Intelligent injection and production allocation method for gas storage based on historical operation data

Abstract

The invention belongs to the technical field of gas storages, and particularly relates to an intelligent injection and production allocation method for a gas storage based on historical operation data, which comprises the following specific steps: establishing an intelligent injection and production model foundation and optimizing a workflow: firstly, establishing constraint conditions for an intelligent injection and production scene of an underground gas storage; then, a data link meeting constraint and feedback of coupling calculation of different models is established on the basis of analysis of coupling calculation of all the models, and then a standard working mode of intelligent injection and production allocation is formulated on the basis of model requirements. According to the gas reservoir-shaft-pipe network integrated coupling model constructed by the method, a mechanism model and an AI model are combined; the mechanism model can simulate the flowing process of gas in a gas reservoir, and the AI model can supplement and optimize complex phenomena which are difficult to accurately describe by the mechanism model through learning and analyzing a large amount of historical production data.

Description

Technical Field

[0001] The present invention relates to the technical field of gas storage, and in particular to an intelligent injection and production method for a gas storage based on historical operation data. Background Art

[0002] Amidst profound changes in the global energy landscape, natural gas, as a clean and efficient energy source, continues to grow in its share of energy consumption. As a key link in the natural gas industry chain (production, supply, storage, and sales), gas storage plays an irreplaceable role in ensuring a stable supply of natural gas, addressing seasonal demand fluctuations, and enhancing national energy security. my country's natural gas industry is developing rapidly, but insufficient storage capacity has long constrained the sustainable and high-quality development of the natural gas business. According to analysis by the International Gas Union (IGU), when natural gas dependence exceeds a certain percentage, clear requirements for the working gas volume of underground gas storage facilities are established. By the end of 2021, although my country had built a number of gas storage facilities, their storage capacity accounted for only 4.4% of that year's natural gas consumption, far below the IGU's target of 12%. Against this backdrop, improving the operational efficiency and management of gas storage facilities is urgent.

[0003] However, the existing gas storage facilities have the following problems:

[0004] 1. Limitations of Relying on Theoretical Model Predictions

[0005] The operation and management of traditional gas storage facilities rely heavily on theoretical models for prediction and analysis. When describing the seepage process of gas in complex geological structures, theoretical models for numerical simulation of gas reservoirs often simplifies the geological body and makes it difficult to accurately obtain all geological parameters, resulting in deviations between the simulation results and the actual situation. In some gas storage facilities with complex pore structures and fracture networks, theoretical models have difficulty accurately depicting the flow path and distribution of gas, making predictions of dynamic changes in gas reservoirs less precise. This may result in the actual injection rate being inconsistent with theoretical expectations during the gas injection process, making it impossible to reach the target reservoir capacity within the expected time; during the gas production phase, the estimated deviation in the gas production capacity of a single well affects the overall production of peak-shaving gas.

[0006] (2) Technical Difficulties in Optimizing Analysis

[0007] Difficulties in collaborative optimization of multiple links: Gas storage production encompasses multiple links, including gas reservoirs, wellbores, and pipeline networks, each of which influences and constrains each other. Current optimization and analysis technologies struggle to comprehensively coordinate the complex relationships among these links and achieve overall collaborative optimization. When adjusting the gas production rate for a particular well, the resulting chain reaction on wellbore pressure distribution and pipeline flow distribution may not be fully considered, resulting in reduced efficiency across the entire system and even posing safety risks.

[0008] Lack of in-depth data mining: After years of operation, gas storage facilities have accumulated a vast amount of historical production data, but existing optimization and analysis technologies fail to fully utilize this data. This valuable information remains unused, making it impossible to extract optimal operating modes and patterns under different operating conditions. Injection and production data from different seasons and production stages contain crucial information about changes in reservoir characteristics and equipment operating status. However, traditional technologies are unable to effectively mine this information, making it difficult to optimize injection and production allocation plans.

[0009] Incomplete emergency response mechanisms: Current emergency injection and production plans and the technology used to trigger and execute them are slow to respond and lack specificity in the face of emergencies, such as sudden increases in gas demand due to extreme weather, sudden wellbore failures, and abnormal gas reservoir production. When a winter cold snap causes a sudden surge in natural gas demand, it's impossible to quickly generate a comprehensive emergency production adjustment plan to meet this sudden demand, potentially posing a threat to energy supply security.

[0010] Based on the above, a method for intelligent injection and production allocation of a gas storage based on historical operation data is invented. Summary of the Invention

[0011] To solve the above technical problems, according to one aspect of the present invention, the present invention provides the following technical solutions:

[0012] A method for intelligent injection and production allocation of a gas storage facility based on historical operation data includes the following specific steps:

[0013] S1. Establish the foundation of the intelligent injection and production model and optimize the workflow: First, establish constraints for the intelligent injection and production scenario of underground gas storage. Then, based on the analysis of the coupled calculations of each model, build a data link that satisfies the constraints and feedback of the coupled calculations of different models. Finally, based on the model requirements, formulate a standard working mode for intelligent injection and production allocation.

[0014] S2, building an intelligent injection and production AI analysis model based on historical data: First, collect and organize gas storage data and model simulation data, then conduct in-depth analysis of the data and build an intelligent injection and production AI analysis model based on the analysis results;

[0015] S3, Injection and Production Allocation Model Construction and Optimization: First, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed based on the mechanism model and AI model. Then, the established integrated coupling model is used to simulate and predict the injection and production plan, and then the injection and production plan is optimized;

[0016] S4, establish intelligent injection and procurement plans: First, clarify the technical boundaries for coordinated optimization between the formulation of injection and production plans and the AI ​​injection and procurement plans. Then, based on the determined technical boundaries, develop a compilation method for intelligent injection and procurement plans based on different gas storage capacity expansion and production stages. Then, based on the compilation method, realize the automatic output of intelligent injection and procurement plans.

