Gas well bubble drainage system optimization method, device, equipment and medium based on machine learning

Abstract

The present invention discloses a method, device, equipment, and medium for optimizing a gas well bubble drainage system based on machine learning, including: screening based on the correlation between the dynamic and static characteristics of the bubble drainage well and the daily gas production to obtain a correlation data set; dividing the data into a training set and a test set; outputting a bubble drainage well daily gas production prediction model when preset training requirements are met; determining the bubble drainage agent daily injection amount range, injection cycle range, and dilution ratio range of the bubble drainage well; inputting optimization parameters into the bubble drainage well daily gas production prediction model to obtain the predicted daily gas production corresponding to the optimization parameters; with the goal of maximizing the predicted daily gas production, screening several groups of optimization parameters to obtain the optimal optimization parameters. The present invention belongs to the field of gas well bubble drainage system optimization. The present invention can improve the management efficiency of bubble drainage wells and improve the implementation efficiency of bubble drainage processes.

Description

Technical Field

[0001] The present invention relates to the field of gas well bubble drainage system optimization, and in particular to a gas well bubble drainage system optimization method, device, equipment and medium based on machine learning. Background Art

[0002] Foam drainage (foam drainage) is a process that injects a surfactant (foaming agent) into the wellbore, agitating the natural gas flow to generate a low-density aqueous foam. This foam is then carried from the wellbore bottom to the surface by the airflow, thereby removing the accumulated liquid. The most important parameters in the foam drainage system are the daily injection rate of the foaming agent, the injection cycle, and the dilution ratio.

[0003] Conventional bubble drainage optimization schemes currently rely primarily on recommended bubble drainage concentrations, combined with calculations of gas well water production and downhole liquid accumulation, to determine the bubble drainage injection schedule. This requires significant analytical effort, and calculations and adjustments rely on historical experience. If the corresponding system is not properly controlled, the inability to effectively remove the downhole liquid accumulation will severely impact its removal and restrict gas well productivity. Therefore, the present invention provides a machine learning-based gas well bubble drainage system optimization method to address this issue. Summary of the Invention

[0004] The present invention solves the technical problem in the prior art that the gas well bubble drainage and injection system is heavily dependent on work experience by providing a method, device, equipment and medium for optimizing the gas well bubble drainage and injection system based on machine learning, and achieves the technical effect of intelligently determining the gas well bubble drainage and injection system.

[0005] In a first aspect, the present invention provides a method for optimizing a gas well bubble discharge system based on machine learning, the method comprising:

[0006] Based on the correlation between the dynamic and static characteristics of the bubble gas wells and the daily gas production, the production parameters, gas reservoir parameters and basic parameters of the bubble gas wells were screened to obtain a correlation data set.

[0007] Based on the associated data set, divide the training set and test set;

[0008] The daily gas production prediction model for the bubbling exhaust well is trained based on the training set, and is verified using the test set. When the preset training requirements are met, the daily gas production prediction model for the bubbling exhaust well is output;

[0009] Determine the daily injection volume range, injection cycle range, and dilution ratio range of the foaming agent for the foaming exhaust well;

[0010] Based on the range of daily injection volume, injection cycle, and dilution ratio of the foaming agent in the gas well, several sets of optimization parameters are constructed. These optimization parameters are input into the daily gas production prediction model of the gas well to obtain the predicted daily gas production corresponding to the optimized parameters. One set of optimization parameters includes a daily injection volume of the foaming agent, an injection cycle, and a dilution ratio.

[0011] With the goal of maximizing the predicted daily gas production, several groups of optimization parameters are screened to obtain the optimal optimization parameters.

[0012] Furthermore, based on the correlation between the dynamic and static characteristics of the bubbling and exhaust gas wells and the daily gas production, the production parameters, gas reservoir parameters, and basic well parameters of the bubbling and exhaust gas wells were screened to obtain a correlation data set, including:

[0013] The basic data set is constructed based on the production parameters of the bubble and exhaust wells, the gas reservoir parameters of the bubble and exhaust wells, and the basic parameters of the wells;

[0014] Based on the Pearson correlation analysis method, the parameters in the basic data set were screened to obtain the parameters related to the daily gas production of the bubbling and exhaust gas wells;

[0015] A correlation data set is constructed based on parameters related to the daily gas production of the bubble exhaust well.

