Natural gas hydrate productivity prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses a natural gas hydrate productivity prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining target parameters of a gas well sample before target time, and inputting the target parameters to an input module; matrix processing is carried out on all the target parameters through a data processing module, and then an input matrix is obtained through merging; inputting the input matrix into a self-attention mechanism module, and processing by using a multi-head self-attention mechanism to obtain the predicted productivity of the target time; on the basis of the predicted capacity and the actual capacity of the target time, constructing a prediction error by using the target function to optimize and adjust the capacity prediction model; and performing natural gas hydrate productivity prediction on the target gas well by using the optimized and adjusted productivity prediction model. According to the method, model optimization adjustment is carried out through a self-attention mechanism in combination with time domain parameter data, and the accuracy and reliability of natural gas hydrate productivity prediction can be improved. The method can be widely applied to the technical field of data processing.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, electronic device and storage medium for predicting natural gas hydrate production capacity. Background Art

[0002] Natural gas hydrate productivity prediction is a key technology in natural gas exploration and development, directly impacting production planning and resource assessment of natural gas hydrate fields. Traditional methods for predicting natural gas hydrate productivity rely primarily on physical models and empirical formulas, such as production instability analysis. However, these methods have limitations when dealing with complex subsurface conditions and nonlinear relationships, particularly in environments with multiple coupled factors and dynamic changes, often exhibiting low prediction accuracy and generalization capabilities. With the rapid development of artificial intelligence (AI) technology, particularly the widespread application of deep learning methods, data-driven modeling approaches have gradually been introduced into natural gas hydrate productivity prediction methods. In recent years, deep learning models such as deep convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory networks (LSTMs) have been applied to natural gas hydrate productivity prediction and have achieved considerable success.

[0003] Although traditional deep learning models have been applied to natural gas hydrate production capacity prediction, these models are usually learned based on fixed local information. Therefore, when dealing with complex problems with long-term dependencies and global characteristics, it is difficult to fully capture the global information in the input data, resulting in poor prediction accuracy. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a natural gas hydrate production capacity prediction method, device, electronic device and storage medium, aiming to solve at least one problem of the prior art.

[0005] To achieve the above objectives, one aspect of the embodiments of the present application provides a method for predicting natural gas hydrate production capacity, the method comprising: Obtain target parameters of the gas well sample before the target time, and input the target parameters into the input module; wherein the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; All target parameters are matrixed through the data processing module and then merged to obtain the input matrix; The input matrix is fed into the self-attention mechanism module, and the multi-head self-attention mechanism is used to process the matrix to obtain the predicted production capacity at the target time. Based on the predicted capacity and the actual capacity at the target time, the capacity forecasting model is optimized and adjusted by constructing the forecast error using the objective function. The capacity forecasting model includes an input module, a data processing module, and a self-attention mechanism module. The optimized and adjusted productivity prediction model is used to predict the natural gas hydrate productivity of the target gas well.

[0006] In some embodiments, before the step of inputting the target parameters into the input module, the method further comprises the following steps: Use Kalman filtering method to filter and reduce noise of production dynamic parameters; Among them, production dynamic parameters include daily production, tubing pressure and casing pressure.

[0007] In some embodiments, matrixing all target parameters and then merging them to obtain an input matrix includes the following steps: Matrixing the production dynamic parameters to obtain a first matrix; wherein each column in the first matrix corresponds to the parameter value of each type of parameter item in the production dynamic parameters at multiple time steps before the target time; Matrixing the gas well reservoir static parameters to obtain a second matrix; wherein each column in the second matrix corresponds to the parameter value of each type of parameter item in the gas well reservoir static parameters at multiple time steps before the target time; Matrixing the engineering fracturing parameters to obtain a third matrix; wherein each column in the third matrix corresponds to the parameter value of each type of parameter item in the engineering fracturing parameters at multiple time steps before the target time; Combine the first, second, and third matrices to create the input matrix.

[0008] In some embodiments, combining the first matrix, the second matrix, and the third matrix to obtain an input matrix comprises the following steps: Horizontally concatenate the first matrix, the second matrix, and the third matrix to obtain an input matrix; The number of columns in the input matrix is equal to the total number of parameters in the target parameter, and the number of rows in the input matrix is equal to the total number of time steps.

[0009] In some embodiments, the target parameters include parameter values of various types of parameter items at multiple time steps before the target time. The input matrix is input into the self-attention mechanism module, and the multi-head self-attention mechanism is used to process the matrix to obtain the predicted production capacity at the target time, including the following steps: Input the input matrix into the self-attention mechanism module, and perform the first linear transformation on the input matrix based on the pre-learned weight matrix to obtain the query matrix, key matrix and value matrix; Perform correlation operations on the query matrix and the key matrix to obtain the correlation score between each time step; Use the normalized exponential function to convert the relevance score into attention weight; Based on the attention weight, the value matrix corresponding to each time step is weighted summed to obtain the predicted output; The product of the predicted output and the preset output weight matrix is added to the preset bias to obtain the predicted production capacity at the target time.

