Shale gas single well AI metering method and system
By combining oil and gas reservoir mechanism models with neural networks, an AI-based measurement method has been developed to address the issues of large measurement errors and high costs in shale gas single-well measurement. This approach achieves high-precision, low-cost measurement and improved operation and maintenance efficiency, providing accurate data support.
Patent Information
- Application Number
- CN202610557625.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-24
- Publication Date
- 2026-07-03
AI Technical Summary
Existing shale gas single-well metering technologies suffer from problems such as large metering errors, high equipment costs, reliance on manual maintenance, and unstable predictions, making it difficult to meet the demand for accurate metering.
By employing AI-based metering methods and combining oil and gas reservoir mechanism models with neural networks, a gas production trend prediction model is constructed through real-time data correction and error judgment. This enables closed-loop correction and intelligent rotation, reducing equipment dependence, improving metering accuracy, and lowering operation and maintenance costs.
It has achieved a significant reduction in the overall error of gas-liquid metering in a single well, reduced equipment investment costs, improved metering accuracy and operation and maintenance efficiency, and provided more accurate data support.
Smart Images
Figure CN122328094A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of oil and gas field development and intelligent metering technology, specifically relating to an AI metering method and system for single shale gas wells. Background Technology
[0002] In shale gas development, the accuracy of single-well metering data directly impacts key development decisions such as dynamic analysis, process selection optimization, and benefit assessment. Currently, the industry primarily uses traditional two-phase flow meters for metering. These are based on the differential pressure metering principle, generating a pressure difference through a throttling device, measuring the pressure difference with a differential pressure sensor, and calculating the gas and liquid flow rates using a built-in algorithm. However, they have significant limitations: 1. Large measurement error: The field conditions are complex, and the data is extremely inaccurate when the range is exceeded, and the production volume is basically meaningless. The allowable error under laboratory conditions is ±(5-10)%. In actual applications, the single well fluid measurement is divided according to the production volume stage, and the accuracy is between ±10% and ±30%, which is difficult to meet the requirements of accurate measurement.
[0003] 2. High equipment and maintenance costs: Traditional two-phase flow meters cost 150,000-200,000 RMB per unit, requiring huge investment for large-scale deployment; the equipment has a complex structure and is susceptible to problems such as water accumulation, blockage, and exceeding the range in the venturi tube, requiring regular manual calibration, and the work of sewage discharge and cleaning is difficult.
[0004] In addition, although some studies in recent years have attempted to use purely data-driven AI models for production forecasting, these methods rely entirely on historical data and lack physical mechanism constraints. When operating conditions change drastically or data quality is poor, the forecast results are unstable and have poor extrapolation, making it difficult to meet the reliability requirements of industrial-grade accurate measurement. Summary of the Invention
[0005] The purpose of this invention is to solve the problems of large errors, high costs and reliance on manual labor in existing metering technologies, and to propose an AI metering method and system for single wells of shale gas.
[0006] The technical solution of the present invention is as follows: Firstly, a shale gas single-well AI metering method, comprising the following steps: A single well is inserted into the test separator to obtain measured data on gas production and liquid production. The measured data on gas production and liquid production are used to determine the liquid production pattern of the single well and to correct the parameters of the oil and gas reservoir mechanism model. The corrected oil and gas reservoir mechanism model is then used to calculate the bottom hole flowing pressure and theoretical gas production. A gas production trend prediction model is constructed based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters. The single well is switched to the production separator for mixed transportation production. The wellhead operating parameters are collected in real time, and the real-time gas production is calculated using the corrected oil and gas reservoir mechanism model. The real-time liquid production is calculated based on the liquid production law and the real-time gas production. Then, the production trend is predicted using the gas production trend prediction model to obtain the predicted values of gas production and liquid production of a single well. Determine the production status of a single well and calculate the error level between the predicted gas production and liquid production of a single well and the measured values; The test rotation plan is automatically generated based on the production status and error level of each well, and the test priority and frequency of each well are adjusted. The new measured data obtained from the rotation test are used to iteratively update the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model, thus completing the AI measurement of shale gas single wells.
[0007] As a preferred option, the reservoir mechanism model includes the vertical pipe flow curve model and the inflow dynamic curve model.
[0008] As a preferred option, the gas production trend prediction model adopts a fully connected feedforward neural network, with bottom hole flowing pressure, theoretical gas production and real-time wellhead operating parameters as input features, and the measured gas production of the test separator as the training target. The feedforward neural network is iteratively updated through an optimization mechanism that minimizes errors.