[0016] Furthermore, based on the associated dataset, the training set and test set are divided into:

[0017] According to the data information dimension, the parameters in the associated data set are spliced ​​to obtain the data wide table at each dimension level;

[0018] The wide data table was sorted based on the correlation between the wide data table and the daily gas production of the bubble gas wells;

[0019] Based on the sorted data wide table, divide the training set and test set.

[0020] Furthermore, the daily gas production prediction model for the bubbling exhaust well is trained based on the training set, and is verified using the test set. When the preset training requirements are met, the daily gas production prediction model for the bubbling exhaust well is output, including:

[0021] Based on the training set, a daily gas production prediction model for bubble exhaust wells is constructed. The daily gas production prediction model for bubble exhaust wells predicts the gas production of bubble exhaust wells according to the time series.

[0022] DILATE is used as the loss function of the daily gas production prediction model for bubble exhaust wells, and Adam is used as the optimizer of the daily gas production prediction model for bubble exhaust wells.

[0023] When DILATE meets the preset application requirements, the test set is input into the daily gas production prediction model for the bubbling and exhaust wells. When the preset training requirements are met, the daily gas production prediction model for the bubbling and exhaust wells is output.

[0024] Furthermore, the daily injection volume range, injection cycle range, and dilution ratio range of the foaming agent for the gas well are determined, including:

[0025] Determine the daily injection amount range of the foaming agent for the air-exhaust well based on the effective concentration range of the foaming agent and the daily water production of the air-exhaust well.

[0026] The filling cycle range and dilution ratio range are determined based on the performance of the filling equipment used in the bubble exhaust well.

[0027] Furthermore, with the goal of maximizing the predicted daily gas production, several groups of optimization parameters were screened to obtain the optimal optimization parameters, including:

[0028] With the goal of maximizing the predicted daily gas production, the optimization parameters corresponding to the maximum predicted daily gas production are screened out from several groups of optimization parameters. If there is only one group of optimization parameters corresponding to the maximum predicted daily gas production, then this group of optimization parameters is taken as the optimal optimization parameters.

[0029] If there are two or more groups of optimization parameters corresponding to the maximum predicted daily gas production, the optimization parameter with the lowest cost is selected from the optimization parameters corresponding to the maximum predicted daily gas production with the lowest cost as the goal;

[0030] If there is only one set of optimization parameters with the lowest cost, then this set of optimization parameters is taken as the optimal optimization parameters;

[0031] If there are two or more groups of optimization parameters with the lowest cost, the best optimization parameters are selected from several optimization parameters with the lowest cost, with the goal of minimizing additional on-site workload.

[0032] Furthermore, the method further comprises:

[0033] Determine the deviation value based on the optimal optimization parameters of the bubble exhaust well;

[0034] When the deviation value is greater than a preset threshold, the daily gas production prediction model of the bubble exhaust well is updated.

[0035] In a second aspect, the present invention provides a device for optimizing the bubble discharge system of a gas well based on machine learning, the device comprising:

[0036] The basic data screening module is used to screen the production parameters, gas reservoir parameters and basic parameters of the bubbling and exhaust gas wells based on the correlation between the dynamic and static characteristics of the bubbling and exhaust gas wells and the daily gas production, and obtain the associated data set;

[0037] The partitioning module is used to divide the training set and the test set based on the associated data set;

[0038] A model training module is used to train the daily gas production prediction model for the bubbling exhaust well based on the training set and verify the daily gas production prediction model for the bubbling exhaust well using the test set. When the preset training requirements are met, the daily gas production prediction model for the bubbling exhaust well is output;

[0039] A range determination module is used to determine the daily injection amount range, injection cycle range and dilution ratio range of the foaming agent in the foaming exhaust well;

[0040] A prediction module is used to construct several sets of optimization parameters based on the range of daily injection volume, injection cycle, and dilution ratio of the foaming and exhausting wells, and input the optimized parameters into the foaming and exhausting well daily gas production prediction model to obtain the predicted daily gas production corresponding to the optimized parameters, wherein a set of optimization parameters includes a daily injection volume of the foaming and exhausting agent, a injection cycle, and a dilution ratio;

[0041] The optimal parameter screening module is used to screen several groups of optimization parameters with the goal of maximizing the predicted daily gas production to obtain the optimal optimization parameters.