[0017] S5, establish an intelligent injection-production index optimization analysis method: first analyze the technical policy and optimal time step for coupling the dynamic data of gas storage production and injection allocation with static data, then analyze the coupling method of automatic driving model update by dynamic data. Finally, through the formation method of intelligent analysis of production dynamic curves, intelligent statistics, comparison, and analysis of production indicators are performed;

[0018] S6, triggering and executing emergency injection and production plans: First, establish emergency injection and production conditions that are suitable for different scenarios. Then, formulate emergency injection and production execution procedures, clarifying each link and operational steps from emergency event triggering to plan execution. Finally, analyze the specific requirements of the emergency production adjustment plan to achieve rapid analysis of the emergency consequences of the gas storage.

[0019] S7, model optimization and iteration: First, based on the historical data of the gas storage, we use big data combined with machine learning technology to develop an intelligent control expert system that integrates the model. Then, we use the data accumulated from years of operation of the existing gas storage to continuously iterate the model. Then, through continuous training and calculation, we deeply explore the parameters and factors that affect the operation of the gas storage.

[0020] As a preferred solution of the method for intelligent injection and production allocation of a gas storage based on historical operation data of the present invention, the specific steps of S1 are as follows:

[0021] S11, build constraints: for the intelligent injection and production scenario of underground gas storage, establish detailed constraints for each node of geology, wellbore, ground station, and external transmission pipeline;

[0022] S12, Optimize data links: Analyze the supporting environment and data transmission requirements for the coupled calculations of each model, and build data links that meet the constraints and feedback of the coupled calculations of different models;

[0023] S13, establish a standard working mode: Based on the model requirements, formulate a complete set of standard working modes that meet the requirements of intelligent injection and procurement.

[0024] As a preferred solution of the intelligent injection and production method of a gas storage based on historical operation data of the present invention, the specific steps of S2 are as follows:

[0025] S21, Data Collection and Collation: Collect development and research data of gas storage facilities, as well as simulation data of gas reservoirs, wellbore, and pipeline network models at different times. At the same time, collate a large amount of historical production data accumulated by different injection and production wells during multiple injection and production cycles, and perform pre-processing operations to ensure the accuracy, completeness, and consistency of the data.

[0026] S22, Big Data Analysis: Use big data analysis technology to conduct in-depth analysis of processed data to establish a database of optimal injection and production conditions for different production wells under different constraints;

[0027] S23, AI model construction and training: First, based on the results of big data analysis, a machine learning algorithm is used to build an intelligent injection and production AI analysis model; then, historical production data is used as training samples, various constraints are used as input features, and the optimal injection and production operating parameters are used as output labels. The model is repeatedly trained and optimized to improve the model's prediction accuracy and generalization ability by adjusting the model structure and parameter settings, so that it can accurately predict the optimal injection and production conditions under given constraints, providing decision support for intelligent injection and production allocation.

[0028] As a preferred solution of the method for intelligent injection and production allocation of a gas storage based on historical operation data of the present invention, the specific steps of S3 are as follows:

[0029] S31, Establishment of an integrated coupling model: Based on the mechanism model and AI model, combined with the material balance, seepage law, and production constraints of the gas reservoir, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed;

[0030] S32, Model Simulation and Prediction: Utilizing the established integrated coupling model, simulate the gas flow process in the gas reservoir, wellbore, and pipeline network under different injection and production schemes. This model predicts changes in gas reservoir pressure and saturation, wellbore pressure distribution and flow rate changes, and pressure and flow rate parameters at each node in the pipeline network. This model simulation compares the advantages and disadvantages of various schemes, providing a scientific basis for optimizing injection and production allocation schemes.

[0031] S33, Scheme Optimization: Using optimization algorithms, the injection and production scheme is optimized with the goal of improving injection and production efficiency, increasing recovery rate, and ensuring safe production. During the optimization process, the constraints of the gas reservoir, wellbore, and pipeline network are used as optimization boundary conditions, and the decision variables of the injection and production wells are adjusted to seek the optimal combination of injection and production schemes. Through continuous iterative optimization, the gas storage facility can maximize its economic and social benefits while ensuring safe operation.

[0032] As a preferred solution of the intelligent injection and production method of a gas storage based on historical operation data of the present invention, the specific steps of S4 are as follows:

[0033] S41, Determine Technical Boundaries: Clarify the technical boundaries for coordinated optimization between the injection and production allocation plan based on the twin model of geology, wellbore, and surface pipeline network, and the AI-based injection and production allocation plan based on historical data. This will allow the respective applicability and weighting of the two plans to be determined at different stages of gas storage capacity expansion and production, taking into account changes in gas reservoir geological characteristics, equipment operating status, and market demand fluctuations.

[0034] S42, Scheme Preparation Method: Based on the determined technical boundaries, formulate a method for preparing intelligent injection and production allocation schemes based on different gas storage expansion and production stages. At the same time, the corresponding technical indicator requirements for intelligent production allocation are given. For each indicator, combining model prediction results and historical data experience, a reasonable value range and adjustment strategy are formulated.

[0035] S43, plan output and automated control: Based on the compilation method, the automatic output of intelligent injection and procurement plans is realized; first, the model calculation results, optimization algorithm output and technical indicator requirements are integrated, and a detailed injection and procurement plan is generated with one click through a friendly human-computer interface; then a connection is established with the automated production allocation process equipment, and the production allocation instructions in the plan are automatically transmitted to the on-site equipment to realize the adjustment of the automated production allocation process conditions, reduce manual intervention, and improve the accuracy and timeliness of plan execution.

[0036] As a preferred solution of the method for intelligent injection and production allocation of a gas storage based on historical operation data of the present invention, the specific steps of S5 are as follows:

[0037] S51, Coupling Dynamic and Static Data: Analyze the technical policies and optimal time steps for coupling dynamic and static data for gas storage production and injection allocation. Dynamic data includes time-varying data, while static data includes relatively fixed data. First, by rationally setting the time step, dynamic and static data are organically integrated to enable the model to more accurately reflect the real-time operating status and long-term evolution trends of the gas storage. Then, when analyzing reservoir pressure changes, the real-time monitored pressure data is combined with the static data in the geological model to more accurately predict future changes in reservoir pressure.