[0010] In some embodiments, the gas well sample includes multiple subsamples; based on the predicted production capacity and the actual production capacity at the target time, the production capacity prediction model is optimized and adjusted by using the objective function to construct the prediction error, including the following steps: Build the root mean square error based on the predicted capacity of each subsample and the actual capacity at the target time; The predicted production capacity of all sub-samples is averaged to obtain the predicted average value. The determination coefficient is constructed based on the predicted average value combined with the predicted production capacity of each sub-sample and the actual production capacity at the target time; The root mean square error and determination coefficient are used as objective functions to optimize and adjust the model parameters of the capacity forecasting model.

[0011] In some embodiments, using the optimized and adjusted productivity prediction model to predict the natural gas hydrate productivity of a target gas well includes the following steps: Obtain target parameters of the target gas well before the time to be predicted; The target parameters corresponding to the target gas well are input into the optimized and adjusted production capacity prediction model. Through matrixization and multi-head self-attention mechanism, the target predicted production capacity of the target gas well at the predicted time is obtained.

[0012] To achieve the above objectives, another aspect of the present invention provides a natural gas hydrate production capacity prediction device, comprising: The data acquisition module is used to obtain target parameters of the gas well sample before the target time and input the target parameters into the input module; wherein the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; A first data processing module is used to matrix all target parameters through the data processing module, and then merge them to obtain an input matrix; The second data processing module is used to input the input matrix into the self-attention mechanism module and use the multi-head self-attention mechanism to obtain the predicted production capacity at the target time; The model training module is used to optimize and adjust the capacity forecasting model based on the predicted capacity and the actual capacity at the target time, using the objective function to construct the prediction error. The capacity forecasting model includes an input module, a data processing module, and a self-attention mechanism module. The model application module is used to predict the natural gas hydrate productivity of the target gas well using the optimized and adjusted productivity prediction model.

[0013] In some embodiments, the apparatus further comprises: The filtering module is used to filter and reduce noise of production dynamic parameters using the Kalman filtering method; Among them, production dynamic parameters include daily production, tubing pressure and casing pressure.

[0014] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor implements the above-mentioned method when executing the computer program.

[0015] To achieve the above-mentioned purpose, another aspect of an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned method is implemented.

[0016] To achieve the above object, another aspect of the present invention provides a computer program product, including a computer program, which implements the above method when executed by a processor. Embodiments of the present application include at least the following beneficial effects: The present application provides a natural gas hydrate production capacity prediction method, apparatus, electronic device, storage medium, and program product. This solution obtains target parameters of a gas well sample before a target time and inputs the target parameters into an input module. The target parameters include production dynamic parameters, gas well reservoir static parameters, and engineering fracturing parameters. A data processing module matrixes all target parameters and merges them to obtain an input matrix. The input matrix is input into a self-attention mechanism module and processed using a multi-head self-attention mechanism to obtain a predicted production capacity at the target time. Based on the predicted production capacity and the actual production capacity at the target time, an objective function is used to construct a prediction error to optimize and adjust the production capacity prediction model. The production capacity prediction model includes an input module, a data processing module, and a self-attention mechanism module. The optimized and adjusted production capacity prediction model is used to predict natural gas hydrate production capacity for the target gas well. This application utilizes a self-attention mechanism and combines time-domain parameter data to optimize and adjust the model. This application aims to improve the accuracy and reliability of natural gas hydrate production capacity prediction by fully exploiting nonlinear relationships and long-range dependencies in the data, thereby providing more accurate decision support for oil and gas field development. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of an implementation environment for the natural gas hydrate production capacity prediction method provided in an embodiment of the present application; Figure 2 This is a flow chart of a method for predicting natural gas hydrate production capacity provided in an embodiment of the present application; Figure 3 is a schematic diagram of the expanded process of step S200 provided in an embodiment of the present application; Figure 4 is a schematic diagram of the expanded process of step S300 provided in an embodiment of the present application; Figure 5 This is a schematic diagram of the overall process of the natural gas hydrate production capacity prediction method provided in the embodiment of the present application; Figure 6 This is a schematic diagram of an example of data comparing daily output before and after Kalman filtering processing provided in an embodiment of the present application; Figure 7 Schematic diagram of the principle structure of the production capacity prediction model provided in the embodiment of the present application; Figure 8 Schematic diagram of a comparison example of actual production and model prediction results provided in the embodiments of the present application; Figure 9 This is a schematic structural diagram of a natural gas hydrate production capacity prediction device provided in an embodiment of the present application; Figure 10 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of devices and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0019] It will be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0020] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0022] In the related art, natural gas hydrate production capacity prediction is influenced by numerous factors. The interrelationships and dependencies between these factors are often nonlinear and subject to complex spatiotemporal dynamics. Traditional methods often fail to fully account for these complex factors, resulting in insufficient prediction accuracy and generalization capabilities.