[0042] In a third aspect, the present invention provides an electronic device, comprising:

[0043] processor;

[0044] a memory for storing processor-executable instructions;

[0045] The processor is configured to execute to implement the gas well bubble drainage system optimization method provided in the first aspect.

[0046] In a fourth aspect, the present invention provides a non-temporary computer-readable storage medium, which, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to implement the machine learning-based gas well bubble drainage system optimization method provided in the first aspect.

[0047] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0048] The present invention realizes online, near real-time and dynamic determination of the injection and production work system in a data-driven manner, effectively solving the contradictions of large workload, low timeliness and low coverage of gas well differential characteristics in analyzing the bubble and exhaust well system, which affect the gas well productivity. It effectively improves the analysis of the effect of the bubble and exhaust well system, optimizes the decision-making efficiency, and helps to effectively utilize the productivity of the bubble and exhaust wells.

[0049] The present invention uses easily accessible parameters to build the model. When building the model, it determines the data usage logic according to different data dimension characteristics, ensures the effective use of data, ensures the effective learning of different data information characteristics, and avoids the influence of subjective factors on the results.

[0050] The present invention fully considers the adaptability of the agent in the region and the conditions of the on-site equipment when setting the boundary conditions, ensuring that the recommendation results formed can be effectively implemented.

[0051] In the context of multi-objective optimization, this invention comprehensively considers the constraints of different objective parameters and determines targeted output logic, avoiding ambiguity in the output results and ensuring the uniqueness of the results. This invention establishes a targeted algorithm performance assurance mechanism, determines a targeted monitoring mechanism based on the two aspects of algorithm operation stability and effectiveness, and determines the corresponding algorithm self-optimization and self-iteration to ensure the algorithm's continued and effective operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0053] Figure 1 A schematic flow chart of the method for optimizing the gas well bubble drainage system based on machine learning provided by the present invention;

[0054] Figure 2 A schematic flow chart of another method for optimizing the gas well bubble drainage system based on machine learning provided by the present invention;

[0055] Figure 3 This is a schematic structural diagram of the gas well bubble drainage system optimization device based on machine learning provided by the present invention. DETAILED DESCRIPTION

[0056] The embodiment of the present invention solves the technical problem in the prior art that the gas well bubble drainage and injection system is heavily dependent on work experience by providing a method for optimizing the gas well bubble drainage system based on machine learning.

[0057] The technical solution of the present invention is to solve the above technical problems, and the overall idea is as follows:

[0058] A method for optimizing a gas well foaming system based on machine learning comprises: screening production parameters, reservoir parameters, and basic parameters of the foaming well based on the correlation between the dynamic and static characteristics of the foaming well and the daily gas production to obtain a correlation data set; dividing the correlation data set into a training set and a test set; training a foaming well daily gas production prediction model based on the training set, and verifying the foaming well daily gas production prediction model with the test set, and outputting the foaming well daily gas production prediction model when preset training requirements are met; determining a foaming agent daily injection amount range, an injection cycle range, and a dilution ratio range for the foaming well; constructing several sets of optimization parameters based on the foaming agent daily injection amount range, injection cycle range, and dilution ratio range for the foaming well, and inputting the optimization parameters into the foaming well daily gas production prediction model to obtain a predicted daily gas production corresponding to the optimization parameters, wherein a set of optimization parameters includes a foaming agent daily injection amount, an injection cycle, and a dilution ratio; and screening the several sets of optimization parameters with the goal of maximizing the predicted daily gas production to obtain the optimal optimization parameters.

[0059] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0060] First, the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. Furthermore, the character " / " in this document generally indicates an "or" relationship between the associated objects.

[0061] Foam drainage gas recovery is a process that injects surfactants (foaming agents) into the gas well, uses the agitation of the natural gas flow to generate low-density water-containing foam, and carries it from the bottom of the well to the surface with the air flow, thereby carrying the bottom-well liquid to the surface and achieving the purpose of removing the bottom-well liquid.