[0038] S52, Dynamic Data Update and Model Coupling: This method uses dynamic data analysis to automatically drive model update coupling. First, the real-time collected dynamic data is used to automatically trigger model update calculations, allowing the model to promptly adapt to changes in the gas storage reservoir's operating status. Then, when a well's gas flow rate changes abnormally, the new flow rate data is automatically input into the integrated coupling model, recalculating relevant parameters of the gas reservoir, wellbore, and pipeline network, and promptly adjusting the injection and production allocation plan to ensure stable system operation.

[0039] S53, production dynamic curve analysis and early warning: Through intelligent analysis of the formation method of production dynamic curve, intelligent statistics, comparison and analysis of production indicators are carried out; first, the curve of production parameters changing over time is analyzed, and the characteristic parameters of the curve are extracted. Then, the curve is compared with historical data and preset standard values ​​to realize abnormal early warning identification; when the slope of a production curve exceeds the normal range, an early warning message will be automatically issued to remind management personnel of possible equipment failure or gas reservoir abnormality, and relevant comparison data will be promptly fed back to injection and production management personnel to enable them to make decisions.

[0040] As a preferred solution of the method for intelligent injection and production allocation of a gas storage based on historical operation data of the present invention, the specific steps of S6 are as follows:

[0041] S61, Emergency Condition Setting: Establish emergency injection and production conditions that adapt to different scenarios, including triggering conditions for emergency injection and production in the event of changes in dispatch instructions, wellbore outburst conditions, and abnormal gas reservoir production conditions. Specific triggering thresholds and judgment criteria are defined for each emergency situation.

[0042] S62, Implementation Procedure Development: Develop detailed emergency injection and production implementation procedures, clarifying each link and operational steps from the triggering of the emergency event to the implementation of the plan, including the activation of the emergency command system, the division of responsibilities among departments, the information transmission process, the equipment operation process, and safety measures;

[0043] S63, Emergency Consequences Analysis and Rapid Response: Analyze the specific requirements of emergency production adjustment plans and realize rapid analysis of emergency consequences of gas storage facilities; utilize the established integrated coupling model to quickly simulate the possible consequences of different emergency production adjustment plans after an emergency event occurs, providing a scientific basis for decision-making.

[0044] As a preferred solution of the method for intelligent injection and production allocation of a gas storage based on historical operation data of the present invention, the specific steps of S7 are as follows:

[0045] S71, Algorithm Fusion and Expert System Development: Based on historical gas storage data, this project utilizes big data and machine learning techniques to develop an intelligent control expert system using the Stacking algorithm fusion model. This system first organically integrates multiple machine learning algorithms to leverage their respective strengths and improve the model's predictive accuracy and adaptability. Then, through learning and analyzing historical data, the expert system simulates the decision-making process of domain experts, providing professional advice and guidance for intelligent injection and production allocation at gas storage facilities.

[0046] S72, iterative model training: Utilizing data accumulated over years of operation at existing gas storage facilities, the model is continuously iteratively trained. New historical data is regularly incorporated into the training set and the model is retrained, enabling it to continuously learn new patterns and characteristics of gas storage operation. The coefficients of the mechanism model are then modified through data inference, making the model more applicable to practical applications.

[0047] S73, Parameter and Factor Optimization: During the continuous training and calculation process, we deeply explore the parameters and factors that affect the operation of the gas storage facility. We first use sensitivity analysis methods to determine the key parameters that have a greater impact on the injection and production allocation effect, and then focus on optimizing and adjusting the parameters. At the same time, we pay attention to newly emerging influencing factors and promptly incorporate them into the model consideration scope to improve the system's computing speed and accuracy, and continuously optimize the intelligent injection and production allocation plan.

[0048] Compared with existing technologies:

[0049] 1. Compensating for the limitations of theoretical model predictions

[0050] The integrated coupling model of "gas reservoir-wellbore-pipeline network" constructed by the present invention combines the mechanism model and the AI ​​model; the mechanism model can simulate the flow process of gas in the gas reservoir, and the AI ​​model can supplement and optimize the complex phenomena that are difficult to accurately describe by the mechanism model by learning and analyzing a large amount of historical production data; for gas storage reservoirs with complex pore structures and fracture networks, the AI ​​model can mine the actual laws of gas flow from historical data, correct the deviations of the theoretical model, and make the prediction of dynamic changes in gas reservoirs more accurate; in the gas injection stage, the gas injection rate can be predicted more accurately to ensure that the target storage capacity is reached within the expected time; in the gas production stage, the gas production capacity of a single well can be accurately estimated to ensure the effective production of the overall peak-shaving gas volume.

[0051] 2. Solve the technical problems of optimization analysis

[0052] Achieve multi-link collaborative optimization: By establishing an intelligent injection-production model foundation and workflow optimization, the constraints of various nodes such as geology, wellbore, surface station, and external transmission pipelines in the intelligent injection-production allocation scenario are clarified, forming a data link between constraints and feedback for the coupled calculations of each model. When adjusting the gas production rate of a well, the chain reaction on the wellbore pressure distribution and pipeline network flow distribution can be fully considered, achieving overall collaborative optimization of multiple links such as gas reservoirs, wellbore, and pipeline network, improving the operating efficiency of the entire system and reducing safety risks.

[0053] Deeply mine the value of historical data: By establishing an intelligent injection and production AI analysis model based on historical data, we can deeply mine the massive amount of historical production data accumulated by gas storage facilities. Using big data analysis methods, we can extract the optimal operating modes and patterns under different operating conditions. From injection and production data in different seasons and production stages, we can mine important information such as changes in gas reservoir characteristics and equipment operating status. Based on this information, we can optimize the injection and production allocation plan in a targeted manner and fully utilize the value of historical data.