[0023] In view of this, an embodiment of the present application provides a method for predicting natural gas hydrate production capacity. This method obtains target parameters of a gas well sample before a target time and inputs the target parameters into an input module. The target parameters include production dynamic parameters, gas well reservoir static parameters, and engineering fracturing parameters. A data processing module matrixes all target parameters and merges them to obtain an input matrix. The input matrix is input into a self-attention mechanism module and processed using a multi-head self-attention mechanism to obtain the predicted production capacity at the target time. Based on the predicted production capacity and the actual production capacity at the target time, an objective function is used to construct a prediction error to optimize and adjust the production capacity prediction model. The production capacity prediction model includes an input module, a data processing module, and a self-attention mechanism module. The optimized and adjusted production capacity prediction model is used to predict the natural gas hydrate production capacity of the target gas well. This application uses the self-attention mechanism to optimize and adjust the model in combination with time-domain parameter data. The purpose is to improve the accuracy and reliability of natural gas hydrate production capacity prediction by fully exploiting the nonlinear relationships and long-range dependencies in the data, thereby providing more accurate decision support for oil and gas field development.

[0024] It is understandable that the natural gas hydrate production capacity prediction method provided in the present application can be applied to any computer device with data processing and computing capabilities, and this computer device can be various terminals or servers. When the computer device in the embodiment is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal is a smart phone, tablet computer, laptop computer, desktop computer, etc., but is not limited to this.

[0025] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present application. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to the network in a wireless or wired manner to complete data transmission and exchange.

[0026] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0027] In addition, server 101 can also be a node server in a blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0028] The terminal 102 may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited thereto. The terminal 102 and the server 101 may be connected directly or indirectly via wired or wireless communication, which is not limited in this embodiment of the present application.

[0029] For example, based on Figure 1 In the implementation environment shown, an embodiment of the present application provides a natural gas hydrate production capacity prediction method. The following is an example of the natural gas hydrate production capacity prediction method being applied in the server 101. It can be understood that the natural gas hydrate production capacity prediction method can also be applied in the terminal 102.

[0030] Reference Figure 2 , Figure 2 This is an optional flowchart of the natural gas hydrate production capacity prediction method provided in an embodiment of the present application. The execution subject of the natural gas hydrate production capacity prediction method can be any of the aforementioned computer devices (including servers or terminals). Figure 2 The method may include but is not limited to steps S100 to S500.

[0031] Step S100, obtaining target parameters of gas well samples before a target time, and inputting the target parameters into an input module; Among them, the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; For example, in some specific embodiments, the input module includes three parts: 1. gas well production dynamic parameters, including gas well daily production (daily gas production), tubing pressure, and casing pressure; 2. gas well reservoir static parameters, including hydrate saturation, reservoir permeability, reservoir porosity, and reservoir depth; 3. engineering fracturing parameters, including fracturing fluid volume, sand filling volume, fracture length, and fracture width.

[0032] In some embodiments, before the step of inputting the target parameters into the input module, the method may further include the following steps: using a Kalman filter method to filter and reduce noise on production dynamic parameters; wherein the production dynamic parameters include daily production, tubing pressure, and casing pressure.

[0033] For example, in some specific implementations, the Kalman filter method is used to filter the data to reduce the data noise value. The specific steps include two parts: In the prediction step of the Kalman filter, an estimate of the system state is predicted based on the state of the previous step: 1. Prediction steps:

[0034] in, It is the current moment Prediction of status, is the optimal estimate at the previous moment, is the state transition matrix, is the control input matrix, is the control input.

[0035] 2. Update steps: When new observation data Upon arrival, the Kalman filter will update the predicted state estimate to incorporate the new observation information and reduce the error: Calculate the Kalman gain:

[0036] in, is the observation matrix, which represents the relationship from state to observation, is the observation noise covariance matrix, which represents the measurement uncertainty.

[0037] Step S200: Matrixing all target parameters through a data processing module, and then merging them to obtain an input matrix; It should be noted that, in some embodiments, Figure 3As shown, matrixing all target parameters and then merging them to obtain an input matrix can include the following steps: S201, matrixing the production dynamic parameters to obtain a first matrix; wherein, each column in the first matrix corresponds to the parameter value of each type of parameter item in the production dynamic parameters multiple time steps before the target time; S202, matrixing the gas well reservoir static parameters to obtain a second matrix; wherein, each column in the second matrix corresponds to the parameter value of each type of parameter item in the gas well reservoir static parameters multiple time steps before the target time; S203, matrixing the engineering fracturing parameters to obtain a third matrix; wherein, each column in the third matrix corresponds to the parameter value of each type of parameter item in the engineering fracturing parameters multiple time steps before the target time; S204, merging the first matrix, the second matrix and the third matrix to obtain an input matrix.

[0038] For example, in some specific implementations, the data processing module takes over the input module and matrixes the data to facilitate model calculation.

