. Among them, the first data contains categorical data and numerical data. The categorical data is the sample layer code, which needs to be processed in an encoded form. The second data includes the initial production corresponding to each target production layer of the multi-layer tight sandstone gas wells that have been put into production; Combine the first data and the production prediction data output by the generator according to the first data to obtain training samples.

[0135] Perform multiple iterative trainings according to the training samples to obtain the pre-trained initial production predictor. Among them, first preprocess the training samples and perform feature dimensionality reduction.

[0136] In view of the actual situation of the data, a method of outlier removal is adopted. The 3σ principle is used for outlier removal, that is, data with a deviation exceeding 3 times the standard deviation from the average value is removed. Perform Pearson feature dimensionality reduction correlation analysis on the determined influencing factors to obtain the correlation coefficients between each dimension and between each dimension and the initial production, so as to obtain the correlation between each factor. Remove the features with a correlation less than a certain value, thereby realizing feature dimensionality reduction.

[0137] S2: Train the XGBOOST prediction model (i.e., the preset production capacity prediction model) and analyze the main control factors;

[0138] The dataset is divided into a training set and a test set according to a certain ratio. The training set is used to train the XGBoost model, while the test set is used to evaluate the generalization ability of the model. A single-well and single-layer initial production capacity prediction model is established through XGBoost, and the main control factor analysis is carried out using the SHAP (SHAPley Additive exPlanations) interpretation and analysis method (i.e., the preset interpretation and analysis algorithm). After normalizing the weight values calculated by the SHAP algorithm for the features, the weight values are passed to the XGBoost training. According to these weights, the weights of each sample are calculated and applied to the model training process, and higher-weighted training is carried out on important features.

[0139] The single-well and single-layer initial production capacity prediction model based on the combination of the XGBoost model and the SHAP (SHAPley Additive exPlanations) interpretation and analysis method can bring a series of significant beneficial effects, especially having advantages in main control factor analysis and feature weighted training. The specific effects are as follows:

[0140] 1. Enhance the interpretability of the model:

[0141] By assigning a weight to each feature, the SHAP value can clearly reveal the contribution of each feature to the model prediction. Compared with traditional black-box models, SHAP can help us understand the internal decision-making process of the model and clarify the importance of each factor in the prediction.

[0142] 2. Feature weighted training improves the model performance:

[0143] The SHAP value provides the importance weights of the features after normalization, and these weights can be fed back into the training of the XGBoost model, enabling the model to assign higher weights to more important features for training.

[0144] 3. Improve the prediction accuracy and generalization ability:

[0145] Traditional training methods often cannot clearly distinguish the importance of features. However, the XGBoost model weighted by SHAP can focus on the optimization training of key features and avoid the interference of unimportant features. This not only improves the fitting effect on the training set but also enhances the generalization ability on the test set and avoids the overfitting problem.

[0146] 4. Improve the existing technologies and methods:

[0147] Compared with traditional methods based on linear regression or simple machine learning, the method of combining XGBoost and SHAP can capture non-linear relationships and complex feature interactions more accurately, and at the same time improve the prediction ability of the model through feature weighted training.

[0148] 5. Solve the problem of imbalanced data:

[0149] For some production capacity prediction problems, the data may be imbalanced (e.g., the production capacity of some wells is high, while that of most wells is low). Through weighted training, the model can pay more attention to rare and highly predictive samples, and solve the prediction bias problem caused by imbalanced data.

[0150] S3: Predict the target sample;

[0151] Obtain the production parameters of the fracturing process (i.e., fracturing characteristics) and geological engineering (geological characteristics) of a target production layer of a target new well, and input the production parameters of the fracturing process and geological engineering of a target production layer of the target new well into the initially trained production capacity predictor. Obtain the initially predicted production capacity value corresponding to a target production layer of the target new well output by the initially trained production capacity predictor, where the initially trained production capacity predictor is obtained by performing multiple iterative trainings using training samples, and the training samples include the production parameters of the fracturing process and geological engineering of each production layer of the already put into production multi-layer tight sandstone gas wells and the initial production capacity.