[0054] Improve the emergency response mechanism: By establishing an intelligent injection and production allocation plan, establishing an intelligent injection and production indicator optimization analysis method, and realizing the triggering and execution of emergency injection and production plans, the emergency injection and production triggering conditions and execution procedures under different scenarios are clarified; using an integrated coupling model, the consequences of different emergency production adjustment plans can be quickly simulated after an emergency event occurs, and a global emergency production adjustment plan can be given in a short time; when encountering a sudden increase in natural gas demand due to a winter cold wave, etc., a rapid response can be made to meet sudden production adjustment needs and ensure energy supply security. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be described in further detail below.

[0056] The present invention provides a method for intelligent injection and production allocation of a gas storage based on historical operation data, comprising the following specific steps:

[0057] S1. Establish the foundation of the intelligent injection and production model and optimize the workflow: First, establish constraints for the intelligent injection and production scenario of underground gas storage. Then, based on the analysis of the coupled calculations of each model, build a data link that satisfies the constraints and feedback of the coupled calculations of different models. Finally, based on the model requirements, formulate a standard working mode for intelligent injection and production allocation.

[0058] S2, building an intelligent injection and production AI analysis model based on historical data: First, collect and organize gas storage data and model simulation data, then conduct in-depth analysis of the data and build an intelligent injection and production AI analysis model based on the analysis results;

[0059] S3, Injection and Production Allocation Model Construction and Optimization: First, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed based on the mechanism model and AI model. Then, the established integrated coupling model is used to simulate and predict the injection and production plan, and then the injection and production plan is optimized;

[0060] S4, establish intelligent injection and procurement plans: First, clarify the technical boundaries for coordinated optimization between the formulation of injection and production plans and the AI ​​injection and procurement plans. Then, based on the determined technical boundaries, develop a compilation method for intelligent injection and procurement plans based on different gas storage capacity expansion and production stages. Then, based on the compilation method, realize the automatic output of intelligent injection and procurement plans.

[0061] S5, establish an intelligent injection-production index optimization analysis method: first analyze the technical policy and optimal time step for coupling the dynamic data of gas storage production and injection allocation with static data, then analyze the coupling method of automatic driving model update by dynamic data. Finally, through the formation method of intelligent analysis of production dynamic curves, intelligent statistics, comparison, and analysis of production indicators are performed;

[0062] S6, triggering and executing emergency injection and production plans: First, establish emergency injection and production conditions that are suitable for different scenarios. Then, formulate emergency injection and production execution procedures, clarifying each link and operational steps from emergency event triggering to plan execution. Finally, analyze the specific requirements of the emergency production adjustment plan to achieve rapid analysis of the emergency consequences of the gas storage.

[0063] S7, Model Optimization and Iteration: First, based on the historical data of the gas storage, we use big data combined with machine learning technology to develop an intelligent control expert system that applies algorithm fusion models such as Stacking. Then, we use the data accumulated from years of operation of the existing gas storage to continuously iterate and train the model. Then, through continuous training and calculation, we deeply explore the parameters and factors that affect the operation of the gas storage.

[0064] The specific steps of S1 are as follows:

[0065] S11, Constructing Constraints: For the intelligent injection and production scenario of underground gas storage, detailed constraints are established for each node of the geology, wellbore, ground station, and transmission pipeline. In terms of geology, factors such as reservoir permeability distribution, porosity changes, and the impact of faults and fractures on gas flow are considered. In terms of wellbore, constraints include wellbore diameter, pipe material properties, and fluid flow resistance in the wellbore. Ground stations focus on equipment processing capacity, pressure control range, safety and environmental protection requirements, etc. In terms of transmission pipelines, pipe diameter, pipe material, pressure loss along the line, and transmission flow restrictions are considered. By accurately constructing constraints, each component of the digital twin model can effectively support intelligent injection and production.

[0066] S12, Optimize Data Links: Analyze the supporting environment and data transmission requirements for the coupled calculations of each model, and build data links that meet the constraints and feedback of the coupled calculations of different models. Use advanced data transmission protocols and interface technologies to ensure that data such as pressure and saturation calculated by the reservoir model can be accurately and in real time transmitted to the wellbore model. Parameters such as pressure drop and flow rate calculated by the wellbore model are promptly fed back to the reservoir model and pipeline network model, forming an efficient data interaction loop and ensuring the accuracy and stability of the collaborative work of each model.

[0067] S13, establish a standard working mode: based on the model requirements, formulate a complete set of standard working modes that meet the requirements of intelligent injection and procurement; for example, in the stages of data collection and preprocessing, model initialization and loading, multi-model coupling calculation process, result analysis and decision output, clarify the detailed operating steps and specifications; in the data collection stage, specify the installation position, collection frequency, and data accuracy requirements of various sensors; in the model calculation stage, determine the convergence conditions of iterative calculations, calculation time step and other key parameters; by establishing a standard working mode, improve the standardization and repeatability of intelligent injection and procurement work.

[0068] The specific steps of S2 are as follows:

[0069] S21, Data Collection and Collation: Collect development and research data on gas storage facilities, as well as simulation data from gas reservoirs, wellbore, and pipeline network models over different periods of time. At the same time, collate a large amount of historical production data accumulated from multiple injection and production cycles of different injection and production wells, including parameters such as pressure, flow, temperature, and produced / injected gas volumes. Perform pre-processing operations such as cleaning, denoising, and normalization on the data to ensure its accuracy, completeness, and consistency.

[0070] S22, Big Data Analysis: Use big data analysis techniques, such as cluster analysis, association rule mining, and time series analysis, to conduct in-depth analysis of processed data. Through cluster analysis, data from different production wells under similar operating conditions are grouped together to identify typical characteristics under each operating condition. Utilize association rule mining to explore potential correlations between production parameters, such as the association rules between injection pressure and injection flow rate, and between gas production and bottomhole pressure. With the help of time series analysis, predict the changing trends of production parameters over time to establish a database of optimal injection and production conditions for different production wells under different constraints.