[0062] The present invention provides Figure 1 The method for optimizing the gas well bubble drainage system based on machine learning includes steps S11-S16:

[0063] Step S11 , based on the correlation between the dynamic and static characteristics of the gas-bubbling and gas-exhaust wells and the daily gas production, the production parameters, gas reservoir parameters and basic parameters of the gas-bubbling and gas-exhaust wells are screened to obtain a correlation data set.

[0064] Specifically, the method includes: constructing a basic data set based on the production parameters of gas wells, gas reservoir parameters of gas wells, and basic well parameters; screening the parameters in the basic data set based on the Pearson correlation analysis method to obtain parameters related to the daily gas production of gas wells; and constructing a related data set based on the parameters related to the daily gas production of gas wells.

[0065] The production parameters of gas wells with bubble drainage can include gas well production data, bubble drainage process system, etc.; gas reservoir parameters of gas wells with bubble drainage can include EUR, pore saturation, etc.; basic well parameters can include wellbore structure and tubing structure.

[0066] The Pearson Correlation Coefficient is a statistical indicator that measures the strength and direction of the linear relationship between two variables.

[0067] Based on the Pearson correlation coefficient, it is possible to measure which of the above parameters are related to the daily gas production of the bubbling and exhaust gas wells. The parameters related to the daily gas production of the bubbling and exhaust gas wells are collected to obtain a correlation data set.

[0068] Related data sets may include: daily gas production, daily water production, gas-water ratio, oil pressure, casing pressure, transmission pressure, wellhead temperature, cumulative production days of gas wells, cumulative gas production, cumulative water production, cumulative duration of foam drainage, daily foam drainage agent injection volume, injection cycle, dilution ratio, foam drainage agent type; well type, reservoir depth, casing inner diameter, tubing outer diameter, tubing inner diameter, tubing depth; formation pressure, EUR, gas production index, and liquid production index.

[0069] It's understandable that parameters related to daily gas production from bubbler-gas wells can be divided into two categories: static well parameters and dynamic production parameters. Static well parameters, such as casing ID, tubing OD, tubing ID, and tubing run depth, do not change over time (not exhaustive). Dynamic production parameters are related to dynamic production conditions and may include cumulative production days, cumulative gas production, and cumulative water production (not exhaustive).

[0070] Step S12: Divide the dataset into a training set and a test set based on the associated dataset.

[0071] Specifically, the method includes: concatenating the parameters in the associated data sets according to the data information dimensions to obtain the wide data tables at each dimensional level; organizing the wide data tables according to the correlation between the wide data tables and the daily gas production of the bubbling and exhaust gas wells; and dividing the data into training and test sets based on the organized wide data tables.

[0072] Based on predefined data information dimensions (such as time, geographic location, and production parameter dimensions), the scattered parameters are horizontally spliced ​​to obtain a wide table. In other words, each row represents an observation unit (such as a daily record of a gas-sucking and gas-exhausting well), and the columns contain all relevant characteristic variables (such as date, well depth, reservoir pressure, and water production). The correlation between the wide data table and the daily gas production of the gas-sucking and gas-exhausting well can also be determined based on the Pearson correlation coefficient. When the Pearson correlation coefficient between a parameter in the wide data table and the daily gas production of the gas-sucking and gas-exhausting well is lower than a threshold, the parameter is removed to obtain a sorted wide data table. (For information on wide data tables, see step S11 in patent document CN119337754B.)

[0073] After obtaining the sorted data wide table, it can be divided into a training set and a test set according to a preset ratio. The training set and the test set are used to train the daily gas production prediction model of the bubble exhaust well described later.

[0074] Step S13: training the daily gas production prediction model for the bubbling exhaust well based on the training set, and verifying the daily gas production prediction model for the bubbling exhaust well using the test set. When the preset training requirements are met, the daily gas production prediction model for the bubbling exhaust well is output.

[0075] Specifically, it includes: constructing a daily gas production prediction model for bubble exhaust wells based on the training set, wherein the daily gas production prediction model for bubble exhaust wells predicts the gas production of bubble exhaust wells according to time series; using DILATE as the loss function of the daily gas production prediction model for bubble exhaust wells, and using Adam as the optimizer of the daily gas production prediction model for bubble exhaust wells; when DILATE meets the preset application requirements, inputting the test set into the daily gas production prediction model for bubble exhaust wells; and outputting the daily gas production prediction model for bubble exhaust wells when the preset training requirements are met.