[0039] Furthermore, the dynamic parameters are matrixed and expressed as a The matrix is as follows:

[0040] in is the number of time steps, 3 represents the three dynamic parameters of each time step, represents daily output (104 / d), Indicates the oil pipe pressure (MPa), Indicates casing pressure (MPa).

[0041] Furthermore, the static parameters of the gas well reservoir are matrixed and expressed as a The matrix is as follows:

[0042] in represents the hydrate saturation (%), represents the reservoir permeability (mD), represents the reservoir porosity (%), represents the reservoir depth (m).

[0043] Furthermore, the gas well engineering parameters are matrixed and expressed as a The matrix is as follows:

[0044] in, Indicates the volume of fracturing fluid (m3), Indicates the amount of sand filling (m3), represents the crack length, Indicates the crack width.

[0045] Furthermore, the three parameters are combined into one to form an overall input matrix .

[0046] In some embodiments, step S204 may include the following steps: horizontally concatenating the first matrix, the second matrix, and the third matrix to obtain an input matrix; wherein the number of columns of the input matrix is equal to the total number of all parameter items in the target parameters, and the number of rows of the input matrix is equal to the total number of time steps.

[0047] For example, in some embodiments, the matrices of the three parameters are horizontally spliced and merged together to form an overall input matrix , forming a The matrix is as follows:

[0048] Step S300: Input the input matrix into the self-attention mechanism module and use the multi-head self-attention mechanism to obtain the predicted production capacity at the target time; It should be noted that the target parameters include parameter values of various types of parameter items at multiple time steps before the target time; in some embodiments, such as Figure 4 As shown, step S300 may include the following steps: S301, inputting the input matrix into the self-attention mechanism module, performing a first linear change processing on the input matrix based on the pre-learned weight matrix to obtain a query matrix, a key matrix and a value matrix; S302, performing a correlation operation on the query matrix and the key matrix to obtain a correlation score between each time step; S303, using a normalized exponential function to convert the correlation score into an attention weight; S304, based on the attention weight, performing a weighted summation on the value matrix corresponding to each time step to obtain a predicted output; S305, adding the product of the predicted output and the preset output weight matrix to the preset bias to obtain the predicted production capacity at the target time.

[0049] For example, in some specific implementations, a self-attention mechanism module, the core idea of which is to calculate the correlation between each input and assign different weights to different inputs, thereby effectively capturing the long-term dependency between input features.

[0050] Furthermore, the input matrix , we get the query ( ),key( ) and value( )matrix.

[0051] Furthermore, let the dimensions of these matrices be ,in The dimension of each vector is calculated by matrix transformation as follows:

[0052] in, 、 、 Is the query ( ),key( ) and value( )matrix, 、 、 is the learned weight matrix.

[0053] Further, calculate, and The correlation score between them is used to measure the correlation importance between each time step.

[0054]

[0055] Through the Softmax function, the score is converted into a weight:

[0056] Furthermore, the attention weight obtained right Perform weighted summation to obtain the weighted output:

[0057] Furthermore, in order to capture different patterns in the data, a multi-head self-attention mechanism is used. By computing multiple attention heads in parallel, each head focuses on different feature representations. Finally, the outputs of multiple heads are concatenated and the final output is obtained through a linear transformation.

[0058] Furthermore, the final gas well productivity prediction value is output, and the calculation formula is:

[0059] in, Predicting production capacity for gas wells, is the weight matrix of the output layer, is the bias.

[0060] Step S400 , based on the predicted capacity and the actual capacity at the target time, the capacity forecast model is optimized and adjusted by constructing the forecast error using the objective function; Among them, the capacity prediction model includes input module, data processing module and self-attention mechanism module; It should be noted that the gas well sample includes multiple sub-samples; in some embodiments, step S400 may include the following steps: constructing a root mean square error based on the predicted production capacity of each sub-sample and the actual production capacity at the target time; averaging the predicted production capacities of all sub-samples to obtain a predicted average value, and constructing a determination coefficient based on the predicted average value combined with the predicted production capacity of each sub-sample and the actual production capacity at the target time; and optimizing and adjusting the model parameters of the production capacity prediction model using the root mean square error and the determination coefficient as objective functions.

[0061] For example, in some embodiments, the root mean square error ( ) and the coefficient of determination ( ) is the objective function, and its calculation formula is: The root mean square error formula is:

[0062] in, is the sample size; is the actual value; is the predicted value.

[0063] The coefficient of determination formula is:

[0064] in, is the sample size; is the actual value; is the predicted value; is the average value.

[0065] In some specific application scenarios, the samples can also be pre-divided into training sets and test sets. First, the relevant processing procedures such as steps S100 to S400 are executed according to the training set to optimize and adjust the training stage of the model; then, the relevant processing procedures such as steps S100 to S400 can be executed again according to the test set. When executing the relevant processing procedures of step S400, only the prediction error obtained by the objective function is used to verify the model prediction accuracy without making any adjustments. If the accuracy meets the requirements, it is directly applied. Otherwise, the relevant processing procedures such as steps S100 to S400 are cyclically performed based on the training set to perform iterative training until the accuracy verification of the test set meets the requirements.