[0071] S23, AI model construction and training: First, based on the results of big data analysis, use machine learning algorithms such as neural networks, decision trees, and support vector machines to build an intelligent injection and production AI analysis model; then use historical production data as training samples, various constraints as input features, and optimal injection and production operating parameters as output labels, and repeatedly train and optimize the model. By adjusting the model structure and parameter settings, the model's prediction accuracy and generalization ability can be improved, enabling it to accurately predict the optimal injection and production conditions under given constraints, providing decision support for intelligent injection and production allocation.

[0072] The specific steps of S3 are as follows:

[0073] S31, Establishment of an integrated coupling model: Based on the mechanism model and AI model, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed in combination with the material balance, seepage law and production constraints of the gas reservoir. In the mechanism model, traditional gas reservoir numerical simulation methods, such as the finite difference method and the finite element method, are used to simulate the three-dimensional seepage process of gas in the gas reservoir, considering the influence of factors such as gas reservoir heterogeneity, capillary force, gravity and other factors on gas flow. The AI ​​model is used to learn and predict complex phenomena that are difficult to accurately describe with the mechanism model, such as complex flow patterns in the wellbore and local resistance loss in the pipeline network. By organically combining the two, accurate simulation and analysis of the entire gas storage system can be achieved.

[0074] S32, Model Simulation and Prediction: Utilizing the established integrated coupling model, the gas flow process in the gas reservoir, wellbore, and pipeline network under different injection and production schemes is simulated. Changes in reservoir pressure and saturation, wellbore pressure distribution and flow rate changes, as well as pressure and flow parameters at each node in the pipeline network are predicted. The model simulation compares the advantages and disadvantages of various schemes, providing a scientific basis for optimizing injection and production allocation. Multiple injection and production schemes are designed to address different seasonal peak-shaving needs, reservoir development stages, and equipment operating conditions.

[0075] S33, Scheme Optimization: Use optimization algorithms, such as genetic algorithms and particle swarm optimization algorithms, to optimize the injection and production scheme with the goal of improving injection and production efficiency, enhancing recovery rates, and ensuring safe production. During the optimization process, the constraints of the gas reservoir, wellbore, and pipeline network are used as optimization boundary conditions to adjust the injection volume, production volume, well opening time, and well opening sequence of the injection and production wells, and seek the optimal combination of injection and production schemes. Through continuous iterative optimization, the gas storage facility can maximize its economic and social benefits while ensuring safe operation.

[0076] The specific steps of S4 are as follows:

[0077] S41, Determine Technical Boundaries: Clarify the technical boundaries for coordinated optimization between the injection and production allocation plan based on twin models of geology, wellbore, and surface pipeline networks and the AI ​​injection and production allocation plan based on historical data. This will allow the applicability and weighting of the two plans to be determined at different stages of gas storage capacity expansion and production, based on changes in gas reservoir geological characteristics, equipment operating status, and market demand fluctuations. In the early stages of gas storage construction, when geological conditions are relatively clear, more reliance can be placed on the injection and production allocation plan based on the geological model. As operation time increases and historical data becomes more abundant, the proportion of AI injection and production allocation plans can be gradually increased.

[0078] S42, Scheme Preparation Method: Based on the determined technical boundaries, a method for preparing intelligent injection and production allocation plans based on different gas storage expansion and production stages is developed. At the same time, the corresponding technical indicator requirements for intelligent production allocation are given, including key indicators such as storage capacity, gas storage formation pressure, bottomhole pressure of individual production wells, gas production per well, stage gas production, and key process parameters at each node. For each indicator, a reasonable value range and adjustment strategy are formulated based on the model prediction results and historical data experience.

[0079] S43, plan output and automated control: Based on the compilation method, the automatic output of intelligent injection and procurement plans is realized; first, the model calculation results, optimization algorithm output and technical indicator requirements are integrated, and a detailed injection and procurement plan is generated with one click through a friendly human-computer interface; then a connection is established with the automated production allocation process equipment, and the production allocation instructions in the plan are automatically transmitted to the on-site equipment to realize the adjustment of the automated production allocation process conditions, reduce manual intervention, and improve the accuracy and timeliness of plan execution.

[0080] The specific steps of S5 are as follows:

[0081] S51, coupling dynamic and static data: Analyze the technical policies and optimal time steps for coupling dynamic and static data for gas storage production and injection allocation. Dynamic data includes real-time monitored data such as pressure, flow, and temperature that changes over time, while static data includes relatively fixed data such as geological structure and equipment parameters. First, by rationally setting the time step, the dynamic and static data are organically integrated to enable the model to more accurately reflect the real-time operating status and long-term evolution trends of the gas storage. Then, when analyzing changes in gas reservoir pressure, the real-time monitored pressure data is combined with static data such as permeability and porosity in the geological model to more accurately predict future changes in gas reservoir pressure.

[0082] S52, Dynamic Data Update and Model Coupling: This method uses dynamic data analysis to automatically drive model update coupling. First, the real-time collected dynamic data is used to automatically trigger model update calculations, allowing the model to promptly adapt to changes in the gas storage reservoir's operating status. Then, when a well's gas flow rate changes abnormally, the new flow rate data is automatically input into the integrated coupling model, recalculating relevant parameters of the gas reservoir, wellbore, and pipeline network, and promptly adjusting the injection and production allocation plan to ensure stable system operation.

[0083] S53, production dynamic curve analysis and early warning: Through intelligent analysis of the formation method of production dynamic curve, intelligent statistics, comparison and analysis of production indicators are carried out; first, the curves of production parameters such as pressure, flow, and output changing over time are analyzed, and the characteristic parameters of the curves, such as peak value, valley value, change slope, etc., are extracted, and then compared with historical data and preset standard values ​​to realize abnormal early warning identification; when the change slope of a production curve exceeds the normal range, an early warning message will be automatically issued to remind management personnel of possible equipment failure or gas reservoir abnormality, and relevant comparison data will be promptly fed back to injection and production management personnel to enable them to make decisions.