[0009] As a preferred method, the production status of a single well is determined by a three-pressure parameter fusion method, based on casing pressure, oil pressure, transmission pressure and oil-casing pressure difference; the production status of a single well includes four types: flowing well, potential liquid accumulation well, liquid accumulation well and abnormal fluctuation well.
[0010] As a preferred method, the specific method for determining the production status of a single well is as follows: Flowing wells: The fluctuation range of casing pressure, oil pressure and oil-casing pressure difference is ≤±5% / h, and there is no continuous trend of increase or decrease; Potential liquid accumulation wells: oil pressure drop rate ≥ 5% / h, oil-casing pressure difference rise rate ≥ 5% / h, and casing pressure fluctuation ≤ ±3% / h; For wells with accumulating liquid: the rate of decrease in oil pressure is ≥10% / h, the rate of increase in oil-casing pressure differential is ≥10% / h, and the casing pressure fluctuation is >±3% / h; Abnormal fluctuation well: Any pressure parameter or oil casing pressure differential change rate ≥ ±15% / h.
[0011] As a preferred option, the specific rules for classifying error levels are as follows: Low error: E_avg < ±5%, where E_avg represents the average relative error between the most recent N measured data and the predicted values of the gas production trend prediction model; Mean error: ±5%≤E_avg≤±10%; High error: E_avg > ±10%.
[0012] As a preferred method, the adjustment method for the testing priority and frequency of a single well is as follows: Wells with accumulating fluid and wells with abnormal fluctuations are classified as Level 1, the highest priority, and the basic testing frequency is once every 24 hours. Potentially liquid-accumulating wells with a high error level are classified as Level 1 priority, and the basic testing frequency is once every 36 hours. Potentially liquid-accumulating wells with an error level of medium or low are classified as Level 2 priority, and the basic testing frequency is once every 48 hours. For self-flowing wells with a high error level (priority level 2), the basic testing frequency is once every 72 hours. For wells that are self-flowing and have an error level of medium error and a priority of level 3, the basic testing frequency is once every 96 hours; For wells that are self-flowing and have a low error level (level 4, lowest priority), the basic testing frequency is once every 168 hours. When testing resources are limited, time windows are allocated according to priority level 1 > level 2 > level 3 > level 4. For wells with the same priority, the wells with higher error levels are prioritized. For level 1 priority wells, the scheduling order is: wells with accumulated fluid > wells with potential accumulated fluid > wells with abnormal fluctuations.
[0013] The beneficial effects of this invention are: 1. Significantly improved accuracy: By using a closed-loop correction and mechanism AI fusion model, the overall error of single-well gas-liquid metering is stably controlled within ±5% from ±10%~30% in the traditional method.
[0014] 2. Significantly reduced costs: By utilizing existing site separators and AI software cores, expensive dedicated two-phase flow meters can be replaced, reducing the investment in single-well metering equipment by more than 60%.
[0015] 3. Intelligent operation and maintenance: Intelligent rotation and automatic model correction reduce the workload of manual calibration, cleaning and maintenance, and reduce the difficulty and cost of operation and maintenance.
[0016] 4. Enhanced decision support: The provided data on bottom hole flowing pressure and production capacity forecasts are more accurate and reliable, providing a solid data foundation for dynamic analysis, production optimization, and benefit evaluation of shale gas fields.
[0017] Secondly, a shale gas single-well AI metering system includes: The first module is used to cut a single well into the test separator, obtain measured data of gas production and liquid production, use the measured data of gas production and liquid production to determine the liquid production law of a single well and correct the parameters of the oil and gas reservoir mechanism model, and use the corrected oil and gas reservoir mechanism model to calculate the bottom hole flowing pressure and theoretical gas production. The second module is used to construct a gas production trend prediction model based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters. The third module is used to switch a single well to the production separator for mixed transportation production, collect wellhead operating parameters in real time, calculate real-time gas production using the corrected oil and gas reservoir mechanism model, calculate real-time liquid production based on the liquid production law and real-time gas production, and then use the gas production trend prediction model to predict the production trend, thereby obtaining the predicted values of single-well gas production and liquid production. The fourth module is used to determine the production status of a single well and to calculate the error level between the predicted gas production and liquid production of a single well and the measured values. The fifth module is used to automatically generate a test rotation plan based on the production status and error level of a single well, adjust the test priority and frequency of a single well, and use the new measured data obtained from the rotation test to iteratively update the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model, thus completing the AI measurement of shale gas single wells.
[0018] Thirdly, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the method as described in the first aspect.