[0076] The preset application requirement may be, for example, the number of training sessions, and the preset training requirement may be, for example, the accuracy of the prediction results. These requirements may be determined based on actual circumstances. It is understood that when inputting the training set into the daily gas production prediction model for the bubbling and exhaust wells, the parameters may be input in chronological order so that the daily gas production prediction model for the bubbling and exhaust wells can predict the gas production of the bubbling and exhaust wells in chronological order.

[0077] DILATE is a combination of Dynamic Time Warping (DTW) and Temporal Localization Loss, designed to address the temporal misalignment problem in time series forecasting tasks. DILATE improves forecasting accuracy by simultaneously minimizing both the shape difference and temporal localization error between the predicted and true series.

[0078] The Adam optimizer is an adaptive learning rate optimization algorithm that combines the advantages of both the AdaGrad and RMSProp optimization algorithms. It effectively accelerates stochastic gradient descent and automatically adjusts the learning rate of each parameter.

[0079] Specifically, based on the training set, a daily gas production prediction model for bubble-displacement wells based on time series prediction is constructed. The correlation features between the parameters in the training set and the daily gas production of bubble-displacement wells are extracted, and the daily gas production of bubble-displacement wells under static conditions of the gas wells and dynamic production conditions are predicted. DILATE is used as the loss function for training the daily gas production prediction model for bubble-displacement wells. Adam is selected as the optimizer, and the state update parameters are changed until DILATE meets the preset application requirements. The preset application requirements can be determined according to actual conditions.

[0080] The test set is input into the daily gas production prediction model of the bubble drainage well to test the effectiveness and accuracy of the established bubble drainage well daily gas production prediction model. If the results meet the preset training requirements, it indicates that the bubble drainage well daily gas production prediction model has effectively learned the influence of various parameters on the daily gas production of the bubble drainage well and meets the basic conditions for model use.

[0081] If the preset training requirements are not met, the formed bubble drainage well daily gas production prediction model is optimized and iterated until the test set is brought into the bubble drainage well daily gas production prediction model to meet the preset training requirements.

[0082] Step S14, determining the daily filling amount range, filling cycle range and dilution ratio range of the foaming agent for the foaming gas well.

[0083] Specifically, it includes: determining the daily filling amount range of the foaming agent for the foaming gas well according to the effective concentration range of the foaming gas well and the daily water production of the foaming gas well; determining the filling cycle range and dilution ratio range according to the performance of the filling equipment used in the foaming gas well.

[0084] In the relevant field, the daily injection amount range of the foaming agent for the air-exhausting well can be determined according to the effective concentration range of the foaming agent and the daily water production of the air-exhausting well.

[0085] Filling equipment performance may include the equipment's maximum and minimum filling rates, stability, and accuracy.

[0086] The filling cycle range and dilution ratio range (water and foaming agent dilution ratio) can be determined based on the performance of the filling equipment.

[0087] Step S15, constructing several sets of optimization parameters based on the daily injection amount range, injection cycle range and dilution ratio range of the foaming and exhausting gas wells, and inputting the optimization parameters into the daily gas production prediction model of the foaming and exhausting gas wells to obtain the predicted daily gas production corresponding to the optimization parameters, wherein a set of optimization parameters includes a daily injection amount of the foaming and exhausting agent, a injection cycle and a dilution ratio.

[0088] A daily injection amount of a foaming agent, a injection cycle range and a dilution ratio are arbitrarily selected from the range of daily injection amount of a foaming agent, an injection cycle range and a dilution ratio of the foaming agent for the gas well to form a set of optimization parameters (several sets of optimization parameters can be obtained by using a particle swarm optimization algorithm).

[0089] By inputting the optimized parameters into the daily gas production prediction model of the bubble exhaust well, the predicted daily gas production corresponding to the optimized parameters can be obtained.

[0090] Step S16, with the goal of maximizing the predicted daily gas production, several groups of optimization parameters are screened to obtain the optimal optimization parameters.