[0084] The specific steps of S6 are as follows:

[0085] S61, Emergency Condition Setting: Establish emergency injection and production conditions that adapt to different scenarios, including triggering conditions for emergency injection and production in situations such as changes in dispatching instructions, wellbore outbursts (such as wellbore leaks and blockages), and abnormal gas reservoir production conditions (such as abnormal pressure fluctuations and sudden drops in gas volume). Specific triggering thresholds and judgment criteria are defined for each emergency situation. When the formation pressure of the gas storage reservoir drops below the preset safety threshold within a short period of time, the emergency gas production plan is automatically triggered. When a superior dispatching instruction is received to significantly increase the gas supply within a short period of time, the corresponding emergency production increase injection and production plan is initiated.

[0086] S62, Implementation Procedure Development: Develop detailed emergency injection and production implementation procedures, clarifying each link and operational steps from the triggering of the emergency event to the execution of the plan, including the activation of the emergency command system, the division of responsibilities among departments, the information transmission process, the equipment operation process, and safety measures. During the emergency gas production process, clearly define the sequence of gas well opening, gas production rate adjustment strategy, and emergency equipment switching operations to ensure that emergency injection and production work can be carried out in an orderly manner.

[0087] S63, Emergency Consequence Analysis and Rapid Response: Analyze the specific requirements of emergency production adjustment plans and realize rapid analysis of emergency consequences of gas storage facilities. Utilize the established integrated coupling model to quickly simulate the possible consequences of different emergency production adjustment plans after an emergency event occurs, such as changes in gas reservoir pressure, wellbore safety conditions, pipeline flow distribution, etc., to provide a scientific basis for decision-making. When faced with emergencies such as emergency peak regulation, ensure that a global emergency production adjustment plan is given within 30-60 minutes to meet sudden production adjustment needs and ensure the stable supply of natural gas and the safe operation of gas storage facilities.

[0088] The specific steps of S7 are as follows:

[0089] S71, Algorithm Fusion and Expert System Development: Based on historical gas storage data, this project utilizes big data and machine learning techniques to develop an intelligent control expert system using the Stacking algorithm fusion model. This system first organically integrates multiple machine learning algorithms to leverage their respective strengths and improve the model's predictive accuracy and adaptability. Then, through learning and analyzing historical data, the expert system simulates the decision-making process of domain experts, providing professional advice and guidance for intelligent injection and production allocation at gas storage facilities.

[0090] S72, iterative model training: The model is continuously iteratively trained using data accumulated over many years of operation at existing gas storage facilities. New historical data is regularly incorporated into the training set and the model is retrained, allowing it to continuously learn new patterns and characteristics during gas storage operation. Then, through data inference, the coefficients of the mechanism model are modified. For example, the permeability coefficient in the gas reservoir seepage model and the friction coefficient in the wellbore flow model are adjusted based on actual production data, making the model more realistic for practical applications.

[0091] S73, Parameter and Factor Optimization: During the continuous training and calculation process, we deeply explore the parameters and factors that affect the operation of the gas storage facility. We first use sensitivity analysis methods to determine the key parameters that have a greater impact on the injection and production allocation effect, and then focus on optimizing and adjusting the parameters. At the same time, we pay attention to emerging influencing factors, such as the impact of climate change on natural gas demand and the impact of new equipment on system operation. We promptly incorporate these into the model considerations, improve the system's computing speed and accuracy, and continuously optimize the intelligent injection and production allocation plan.

[0092] In the present invention:

[0093] Establishing the foundation of an intelligent injection-production model and optimizing workflows is the cornerstone of the entire system. It establishes constraints for each node, forms data links, and standard operating modes, providing the basic environment and regulatory framework for the operation of other components. Establishing an intelligent injection-production AI analysis model based on historical data relies on the data transmission requirements and operating modes provided by this foundation, enabling effective analysis of massive amounts of historical data and establishing optimal injection-production conditions for different production wells. At the same time, this foundation also provides a supporting environment and data links for the coupled calculations of various models in the construction and optimization of the injection-production matching model, ensuring the smooth construction of an integrated coupled model of "gas reservoir-wellbore-pipeline network."

[0094] Establishing an intelligent injection and production AI analysis model based on historical data provides data support and model reference for the construction and optimization of injection and production matching models. The optimal injection and production conditions under different constraints, derived through big data analysis, are an important basis for the construction and optimization of injection and production matching models. The results of the construction and optimization of the injection and production matching models can further enrich the learning samples of the AI ​​analysis model and enhance its analytical capabilities. In addition, the AI ​​analysis model also provides a reference for the optimal injection and production mode at the historical data level for the formulation of intelligent injection and production matching plans, helping to make the plans more scientific and reasonable.

[0095] The construction and optimization of the injection-production allocation model is the core link between the basic model and plan formulation. The resulting integrated coupling model can simulate and predict system dynamics under different injection-production schemes, providing critical simulation data and optimization directions for developing intelligent injection-production allocation plans. Furthermore, the model's calculation results provide analytical targets for developing intelligent injection-production indicator optimization analysis methods, facilitating in-depth research and optimization of production indicators.

[0096] The method for developing intelligent injection and production allocation plans integrates the results of intelligent injection and production model foundations, AI analysis models, and injection and production allocation models. The technical boundaries and plan development methods it defines provide a reference standard for conventional injection and production plans for triggering and executing emergency injection and production plans, enabling better coordination between emergency plans and conventional plans during development. Furthermore, the plans generated by this plan development method serve as the analysis targets for the intelligent injection and production indicator optimization analysis method. Through the analysis of indicators during plan execution, the plans are continuously refined.