[0152] Based on the above embodiments, aiming at the characteristics of multi-layer tight gas reservoirs, considering the differences between different layers, after data preprocessing and dimensionality reduction, the XGBOOST (extreme gradient boosting) algorithm with the advantages of fast running speed, supporting classification and regression, high accuracy, having regularization, preventing overfitting, etc. is used for prediction. On the basis of the prediction results, the SHAP (SHAPley Additive exPlanations) interpretation and analysis method is used to carry out the main control factor analysis. And on the basis of calculating the weights of features by the SHAP algorithm, the weights are passed to the XGBOOST training. A prediction method for the initial production capacity of a single well and a single layer with high accuracy and strong interpretability is obtained, which fully considers the characteristics of multi-layer exploitation of tight gas reservoirs, and the prediction results are more accurate and reliable, and can promote the development of multi-layer exploitation of natural gas towards a more efficient development direction.

[0153] In some embodiments, taking a certain gas reservoir in Ordos as an example, it is a typical gas reservoir with "multi-layers, low pressure, low porosity, low permeability, and strong heterogeneity".

[0154] S1: Preprocessing of training samples, including production capacity splitting, outlier removal, and dimensionality reduction by person feature analysis;

[0155] In this example, the factors affecting the initial gas production of a single well in a single layer after fracturing are divided into two categories: fracturing technology and geological engineering. Geological factors include formation code, effective thickness, total porosity, permeability, gas saturation, shale content, formation coefficient, and energy storage coefficient. Among them, the formation code is categorical data and needs to be processed in the form of encoding. The processing codes are shown in Table 1. Fracturing technology factors include sand addition per unit thickness, sand-carrying fluid per unit thickness, fracture pressure, and net pressure.

[0156] Table 1

[0157]

[0158] There are some problems with the data from the field, such as being incomplete, noisy, and redundant. These problems will cause the data to be unable to be directly analyzed or the analysis effect to be poor. Therefore, it is necessary to preprocess the data. According to the actual situation of the data, the method of outlier removal is adopted. The 3σ principle is used for outlier removal, that is, the data with a deviation from the average value exceeding 3 times the standard deviation is removed. Pearson correlation analysis is performed on the determined influencing factors to obtain the correlation coefficients between each dimension and between each dimension and the initial production, so as to obtain the correlation heat map between the factors as shown in Figure 5 shown. Among them, the sand addition per unit thickness and the sand-carrying fluid volume per unit thickness have a very strong correlation, and the factors that have a greater impact on the initial production capacity of a single well in a single layer are the effective thickness and the energy storage coefficient. The features with a correlation between the feature variables and the dependent variable less than or equal to 0.1 are removed to achieve feature dimensionality reduction.

[0159] After preprocessing, the initial production of the single well in this gas reservoir after batch scoring according to the contribution rate of the producing layer and the factors affecting the production capacity of each layer are shown in Table 2.

[0160] Table 2

[0161]

[0162] S2: Train the XGBOOST prediction model to obtain the prediction model for the initial production of a single well in a single layer of a tight sandstone gas reservoir.

[0163] In this embodiment, the data set is divided into a training set and a test set in a ratio of 7:3 for training and evaluating the model respectively. The features such as formation code, sand addition per unit thickness, sand-carrying fluid per unit thickness, net pressure, effective thickness, formation coefficient, and energy storage coefficient are used as independent variables, and the cumulative gas production in 120 days is used as the dependent variable. An initial production capacity prediction model for a single well in a single layer is established through XGBOOST, and then the SHAP tool is used to explain the XGBOOST machine learning model. The SHAP value is obtained after the model predicts the sample data, indicating the contribution of each feature in each sample. The feature importance diagram of the visualized SHAP values of each feature is shown in Figure 6As shown, the average SHAP values of each feature are obtained and normalized. The feature importance weights are shown in Table 3. The effective thickness has a relatively high impact on the initial production of a single layer in this embodiment.