[0019] Fourthly, a non-transitory computer-readable storage medium is provided that stores computer instructions for causing a computer to perform the method as described in the first aspect. Attached Figure Description
[0020] Figure 1 The diagram shows a flowchart of an AI metering method for a single shale gas well.
[0021] Figure 2 The diagram shown is an architecture diagram of the real-time closed-loop system provided in Example 2.
[0022] Figure 3 The diagram shown is a schematic diagram of the gas production rate provided in Example 2. Detailed Implementation
[0023] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the invention, and are not intended to limit the scope of the invention.
[0024] Example 1: like Figure 1 As shown, a single-well AI metering method for shale gas includes the following steps: S1. Connect the single well to the test separator, obtain the measured data of gas production and liquid production, use the measured data of gas production and liquid production to determine the liquid production law of the single well and correct the parameters of the oil and gas reservoir mechanism model, and use the corrected oil and gas reservoir mechanism model to calculate the bottom hole flowing pressure and theoretical gas production. S2. Construct a gas production trend prediction model based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters; S3. Switch the single well to the production separator for mixed transportation production, collect the wellhead operating parameters in real time, and use the corrected oil and gas reservoir mechanism model to calculate the real-time gas production. Calculate the real-time liquid production based on the liquid production law and the real-time gas production. Then use the gas production trend prediction model to predict the production trend and obtain the predicted values of single-well gas production and liquid production. S4. Determine the production status of a single well and calculate the error level between the predicted gas production and liquid production of a single well and the measured values. S5. Automatically generate test rotation plans based on the production status and error level of individual wells, adjust the test priority and frequency of individual wells; and use the new measured data obtained from the rotation tests to iteratively update the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model, and complete the AI metering of shale gas individual wells.
[0025] In this embodiment, measured gas production and liquid production data are used to correct the parameters of the oil and gas reservoir mechanism model: key parameters such as formation pressure are corrected by methods such as the gas reservoir material balance equation; an accurate vertical pipe flow (VLP) curve is established by combining the wellbore multiphase flow model; and parameters are inverted and corrected for the inflow dynamics (IPR) curve model describing the formation's liquid supply capacity using measured production data.
[0026] In this embodiment, the gas production trend prediction model integrating mechanism and artificial intelligence is constructed using a neural network algorithm as follows: Based on the calibrated reservoir mechanism model (IPR / VLP), key state parameters such as bottom hole flowing pressure and theoretical gas production under corresponding operating conditions are calculated to provide a physical constraint basis for fusion modeling. On this basis, a data-driven modeling method is introduced to learn and correct the deviation relationship between the mechanism model and actual production data, thereby establishing a gas production trend prediction model capable of characterizing complex nonlinear features.
[0027] The neural network algorithm uses the calculation results of the mechanism model and the real-time wellhead operating parameters as input features, and the measured gas production of the test separator as the training target. Through the optimization mechanism of minimizing error, the model parameters are iteratively updated to gradually approximate the actual production change law, thereby constructing a gas production trend prediction model with high prediction accuracy and operating condition adaptability.
[0028] The neural network is a feedforward neural network with a fully connected structure. The functions and parameters of each layer node are as follows: Input Layer: The input layer receives multi-dimensional operational features, including key state parameters calculated from the reservoir mechanism model and real-time wellhead operational variables. Before model training, the input features are processed to achieve a uniform scale to reduce the impact of different physical dimensions on model convergence.
[0029] Hidden layers: Hidden layers are used to extract higher-order nonlinear relationships between input variables, enhancing the model's ability to express complex coupled conditions through nonlinear activation mechanisms. The number of hidden layers and the size of the nodes can be adaptively configured according to the data scale and prediction accuracy requirements.
[0030] Output layer: The output layer is used to generate the target gas production prediction results and maintain the numerical consistency between the output results and the actual physical quantities.
[0031] Through the above structural design, the fusion model retains the physical constraint characteristics of the mechanistic model, and further introduces a data-driven method to compensate for nonlinear deviations, thereby improving the accuracy and stability of gas production trend prediction under variable operating conditions.
[0032] In this embodiment, the predicted gas production and liquid production of a single well are calculated using a corrected reservoir mechanism model and a gas production trend prediction model. Specifically, the VLP model is updated based on the new production backpressure, and combined with the corrected IPR model, gas production is calculated in real time using nodal analysis. Liquid production is calculated based on the liquid production patterns determined in step S1 (such as gas-liquid ratio and temperature-liquid volume relationship). The gas production trend prediction model is simultaneously invoked to predict production trends, and the results are cross-validated with the calculation results from the mechanism model.