[0091] Specifically, it includes: with the goal of maximizing the predicted daily gas production, selecting the optimization parameters corresponding to the maximum predicted daily gas production from several groups of optimization parameters; if there is only one group of optimization parameters corresponding to the maximum predicted daily gas production, then this group of optimization parameters is used as the optimal optimization parameters; if there are two or more groups of optimization parameters corresponding to the maximum predicted daily gas production, with the goal of lowest cost, selecting the optimization parameters with the lowest cost from several optimization parameters corresponding to the maximum predicted daily gas production; if there is only one group of optimization parameters with the lowest cost, then this group of optimization parameters is used as the optimal optimization parameters; if there are two or more groups of optimization parameters with the lowest cost, with the goal of minimizing the additional on-site workload, selecting the optimal optimization parameters from several optimization parameters with the lowest cost.

[0092] The lowest cost means the lowest amount of foaming agent is used, and the minimum additional on-site workload means the least amount of action required by relevant personnel to adjust the filling equipment.

[0093] The method further includes: determining a deviation value according to the optimal optimization parameters of the bubble exhaust well; and updating the daily gas production prediction model of the bubble exhaust well when the deviation value is greater than a preset threshold.

[0094] It is understandable that when predicting the same gas production well, there should not be too large a parameter gap between two optimal optimization parameters with similar time series. If the gap between the two parameters is too large, it means that the deviation is too large, and the daily gas production prediction model of the gas production well needs to be retrained.

[0095] To ensure the stability and effectiveness of the algorithm, when deviation occurs, the self-iteration mechanism is triggered to update and iterate the algorithm to ensure the performance of the algorithm.

[0096] All bubble system characteristics can be brought into the model again and the model can be retrained to solve the problem of reduced model stability after changes in production characteristics. When the algorithm effectiveness deviates, combined with the algorithm effectiveness feedback results, on the basis of inheriting the accurate recommendations in the early stage, the data features with invalid recommendations are analyzed, and the data features with poor effects are re-learned and trained, thereby improving the effectiveness of the algorithm results.

[0097] The gas well bubble discharge system optimization method based on machine learning provided by the present invention can also refer to Figure 2 .

[0098] In summary, the present invention provides a method for optimizing the gas well bubble drainage system based on machine learning, the method comprising: screening the bubble drainage well production parameters, bubble drainage well gas reservoir parameters and well basic parameters based on the gas production correlation of the bubble drainage well daily gas production to obtain an associated data set; dividing the associated data set into a training set and a test set; training the bubble drainage well daily gas production prediction model based on the training set, and verifying the bubble drainage well daily gas production prediction model with the test set, and outputting the bubble drainage well daily gas production prediction model when the preset training requirements are met; determining the bubble drainage agent daily injection amount range, injection cycle range and dilution ratio range of the bubble drainage well; constructing several groups of optimization parameters according to the bubble drainage agent daily injection amount range, injection cycle range and dilution ratio range of the bubble drainage well, and inputting the optimization parameters into the bubble drainage well daily gas production prediction model to obtain the predicted daily gas production corresponding to the optimization parameters, wherein a group of optimization parameters includes a bubble drainage agent daily injection amount, a injection cycle and a dilution ratio; with the goal of maximizing the predicted daily gas production, screening several groups of optimization parameters to obtain the optimal optimization parameters. The present invention uses readily available parameters to build a model. When building the model, it determines the data usage logic based on different data dimension characteristics, ensuring the effective use of data, ensuring the effective learning of different data information characteristics, and avoiding the influence of subjective factors on the results. When setting boundary conditions, the present invention fully considers the adaptability of the agent in the region and the conditions of the on-site equipment to ensure that the recommended results can be effectively implemented and applied. In the case of multi-objective optimization, the present invention comprehensively considers the constraints of different target parameters and determines targeted output logic to avoid ambiguity in the output results and ensure the uniqueness of the results. The present invention establishes a targeted algorithm performance guarantee mechanism, determines a targeted monitoring mechanism from the two aspects of algorithm operation stability and effectiveness, and determines the corresponding algorithm self-optimization and self-iteration to ensure the continuous and effective operation of the algorithm. The present invention realizes online, near real-time, and dynamic determination of the injection and production work system through data-driven methods, effectively solving the contradictions of large workload, low timeliness, and low coverage of gas well differential characteristics in analyzing the bubble and exhaust well system, which affect the gas well productivity. It effectively improves the efficiency of bubble and exhaust well system effect analysis and optimization decision-making, and helps bubble and exhaust well productivity to be effectively utilized.