[0066] Step S500, using the optimized and adjusted productivity prediction model to predict the natural gas hydrate productivity of the target gas well; It should be noted that, in some embodiments, step S500 may include the following steps: obtaining the target parameters of the target gas well before the time to be predicted; inputting the target parameters corresponding to the target gas well into the optimized and adjusted production capacity prediction model, and obtaining the target predicted production capacity of the target gas well at the time to be predicted through matrixing and multi-head self-attention mechanism.

[0067] For example, in some specific implementations, during the actual application of the production capacity forecasting model, before the target parameters are input into the model, the Kalman filtering method can also be used to suppress noise and filter data on the production dynamic data in the target parameters to ensure the accuracy and stability of the input data, thereby providing high-quality input data for subsequent model predictions.

[0068] In order to explain the principle of the technical solution of the present invention in detail, the overall process of the present invention is described below in combination with some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and cannot be regarded as a limitation of the present invention.

[0069] First, it's important to note that natural gas hydrate production capacity prediction involves numerous factors. The interrelationships and dependencies between these factors are often nonlinear and subject to complex spatiotemporal dynamics. Traditional methods often fail to fully account for these complex factors, resulting in limited prediction accuracy and generalization capabilities. Therefore, effectively mining the complex dependencies within natural gas hydrate production data, particularly long-term dependencies and capturing global information, remains a major challenge in current technologies.

[0070] In view of this, the present invention optimizes and adjusts the model through the self-attention mechanism combined with time domain parameter data, aiming to improve the accuracy and reliability of natural gas hydrate production capacity prediction by fully exploring the nonlinear relationships and long-distance dependencies in the data, thereby providing more accurate decision-making support for oil and gas field development.

[0071] In some specific application scenarios, such as Figure 5 As shown, the natural gas hydrate production capacity prediction method provided by the present invention can be implemented through the following process steps: (1) Data preprocessing: Get the dynamic production data of a gas well, including the gas well tubing pressure, casing pressure and other data.

[0072] Specifically, the Kalman filter method is used to filter the data to reduce the data noise value. The specific steps include two parts: In the prediction step of the Kalman filter, an estimate of the system state is predicted based on the state of the previous step: 1. Prediction steps:

[0073] in, It is the current moment Prediction of status, is the optimal estimate at the previous moment, is the state transition matrix, is the control input matrix, is the control input.

[0074] 2. Update steps: When new observation data Upon arrival, the Kalman filter will update the predicted state estimate to incorporate the new observation information and reduce the error: Calculate the Kalman gain:

[0075] in, is the observation matrix, which represents the relationship from state to observation, is the observation noise covariance matrix, which represents the uncertainty of measurement. The dynamic effects of gas well production before and after Kalman filter processing are as follows: Figure 6 shown.

[0076] (2) Design of gas well productivity prediction model structure (e.g. Figure 7 The following is only an example. If the data processing flow is clear, the specific structure details are not limited): To enhance the physical constraint capability of the model, the present invention designs three modules: The first module: input module, which consists of three parts: 1. Gas well production dynamic parameters (dynamic data) after Kalman filtering, including gas well daily production, tubing pressure, and casing pressure; 2. Gas well reservoir static parameters (static data), including hydrate saturation, reservoir permeability, reservoir porosity, and reservoir depth; 3. Engineering fracturing parameters (fracturing data), including fracturing fluid volume, sand filling volume, fracture length, and fracture width.

[0077] The second module: data processing module, which takes over the input module and matrixes the data to facilitate model calculation.

[0078] Furthermore, the dynamic parameters are matrixed and expressed as a The matrix is as follows:

[0079] in is the number of time steps, 3 represents the three dynamic parameters of each time step, represents daily output (104 / d), Indicates the oil pipe pressure (MPa), Indicates casing pressure (MPa).

[0080] Furthermore, the static parameters of the gas well reservoir are matrixed and expressed as a The matrix is as follows:

[0081] in represents the hydrate saturation (%), represents the reservoir permeability (mD), represents the reservoir porosity (%), represents the reservoir depth (m).

[0082] Furthermore, the gas well engineering parameters are matrixed and expressed as a The matrix is as follows:

[0083] in, Indicates the volume of fracturing fluid (m3), Indicates the amount of sand filling (m3), represents the crack length, Indicates the crack width.

[0084] Furthermore, the three parameters are combined into one to form an overall input matrix , forming a The matrix is as follows:

[0085] The third module is the self-attention mechanism module. The core idea of this module is to calculate the correlation between each input and assign different weights to different inputs, thereby effectively capturing the long-term dependency between input features.

[0086] Furthermore, the input matrix , we get the query ( ),key( ) and value( )matrix.