[0164] Table 3

[0165]

[0166] After normalizing the weight values calculated by the SHAP algorithm for the features, the weight values are passed to the XGBOOST training. When the model is trained, it will determine the importance of each feature during the training process according to the feature weights. Based on these weights, the weights of each sample are calculated and applied to the model training process. For samples with a higher feature contribution degree, higher weights are assigned to them, so that the model pays more attention to these samples during training, thereby improving its learning ability for important features, and thus improving the prediction accuracy and effect as a whole. In this embodiment, the R2 of the training set of this model is 0.96. The comparison line chart and dot line chart of the actual values and predicted values of the training set of the prediction model are as Figure 7 and Figure 8 shown, both indicating that the model has a relatively high prediction accuracy. Among them, Sample Index represents the sample index, Values represents the predicted value, Train: Actual vs Predicted represents the training set: actual value compared with the predicted value, Actual Values represents the actual value, Predicted Values represents the predicted value, and LinePlot represents the line chart.

[0167] S3 Prediction of the target sample: The unknown sample is input into the single-well single-layer initial production capacity prediction model of the tight sandstone gas reservoir established in the above steps to predict the initial production of this sample.

[0168] In this embodiment, the initial production of a single layer of the target sample and the influencing factors are shown in Table 4 below. The production parameters of the fracturing process and geological engineering of a certain target pay zone of the target new well are input into the pre-trained initial production predictor, and the predicted value of the initial production of 120 days corresponding to a certain target pay zone of the target new well output by the initial production predictor is 579,700 cubic meters.

[0169] Table 4

[0170]

[0171]

[0172] Compared with the prior art, the beneficial effects of the present invention include: a single-well and single-layer initial production capacity prediction model based on the combination of the XGBOOST model and the SHAP (SHAPley Additive exPlanations) interpretation and analysis method has advantages in the analysis of main control factors and feature weighted training, enhances the interpretability of the model, improves the model performance through feature weighted training, improves the prediction accuracy and generalization ability, improves the existing technologies and methods, and solves the problem of unbalanced data. Compared with the traditional methods based on linear regression or simple machine learning, the method of combining XGBOOST and SHAP can capture non-linear relationships and complex feature interactions more accurately, and at the same time improves the prediction ability of the model through feature weighted training.

[0173] Although this specification provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative means. The order of steps listed in the embodiments is only one way among the execution orders of numerous steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in the order of the method shown in the embodiments or the drawings or in parallel (for example, in an environment of parallel processors or multi-threaded processing, or even in a distributed data processing environment). The terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, product or device including a series of elements not only includes those elements but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, product or device. Without further limitations, it does not exclude the existence of additional identical or equivalent elements in the process, method, product or device including the said elements. The terms such as "first", "second" are used to denote names and do not denote any specific order.

[0174] Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, the method steps can be logically programmed to enable the controller to implement the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Therefore, such a controller can be regarded as a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and the structures within the hardware component.

[0175] From the description of the above embodiments, those skilled in the art can clearly understand that this specification can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of this specification can essentially be embodied in the form of a software product, and this computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions for causing a computer device (which can be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this specification.

[0176] Although this specification is depicted through embodiments, those of ordinary skill in the art know that this specification has many variations and changes without departing from the spirit of this specification, and it is hoped that the appended claims will cover these variations and changes without departing from the spirit of this specification.

Claims

1. A method for predicting the single-layer productivity of a multi-layer gas reservoir, characterized in that: include: Acquire target features of a target layer of a target well; wherein the target features include geological features and fracturing features; Using a preset capacity prediction model, based on the target features of the target layer, an initial predicted capacity corresponding to the target layer is obtained; wherein the preset capacity model is obtained by pre-training using sample data in combination with contribution analysis and feature weight adjustment; According to the initial predicted production capacity corresponding to the target layer, a production strategy for the target layer of the target well is determined.