[0099] Based on the same inventive concept, the present invention provides Figure 3 The device for optimizing the gas well bubble discharge system based on machine learning is shown, and the device includes:

[0100] A basic data screening module 31 is used to screen the production parameters, gas reservoir parameters and basic parameters of the bubbling and exhaust gas wells based on the correlation between the dynamic and static characteristics of the bubbling and exhaust gas wells and the daily gas production to obtain a related data set;

[0101] A partitioning module 32 is used to partition a training set and a test set based on the associated data set;

[0102] The model training module 33 is used to train the daily gas production prediction model of the bubbling exhaust well based on the training set, and verify the daily gas production prediction model of the bubbling exhaust well using the test set. When the preset training requirements are met, the daily gas production prediction model of the bubbling exhaust well is output;

[0103] A range determination module 34 is used to determine the daily injection amount range, injection cycle range and dilution ratio range of the foaming agent for the foaming gas well;

[0104] The prediction module 35 is used to construct a plurality of sets of optimization parameters based on the range of daily injection amount, injection cycle, and dilution ratio of the foaming and exhausting wells, and input the optimization parameters into the foaming and exhausting well daily gas production prediction model to obtain the predicted daily gas production corresponding to the optimization parameters, wherein a set of optimization parameters includes a daily injection amount of the foaming and exhausting agent, a injection cycle, and a dilution ratio;

[0105] The optimal parameter screening module 36 is used to screen several groups of optimization parameters with the goal of maximizing the predicted daily gas production to obtain the optimal optimization parameters.

[0106] Based on the same inventive concept, the present invention further provides an electronic device, comprising:

[0107] processor;

[0108] a memory for storing processor-executable instructions;

[0109] The processor is configured to execute to implement the gas well bubble discharge system optimization method based on machine learning as provided above.

[0110] Based on the same inventive concept, the present invention also provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor of an electronic device, the electronic device is able to implement the gas well bubble drainage system optimization method based on machine learning as provided above.

[0111] Since the electronic device described in this embodiment is an electronic device used to implement the information processing method in the embodiment of the present invention, based on the information processing method described in the embodiment of the present invention, those skilled in the art will be able to understand the specific implementation of the electronic device of this embodiment and its various variations. Therefore, how the electronic device implements the method in the embodiment of the present invention will not be described in detail here. As long as the electronic device used by those skilled in the art to implement the information processing method in the embodiment of the present invention falls within the scope of protection of the present invention.

[0112] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0114] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1A step that specifies a function in one or more boxes.

[0116] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0117] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for optimizing the bubble drainage system of gas wells based on machine learning, characterized in that: The method comprises: Based on the correlation between the dynamic and static characteristics of the bubbling and exhaust gas wells and the daily gas production, the production parameters, gas reservoir parameters and basic parameters of the bubbling and exhaust gas wells are screened to obtain a correlation data set, including: constructing a basic data set based on the production parameters, gas reservoir parameters and basic parameters of the bubbling and exhaust gas wells; screening the parameters in the basic data set based on the Pearson correlation analysis method to obtain parameters related to the daily gas production of the bubbling and exhaust gas wells; and constructing a correlation data set based on the parameters related to the daily gas production of the bubbling and exhaust gas wells; Based on the associated data set, dividing the data into a training set and a test set; The daily gas production prediction model for the bubble exhaust well is trained based on the training set, and the daily gas production prediction model for the bubble exhaust well is verified using the test set. When the preset training requirements are met, the daily gas production prediction model for the bubble exhaust well is output. The method includes: constructing the daily gas production prediction model for the bubble exhaust well based on the training set, wherein the daily gas production prediction model for the bubble exhaust well predicts the gas production of the bubble exhaust well according to the time series; using DILATE as the loss function of the daily gas production prediction model for the bubble exhaust well, and using Adam as the optimizer of the daily gas production prediction model for the bubble exhaust well; when DILATE meets the preset application requirements, inputting the test set into the daily gas production prediction model for the bubble exhaust well, and when the preset training requirements are met, outputting the daily gas production prediction model for the bubble exhaust well; Determine the daily injection volume range, injection cycle range, and dilution ratio range of the foaming agent for the foaming exhaust well; Constructing several sets of optimization parameters based on the daily injection amount range, injection cycle range, and dilution ratio range of the foaming and exhausting wells, and inputting the optimization parameters into the daily gas production prediction model of the foaming and exhausting wells to obtain the predicted daily gas production corresponding to the optimization parameters, wherein a set of optimization parameters includes a daily injection amount of the foaming and exhausting agent, a injection cycle, and a dilution ratio; With the goal of maximizing the predicted daily gas production, several groups of optimization parameters are screened to obtain the optimal optimization parameters, including: with the goal of maximizing the predicted daily gas production, the optimization parameters corresponding to the maximum predicted daily gas production are screened from several groups of optimization parameters; if there is only one group of optimization parameters corresponding to the maximum predicted daily gas production, then this group of optimization parameters is used as the optimal optimization parameters; if there are two or more groups of optimization parameters corresponding to the maximum predicted daily gas production, then with the goal of minimizing cost, the optimization parameters with the lowest cost are screened from several optimization parameters corresponding to the maximum predicted daily gas production; if there is only one group of optimization parameters with the lowest cost, then this group of optimization parameters is used as the optimal optimization parameters; if there are two or more groups of optimization parameters with the lowest cost, then with the goal of minimizing on-site additional workload, the optimal optimization parameters are screened from several optimization parameters with the lowest cost.