[0087] Furthermore, let the dimensions of these matrices be ,in The dimension of each vector is calculated by matrix transformation as follows:

[0088] in, 、 、 Is the query ( ),key( ) and value( )matrix, 、 、 is the learned weight matrix.

[0089] Further, calculate, and The correlation score between them is used to measure the correlation importance between each time step.

[0090]

[0091] Through the Softmax function, the score is converted into a weight:

[0092] Furthermore, the attention weight obtained right Perform weighted summation to obtain the weighted output:

[0093] Furthermore, in order to capture different patterns in the data, a multi-head self-attention mechanism is used. By computing multiple attention heads in parallel, each head focuses on different feature representations. Finally, the outputs of multiple heads are concatenated and the final output is obtained through a linear transformation.

[0094] Furthermore, the final gas well productivity prediction value is output, and the calculation formula is:

[0095] in, Predicting production capacity for gas wells, is the weight matrix of the output layer, is the bias.

[0096] (3) Model training and validation (testing): The specific steps of model training are as follows: First, the data set is divided into training set and test set, such as Figure 8 This example compares actual production with the model prediction results achieved using different models (LSTM and SAM) using the attention mechanism and training optimization process of this application. Specifically, the production rate can be adjusted based on the actual production dynamic time of the gas well, with a ratio between 0.5 and 0.8.

[0097] Furthermore, the root mean square error ( ) and the coefficient of determination ( ) is the objective function, and its calculation formula is: The root mean square error formula is:

[0098] in, is the sample size; is the actual value; is the predicted value.

[0099] The coefficient of determination formula is:

[0100] in, is the sample size; is the actual value; is the predicted value; is the average value.

[0101] In summary, the present invention proposes a natural gas hydrate production capacity prediction method based on the self-attention mechanism, which uses the self-attention mechanism to capture the complex dependencies that affect production capacity. Specifically, by constructing a deep learning model based on the Transformer architecture, the long-term dependencies and global features in the gas well production data are fully explored, and the relationship weights between the input features are automatically learned and adjusted. This method can not only effectively deal with multi-factor coupling and nonlinear problems, but also provide higher prediction accuracy with less historical data support. Through experimental verification, the method proposed in the present invention can significantly improve the accuracy of natural gas hydrate production capacity prediction, and provide more reliable technical support for production planning, resource assessment and optimization decision-making of oil and gas fields.

[0102] Specifically, the core principles implemented by the technical solution of the present invention include: 1. Natural Gas Hydrate Productivity Prediction Method Based on Self-Attention Mechanism: The core technology of this invention lies in the use of the self-attention mechanism to process gas well production dynamics data. By assigning different weights to input features, the model can dynamically adjust and capture long-term dependencies and global information between data, thereby improving prediction accuracy and generalization capabilities.

[0103] 2. Kalman filter preprocessing step: The present invention adopts the Kalman filter method to suppress noise and filter data of gas well production dynamic data to ensure the accuracy and stability of input data and provide high-quality input data for subsequent model training.

[0104] 3. Modular Design of the Gas Well Productivity Prediction Model: This invention incorporates three modules: an input module, a data processing module, and a self-attention mechanism module. Each module independently and efficiently processes different types of input data (such as dynamic parameters, static parameters, and engineering fracturing parameters). The self-attention mechanism is then used to perform comprehensive modeling, improving overall prediction accuracy.

[0105] 4. Application of a Multi-Head Self-Attention Mechanism: To further enhance model performance, this paper employs a multi-head self-attention mechanism, enabling the model to weight input data from multiple perspectives, capturing information about the expression of different features. Each "attention head" focuses on a different feature representation, effectively enhancing the model's ability to handle complex problems.

[0106] 5. Model Training and Validation Methods: This paper incorporates a well-defined training process, including dataset partitioning, objective function selection (RMS error and coefficient of determination), and implementation of training and validation methods. This rigorous validation and evaluation of the model training process ensures the accuracy and reliability of the model in practical applications.

[0107] Compared with the prior art, the present invention has at least the following beneficial effects: First, traditional methods rely heavily on physical models and empirical formulas, failing to effectively handle the nonlinear relationships and long-term dependencies between multiple factors. This leads to low prediction accuracy, especially under complex underground conditions. However, this new method introduces a self-attention mechanism that dynamically adjusts the weights between input features, effectively capturing the long-term dependencies between features and improving the model's expressiveness and prediction accuracy.

[0108] Secondly, the present invention uses Kalman filtering to remove noise from production dynamics data, ensuring data accuracy and stability, further improving the reliability of model training. By combining Kalman filtering with a self-attention mechanism, it achieves more robust processing of noisy data and fully utilizes information such as gas well production dynamics, reservoir static parameters, and engineering fracturing parameters to achieve effective fusion of multi-dimensional data.