2. The method according to claim 1, characterized in that Before obtaining the initial predicted capacity corresponding to the target layer based on the target characteristics of the target layer by using the preset capacity prediction model, the method further includes: Obtain and utilize the target features of the sample layer and the corresponding first initial capacity to construct sample data; Build an initial capacity forecasting model based on the XGBOOST algorithm; Performing a first training on the initial capacity prediction model using the sample data to obtain a first intermediate model; Detecting whether the first intermediate model meets the requirements according to the sample data; When it is determined that the first intermediate model does not meet the requirements, determining a first weight parameter for the target feature by performing a SHAP value analysis according to the sample data; Adjusting the first intermediate model according to a first weight parameter of the target feature to obtain an adjusted first intermediate model; The adjusted first intermediate model is trained for the second time using the sample data to obtain a preset capacity prediction model that meets the requirements.

3. The method according to claim 2, characterized in that When constructing the initial capacity prediction model, it also includes: According to the sample data, determining the initial weight parameters of the target features by using a preset hierarchical analysis method; According to the initial weight parameters and based on the XGBOOST algorithm, an initial capacity prediction model is constructed.

4. The method according to claim 3, characterized in that The second training of the adjusted first intermediate model using the sample data to obtain a preset capacity prediction model that meets the requirements includes: Performing a second training on the adjusted first intermediate model using the sample data to obtain a second intermediate model; Detecting whether the second intermediate model meets the requirements based on the sample data; When it is determined that the second intermediate model does not meet the requirements, determining a second weight parameter for the target feature by performing a SHAP value analysis based on the sample data; According to a second weight parameter of the target feature, adjusting the second intermediate model to obtain an adjusted second intermediate model; The adjusted second intermediate model is trained for the third time using the sample data to obtain a preset capacity prediction model that meets the requirements.

5. The method according to claim 4, characterized in that The determining a first weight parameter about the target feature by performing SHAP value analysis according to the sample data includes: By performing SHAP value analysis on the sample data, the marginal contribution of each target feature to the capacity forecast is normalized, and a first weight parameter for the target feature is determined based on the marginal contribution obtained after normalization.

6. The method according to claim 5, characterized in that The obtaining of the target feature corresponding to the sample layer includes: Acquire the first features corresponding to the sample layer, and construct a thermal matrix between the first features; Determining, according to the thermal matrix, the correlation between the first characteristics and the correlation between each first characteristic and the first initial capacity; According to the correlation between the first features and the correlation between each first feature and the first initial production capacity, the first features are screened and processed to obtain the target features corresponding to the sample layer. The target features corresponding to the sample layer include layer identification, sand addition per unit thickness, sand-carrying fluid per unit thickness, net pressure, effective thickness, formation coefficient, and energy storage coefficient.

7. The method according to claim 1, characterized in that The step of obtaining target characteristics of a target layer of a target well includes: Acquire initial target features corresponding to the target layer; Obtaining the mean value and standard deviation corresponding to the initial target feature; The difference between each of the initial target features and the average value is obtained, and the initial target features whose difference is greater than a preset multiple of the standard deviation are eliminated to obtain the target feature of the target layer of the target well.

8. A device for predicting the productivity of a single layer of a multi-layer gas reservoir, characterized in that: include: A feature acquisition module, used to acquire target features of a target layer of a target well; wherein the target features include geological features and fracturing features; A capacity prediction module is used to obtain the initial predicted capacity corresponding to the target layer based on the target characteristics of the target layer by using a preset capacity prediction model; wherein the preset capacity model is obtained by pre-training with sample data in combination with contribution analysis and feature weight adjustment; The strategy determination module is used to determine the production strategy for the target layer of the target well according to the initial predicted production capacity corresponding to the target layer.

9. An electronic device, characterized in that: It comprises a processor and a memory for storing instructions executable by the processor, and when the processor executes the instructions, the steps of the method for predicting the single-layer productivity of a multi-layer gas reservoir as claimed in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the instructions are executed by a processor, the steps of the method for predicting the single-layer productivity of a multi-layer gas reservoir described in any one of claims 1 to 7 are implemented.

Images