2. The method for optimizing the gas well bubble discharge system based on machine learning according to claim 1, characterized in that: Based on the associated data set, the training set and the test set are divided, including: Performing data splicing on the parameters in the associated data set according to the data information dimension to obtain a wide data table at each dimensional level; The wide data table was sorted based on the correlation between the wide data table and the daily gas production of the bubble gas wells; Based on the sorted data wide table, divide the training set and test set.

3. The method for optimizing the gas well bubble discharge system based on machine learning according to claim 1, characterized in that: Determine the daily injection volume range, injection cycle range, and dilution ratio range of the foaming agent for the foaming exhaust well, including: Determine the daily injection amount range of the foaming agent for the air-exhaust well based on the effective concentration range of the foaming agent and the daily water production of the air-exhaust well. The filling cycle range and dilution ratio range are determined based on the performance of the filling equipment used in the bubble exhaust well.

4. The method for optimizing the gas well bubble discharge system based on machine learning according to claim 1, wherein: The method further comprises: Determine the deviation value based on the optimal optimization parameters of the bubble exhaust well; When the deviation value is greater than a preset threshold, the daily gas production prediction model of the bubble exhaust well is updated.

5. A device for optimizing the gas well bubble discharge system based on machine learning, characterized in that: The method for optimizing the gas well bubble discharge system based on machine learning as claimed in any one of claims 1 to 4, wherein the device comprises: The basic data screening module is used to screen the production parameters, gas reservoir parameters and basic parameters of the bubbling and exhaust gas wells based on the correlation between the dynamic and static characteristics of the bubbling and exhaust gas wells and the daily gas production, and obtain the associated data set; A partitioning module, configured to partition the associated dataset into a training set and a test set; a model training module, configured to train a daily gas production prediction model for a bubbling exhaust well based on the training set, and verify the daily gas production prediction model for the bubbling exhaust well using a test set, and output the daily gas production prediction model for the bubbling exhaust well when preset training requirements are met; A range determination module is used to determine the daily injection amount range, injection cycle range and dilution ratio range of the foaming agent in the foaming exhaust well; a prediction module for constructing a plurality of sets of optimization parameters based on a range of daily injection amount, injection cycle, and dilution ratio of a foaming agent for a gas well, and inputting the optimization parameters into a daily gas production prediction model for the gas well to obtain a predicted daily gas production corresponding to the optimization parameters, wherein a set of optimization parameters includes a daily injection amount of a foaming agent, a injection cycle, and a dilution ratio; The optimal parameter screening module is used to screen several groups of optimization parameters with the goal of maximizing the predicted daily gas production to obtain the optimal optimization parameters.

6. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute to implement the gas well bubble discharge system optimization method based on machine learning as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium, characterized in that When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to implement the method for optimizing the gas well bubble discharge system based on machine learning as described in any one of claims 1 to 4.

Images