[0109] Furthermore, the use of a multi-head self-attention mechanism enables the model to focus on different features of the input data from different perspectives, further enhancing its ability to handle complex nonlinear relationships. The model's weighted output, calculated through linear transformation, allows each feature to be dynamically adjusted based on its importance, improving the accuracy and generalization of gas well productivity predictions.

[0110] Finally, by optimizing the objective function (root mean square error and coefficient of determination) during the training process, the present invention can not only predict the gas well productivity with high precision, but also effectively measure the model fitting effect and prediction accuracy, ensuring the effectiveness of its application under different production conditions.

[0111] In summary, the present invention innovatively improves the accuracy, robustness, and generalization ability of natural gas hydrate production capacity prediction by combining Kalman filtering and self-attention mechanism, solves the problem of difficulty in handling multi-factor interactions and long-term dependencies in existing technologies, and has important practical application value and economic benefits.

[0112] like Figure 9 As shown, the embodiment of the present application further provides a natural gas hydrate production capacity prediction device 900, which can implement the above method, and the device includes: The data acquisition module 901 is used to obtain target parameters of the gas well sample before the target time and input the target parameters into the input module; wherein the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; A first data processing module 902 is used to matrix all target parameters through the data processing module and then merge them to obtain an input matrix; The second data processing module 903 is used to input the input matrix into the self-attention mechanism module and use the multi-head self-attention mechanism to obtain the predicted production capacity at the target time; Model training module 904 is used to optimize and adjust the capacity prediction model based on the predicted capacity and the actual capacity at the target time, using the objective function to construct the prediction error. The capacity prediction model includes an input module, a data processing module, and a self-attention mechanism module. The model application module 905 is used to use the optimized and adjusted productivity prediction model to predict the natural gas hydrate productivity of the target gas well.

[0113] In some embodiments, the apparatus may further include: The filtering module is used to filter and reduce noise of production dynamic parameters using the Kalman filtering method; Among them, production dynamic parameters include daily production, tubing pressure and casing pressure.

[0114] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0115] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the above method when executing the computer program. The electronic device can be any smart terminal including a tablet computer, an in-vehicle computer, or the like.

[0116] It can be understood that the contents of the above method embodiments are applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0117] like Figure 10 As shown, Figure 10 The hardware structure of an electronic device 1000 according to another embodiment is shown. The electronic device 1000 includes: The processor 1001 may be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided by the embodiments of the present invention. The memory 1002 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called by the processor 1001 to execute the network node population optimization method of the embodiment of the present invention. Input / output interface 1003, used to implement information input and output; Communication interface 1004, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.); Bus 1005 , which transmits information between various components of the device (e.g., processor 1001 , memory 1002 , input / output interface 1003 , and communication interface 1004 ); The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0118] The electronic device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one location or distributed across multiple network units. Some or all of these modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0119] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above method is implemented.

[0120] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0121] An embodiment of the present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0122] It is understandable that the contents of the above method embodiments are all applicable to the present program product embodiments, the functions specifically implemented by the present program product embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0123] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0124] The embodiments of the present application provide a natural gas hydrate production capacity prediction method, device, electronic device, storage medium, and program product. The method obtains target parameters of a gas well sample before a target time and inputs the target parameters into an input module. The target parameters include production dynamic parameters, gas well reservoir static parameters, and engineering fracturing parameters. A data processing module matrixes all target parameters and merges them to obtain an input matrix. The input matrix is input into a self-attention mechanism module and processed using a multi-head self-attention mechanism to obtain a predicted production capacity at the target time. Based on the predicted production capacity and the actual production capacity at the target time, a prediction error is constructed using an objective function to optimize and adjust the production capacity prediction model. The production capacity prediction model includes an input module, a data processing module, and a self-attention mechanism module. The optimized and adjusted production capacity prediction model is used to predict the natural gas hydrate production capacity of the target gas well. This application utilizes a self-attention mechanism to optimize and adjust the model in combination with time-domain parameter data. This application aims to improve the accuracy and reliability of natural gas hydrate production capacity prediction by fully exploiting nonlinear relationships and long-range dependencies in the data, thereby providing more accurate decision support for oil and gas field development.

[0125] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0126] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0127] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0128] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0129] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0130] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0131] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as 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 mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0132] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0133] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0134] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0135] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for predicting natural gas hydrate production capacity, characterized in that: The method comprises the following steps: Obtain target parameters of the gas well sample before the target time, and input the target parameters into an input module; wherein the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; Matrixing all the target parameters through a data processing module, and then merging them to obtain an input matrix; The matrixing of all the target parameters and then merging them to obtain an input matrix comprises the following steps: Matrixing the production dynamic parameters to obtain a first matrix; wherein each column in the first matrix corresponds to a parameter value of each type of parameter item in the production dynamic parameters at multiple time steps before the target time; Matrixing the gas well reservoir static parameters to obtain a second matrix; wherein each column in the second matrix corresponds to a parameter value of each type of parameter item in the gas well reservoir static parameters at multiple time steps before the target time; Matrixing the engineering fracturing parameters to obtain a third matrix; wherein each column in the third matrix corresponds to a parameter value of each type of parameter item in the engineering fracturing parameters at multiple time steps before the target time; Combining the first matrix, the second matrix and the third matrix to obtain the input matrix; Input the input matrix into the self-attention mechanism module, and use the multi-head self-attention mechanism to obtain the predicted production capacity at the target time; Based on the predicted capacity and the actual capacity at the target time, the capacity prediction model is optimized and adjusted by constructing a prediction error using an objective function; wherein the capacity prediction model includes the input module, the data processing module, and the self-attention mechanism module; The optimized and adjusted productivity prediction model is used to predict the natural gas hydrate productivity of the target gas well.