[0097] An intelligent injection-production indicator optimization analysis method was established to monitor and provide feedback on the results of the previous sections. By analyzing production dynamic curves and providing early warning of anomalies, it can promptly identify issues within the intelligent injection-production model, the AI ​​analysis model, the injection-production matching model, and the injection-production matching plan. This information is then fed back to these components to provide a basis for optimization. For example, if an abnormal injection-production indicator is detected for a particular well, this may indicate that the AI ​​analysis model's analysis of the well's operating conditions is biased, necessitating relearning of historical data. It may also indicate that the injection-production matching model's simulation results differ from actual results, requiring further optimization.

[0098] Triggering and executing emergency injection and production plans and plans ensures the entire system's ability to respond to emergencies. Based on the models, plans, and analysis methods developed in the previous sections, it enables rapid response in emergencies. Furthermore, the execution data of emergency plans becomes part of historical data, providing new learning samples for intelligent injection and production AI analysis models, model optimization, and iteration based on historical data, thereby enhancing the system's ability to respond to similar emergencies.

[0099] Model optimization and iteration are integrated throughout every aspect of the system. It leverages new data generated during the operation of each component to continuously revise and optimize the foundation of the intelligent injection and production model, the AI ​​analysis model, and the injection and production allocation model. By inferring the data, the mechanism model coefficients are corrected to make each model more realistic. Continuous training of the AI ​​model improves its analytical accuracy. This iterative optimization enables the entire intelligent injection and production allocation system to continuously adapt to changes in the gas storage facility, continuously improving operational efficiency and safety.

[0100] Although the present invention has been described above with reference to embodiments, various modifications may be made thereto and equivalent components may be substituted without departing from the scope of the present invention. In particular, as long as there are no structural conflicts, the various features of the embodiments disclosed herein may be combined with each other in any manner, and the omission of an exhaustive description of such combinations in this specification is solely for the sake of space and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for intelligent injection and production allocation of a gas storage facility based on historical operation data, characterized in that: The specific steps are as follows: S1. Establish the foundation of the intelligent injection and production model and optimize the workflow: First, establish constraints for the intelligent injection and production scenario of underground gas storage. Then, based on the analysis of the coupled calculations of each model, build a data link that satisfies the constraints and feedback of the coupled calculations of different models. Finally, based on the model requirements, formulate a standard working mode for intelligent injection and production allocation. S2, building an intelligent injection and production AI analysis model based on historical data: First, collect and organize gas storage data and model simulation data, then conduct in-depth analysis of the data and build an intelligent injection and production AI analysis model based on the analysis results; S3, Injection and Production Allocation Model Construction and Optimization: First, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed based on the mechanism model and AI model. Then, the established integrated coupling model is used to simulate and predict the injection and production plan, and then the injection and production plan is optimized; S4, establish intelligent injection and procurement plans: First, clarify the technical boundaries for coordinated optimization between the formulation of injection and production plans and the AI ​​injection and procurement plans. Then, based on the determined technical boundaries, develop a compilation method for intelligent injection and procurement plans based on different gas storage capacity expansion and production stages. Then, based on the compilation method, realize the automatic output of intelligent injection and procurement plans. S5, establish an intelligent injection-production index optimization analysis method: first analyze the technical policy and optimal time step for coupling the dynamic data of gas storage production and injection allocation with static data, then analyze the coupling method of automatic driving model update by dynamic data. Finally, through the formation method of intelligent analysis of production dynamic curves, intelligent statistics, comparison, and analysis of production indicators are performed; S6, triggering and executing emergency injection and production plans: First, establish emergency injection and production conditions that are suitable for different scenarios. Then, formulate emergency injection and production execution procedures, clarifying each link and operational steps from emergency event triggering to plan execution. Finally, analyze the specific requirements of the emergency production adjustment plan to achieve rapid analysis of the emergency consequences of the gas storage. S7, model optimization and iteration: First, based on the historical data of the gas storage, we use big data combined with machine learning technology to develop an intelligent control expert system that integrates the model. Then, we use the data accumulated from years of operation of the existing gas storage to continuously iterate the model. Then, through continuous training and calculation, we deeply explore the parameters and factors that affect the operation of the gas storage.

2. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S1 are as follows: S11, build constraints: for the intelligent injection and production scenario of underground gas storage, establish detailed constraints for each node of geology, wellbore, ground station, and external transmission pipeline; S12, Optimize data links: Analyze the supporting environment and data transmission requirements for the coupled calculations of each model, and build data links that meet the constraints and feedback of the coupled calculations of different models; S13, establish a standard working mode: Based on the model requirements, formulate a complete set of standard working modes that meet the requirements of intelligent injection and procurement.

3. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S2 are as follows: S21, Data Collection and Collation: Collect development and research data of gas storage facilities, as well as simulation data of gas reservoirs, wellbore, and pipeline network models at different times. At the same time, collate a large amount of historical production data accumulated by different injection and production wells during multiple injection and production cycles, and perform pre-processing operations to ensure the accuracy, completeness, and consistency of the data. S22, Big Data Analysis: Use big data analysis technology to conduct in-depth analysis of processed data to establish a database of optimal injection and production conditions for different production wells under different constraints; S23, AI model construction and training: First, based on the results of big data analysis, a machine learning algorithm is used to build an intelligent injection and production AI analysis model; then, historical production data is used as training samples, various constraints are used as input features, and the optimal injection and production operating parameters are used as output labels. The model is repeatedly trained and optimized to improve the model's prediction accuracy and generalization ability by adjusting the model structure and parameter settings, so that it can accurately predict the optimal injection and production conditions under given constraints, providing decision support for intelligent injection and production allocation.

4. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S3 are as follows: S31, Establishing an integrated coupling model: Based on the mechanism model and AI model, combined with the material balance, seepage laws, and production constraints of the gas reservoir, an integrated coupling model of "gas reservoir-wellbore-pipeline network" is constructed; S32, Model Simulation and Prediction: Utilizing the established integrated coupling model, simulate the gas flow process in the gas reservoir, wellbore, and pipeline network under different injection and production schemes. This model predicts changes in gas reservoir pressure and saturation, wellbore pressure distribution and flow rate changes, and pressure and flow rate parameters at each node in the pipeline network. This model simulation compares the advantages and disadvantages of various schemes, providing a scientific basis for optimizing injection and production allocation schemes. S33, Scheme Optimization: Using optimization algorithms, the injection and production scheme is optimized with the goal of improving injection and production efficiency, increasing recovery rate, and ensuring safe production. During the optimization process, the constraints of the gas reservoir, wellbore, and pipeline network are used as optimization boundary conditions, and the decision variables of the injection and production wells are adjusted to seek the optimal combination of injection and production schemes. Through continuous iterative optimization, the gas storage facility can maximize its economic and social benefits while ensuring safe operation.

5. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S4 are as follows: S41, Determine Technical Boundaries: Clarify the technical boundaries for coordinated optimization between the injection and production allocation plan based on the twin model of geology, wellbore, and surface pipeline network, and the AI-based injection and production allocation plan based on historical data. This will allow the respective applicability and weighting of the two plans to be determined at different stages of gas storage capacity expansion and production, taking into account changes in gas reservoir geological characteristics, equipment operating status, and market demand fluctuations. S42, Scheme Preparation Method: Based on the determined technical boundaries, formulate a method for preparing intelligent injection and production allocation schemes based on different gas storage expansion and production stages. At the same time, the corresponding technical indicator requirements for intelligent production allocation are given. For each indicator, combining model prediction results and historical data experience, a reasonable value range and adjustment strategy are formulated. S43, plan output and automated control: Based on the compilation method, the automatic output of intelligent injection and procurement plans is realized; first, the model calculation results, optimization algorithm output and technical indicator requirements are integrated, and a detailed injection and procurement plan is generated with one click through a friendly human-computer interface; then a connection is established with the automated production allocation process equipment, and the production allocation instructions in the plan are automatically transmitted to the on-site equipment to realize the adjustment of the automated production allocation process conditions, reduce manual intervention, and improve the accuracy and timeliness of plan execution.

6. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S5 are as follows: S51, Coupling Dynamic and Static Data: Analyze the technical policies and optimal time steps for coupling dynamic and static data for gas storage production and injection allocation. Dynamic data includes time-varying data, while static data includes relatively fixed data. First, by rationally setting the time step, dynamic and static data are organically integrated to enable the model to more accurately reflect the real-time operating status and long-term evolution trends of the gas storage. Then, when analyzing reservoir pressure changes, the real-time monitored pressure data is combined with the static data in the geological model to more accurately predict future changes in reservoir pressure. S52, Dynamic Data Update and Model Coupling: This method uses dynamic data analysis to automatically drive model update coupling. First, the real-time collected dynamic data is used to automatically trigger model update calculations, allowing the model to promptly adapt to changes in the gas storage reservoir's operating status. Then, when a well's gas flow rate changes abnormally, the new flow rate data is automatically input into the integrated coupling model, recalculating relevant parameters of the gas reservoir, wellbore, and pipeline network, and promptly adjusting the injection and production allocation plan to ensure stable system operation. S53, production dynamic curve analysis and early warning: Through intelligent analysis of the formation method of production dynamic curve, intelligent statistics, comparison and analysis of production indicators are carried out; first, the curve of production parameters changing over time is analyzed, and the characteristic parameters of the curve are extracted. Then, the curve is compared with historical data and preset standard values ​​to realize abnormal early warning identification; when the slope of a production curve exceeds the normal range, an early warning message will be automatically issued to remind management personnel of possible equipment failure or gas reservoir abnormality, and relevant comparison data will be promptly fed back to injection and production management personnel to enable them to make decisions.

7. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S6 are as follows: S61, Emergency Condition Setting: Establish emergency injection and production conditions that adapt to different scenarios, including triggering conditions for emergency injection and production in the event of changes in dispatch instructions, wellbore outburst conditions, and abnormal gas reservoir production conditions. Specific triggering thresholds and judgment criteria are defined for each emergency situation. S62, Implementation Procedure Development: Develop detailed emergency injection and production implementation procedures, clarifying each link and operational steps from the triggering of the emergency event to the implementation of the plan, including the activation of the emergency command system, the division of responsibilities among departments, the information transmission process, the equipment operation process, and safety measures; S63, Emergency Consequences Analysis and Rapid Response: Analyze the specific requirements of emergency production adjustment plans and realize rapid analysis of emergency consequences of gas storage facilities; utilize the established integrated coupling model to quickly simulate the possible consequences of different emergency production adjustment plans after an emergency event occurs, providing a scientific basis for decision-making.

8. The intelligent injection and production method for a gas storage based on historical operation data according to claim 1 is characterized in that: The specific steps of S7 are as follows: S71, Algorithm Fusion and Expert System Development: Based on historical gas storage data, this project utilizes big data and machine learning techniques to develop an intelligent control expert system using the Stacking algorithm fusion model. This system first organically integrates multiple machine learning algorithms to leverage their respective strengths and improve the model's predictive accuracy and adaptability. Then, through learning and analyzing historical data, the expert system simulates the decision-making process of domain experts, providing professional advice and guidance for intelligent injection and production allocation at gas storage facilities. S72, iterative model training: Utilizing data accumulated over years of operation at existing gas storage facilities, the model is continuously iteratively trained. New historical data is regularly incorporated into the training set and the model is retrained, enabling it to continuously learn new patterns and characteristics of gas storage operation. The coefficients of the mechanism model are then modified through data inference, making the model more applicable to practical applications. S73, Parameter and Factor Optimization: During the continuous training and calculation process, we deeply explore the parameters and factors that affect the operation of the gas storage facility. We first use sensitivity analysis methods to determine the key parameters that have a greater impact on the injection and production allocation effect, and then focus on optimizing and adjusting the parameters. At the same time, we pay attention to newly emerging influencing factors and promptly incorporate them into the model consideration scope to improve the system's computing speed and accuracy, and continuously optimize the intelligent injection and production allocation plan.