2. The method according to claim 1, characterized in that Before the step of inputting the target parameters into the input module, the method further comprises the following steps: Using the Kalman filter method to filter and reduce noise of the production dynamic parameters; The production dynamic parameters include daily production, tubing pressure and casing pressure.

3. The method according to claim 1, characterized in that The step of merging the first matrix, the second matrix, and the third matrix to obtain the input matrix comprises the following steps: horizontally concatenating the first matrix, the second matrix, and the third matrix to obtain the input matrix; The number of columns of the input matrix is equal to the total number of all parameter items in the target parameters, and the number of rows of the input matrix is equal to the total number of time steps.

4. The method according to claim 1, wherein The target parameters include parameter values of various types of parameter items at multiple time steps before the target time; the input matrix is input into the self-attention mechanism module, and the multi-head self-attention mechanism is used to process the matrix to obtain the predicted production capacity at the target time, including the following steps: Inputting the input matrix into a self-attention mechanism module, performing a first linear transformation on the input matrix based on a pre-learned weight matrix to obtain a query matrix, a key matrix, and a value matrix; performing a correlation operation on the query matrix and the key matrix to obtain a correlation score between each of the time steps; Converting the relevance score into an attention weight using a normalized exponential function; Based on the attention weight, performing weighted summation on the value matrix corresponding to each time step to obtain a predicted output; The product of the predicted output and the preset output weight matrix is added to the preset bias to obtain the predicted production capacity at the target time.

5. The method according to claim 1, wherein The gas well sample includes a plurality of subsamples; and based on the predicted production capacity and the actual production capacity at the target time, optimizing and adjusting the production capacity prediction model by constructing a prediction error using an objective function includes the following steps: constructing a root mean square error based on the predicted capacity of each subsample and the actual capacity at the target time; Performing an average calculation on the predicted production capacities of all the subsamples to obtain a predicted average value, and constructing a determination coefficient based on the predicted average value in combination with the predicted production capacity of each subsample and the actual production capacity at the target time; The root mean square error and the determination coefficient are used as the objective function to optimize and adjust the model parameters of the production capacity prediction model.

6. The method according to any one of claims 1 to 5, characterized in that The method of using the optimized and adjusted productivity prediction model to predict the natural gas hydrate productivity of the target gas well includes the following steps: Obtaining the target parameters of the target gas well before the time to be predicted; The target parameters corresponding to the target gas well are input into the optimized and adjusted production capacity prediction model, and the target predicted production capacity of the target gas well at the time to be predicted is obtained through the matrixization and the multi-head self-attention mechanism.

7. A natural gas hydrate production capacity prediction device, characterized in that: The device comprises: A data acquisition module is used to obtain target parameters of gas well samples before a target time and input the target parameters into an input module; wherein the target parameters include production dynamic parameters, gas well reservoir static parameters and engineering fracturing parameters; A first data processing module, configured to matrix all the target parameters through a data processing module, and then combine them to obtain an input matrix; The matrixing of all the target parameters and then merging them to obtain an input matrix comprises the following steps: Matrixing the production dynamic parameters to obtain a first matrix; wherein each column in the first matrix corresponds to a parameter value of each type of parameter item in the production dynamic parameters at multiple time steps before the target time; Matrixing the gas well reservoir static parameters to obtain a second matrix; wherein each column in the second matrix corresponds to a parameter value of each type of parameter item in the gas well reservoir static parameters at multiple time steps before the target time; Matrixing the engineering fracturing parameters to obtain a third matrix; wherein each column in the third matrix corresponds to a parameter value of each type of parameter item in the engineering fracturing parameters at multiple time steps before the target time; Combining the first matrix, the second matrix and the third matrix to obtain the input matrix; A second data processing module is configured to input the input matrix into a self-attention mechanism module, and use a multi-head self-attention mechanism to obtain the predicted production capacity at the target time; a model training module, configured to optimize and adjust the capacity prediction model by constructing a prediction error using an objective function based on the predicted capacity and the actual capacity at the target time; wherein the capacity prediction model includes the input module, the data processing module, and the self-attention mechanism module; The model application module is used to use the optimized and adjusted productivity prediction model to predict the natural gas hydrate productivity of the target gas well.

8. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images