[0033] In this embodiment, the method for determining the production status of a single well based on real-time pressure data and calculating the error level between the model prediction value and the recent measured value is as follows: Based on real-time pressure data (such as oil pressure, casing pressure, and transmission pressure) and their changing trends, the system automatically determines whether a single well is in a state of "self-flowing," "potential fluid accumulation," "fluid accumulation," or "abnormal fluctuation." The production status of a single well is determined using a "three-pressure parameter fusion judgment method," which is based solely on casing pressure (P_casing), oil pressure (P_oil), transmission pressure (P_trans), and the derived oil-casing pressure difference (ΔP_oil-casing=P_oil-P_casing). The determination is made through the stability of a single parameter or the rate of change of a combination of parameters, as shown in Table 1.
[0034] Table 1 Criteria for Determining the Production Status of a Single Well
[0035] Simultaneously, the error between the model's predicted values and recent measured values is calculated and categorized into "high," "medium," and "low" levels. The prediction error level classification uses the "average relative error between the three most recent measured data (test separator) and the model's predicted data" as the core evaluation indicator (E_avg), combined with the single maximum error (E_max) to classify the levels, as shown in Table 2.
[0036] Table 2. Classification of Prediction Error Levels and Threshold Standards
[0037] In this embodiment, step S5 specifically involves: automatically generating a test rotation plan based on the above evaluation results. The test rotation plan adopts a "priority-frequency" dual-dimensional scheduling strategy, combining the production status and error level determined by the unified three pressure parameters to determine the single-well test plan. The basic principle is: for wells with unstable production (such as wells with fluid accumulation) or high metering errors, the test priority and frequency are increased; for wells with stable production and low errors, the test frequency is reduced. Wellhead switching is automatically executed through an intelligent diversion skid. The rules for test frequency and priority arrangement are shown in Table 3.
[0038] Table 3. Test Frequency and Priority Arrangement Rules
[0039] Priority execution logic: When testing resources are limited, time windows are allocated according to level 1 > level 2 > level 3 > level 4; for wells with the same priority, the wells with higher errors are prioritized, and level 1 priority is scheduled according to "liquid accumulation wells > potential liquid accumulation wells > abnormal fluctuation wells" to ensure the measurement accuracy of high-risk wells.
[0040] Finally, using the new measured data obtained from the rotation test, the mechanism model parameters and AI prediction model are iteratively corrected and trained again, so that the entire system can continuously improve its accuracy as it adapts to changes in working conditions.
[0041] Example 2: Based on Example 1, this embodiment of the invention takes four wells of a shale gas platform as an example to provide a detailed description of the AI metering method for single shale gas wells proposed in this invention.
[0042] This invention employs a "cloud-edge-device architecture" to construct a real-time closed-loop system, such as... Figure 2As shown, wellhead data acquisition equipment, including pressure sensors, temperature sensors, and separator liquid level sensors, is deployed on the shale gas well platform to collect key production data such as single-well pressure, temperature, and separator liquid level in real time. Edge computing nodes are deployed on-site to complete data preprocessing and control logic distribution. An intelligent diversion skid system and test and production separators are configured to achieve automatic switching of flow direction for single wells. An AI metering core module and data management platform are deployed in the cloud, integrating historical production data, model training tools (including neural network training modules), and visualization monitoring functions, supporting the training and optimization of IPR / VLP curve correction, gas / liquid production prediction models, and gas production trend prediction models. The system works collaboratively from the cloud to the edge: the edge computing module is responsible for real-time data preprocessing, rapid model inference, and control command distribution, ensuring low-latency response; the cloud data processing module aggregates historical and real-time data from each well, performs large-scale, in-depth model training and strategy optimization, and distributes the updated model parameters to the edge side. This architecture ensures both real-time computation and continuous model evolution.
[0043] First, initial configuration is performed: input the original geological reserves, wellbore structure, initial formation pressure and other basic parameters of the four wells, deploy various sensors and edge computing nodes, and complete the initial training of the mechanism model and prediction model.
[0044] Actual measurement correction: In accordance with the intelligent rotation system, potential liquid accumulation wells are given priority for testing separator metering, collecting pressure and temperature data, calculating formation pressure through the gas reservoir material balance equation, and obtaining the actual liquid production and gas-liquid ratio by combining magnetic float liquid level data, and correcting the IPR / VLP curve; a neural network algorithm is used to learn the correlation between the actual gas production and the calculation results of the mechanism model, pressure, and temperature, to complete the training of the gas production trend prediction model and establish a liquid volume prediction model.
[0045] Mixed transport metering: The calibrated single well is switched to the production separator for mixed transport, and back pressure, pressure and temperature data are collected in real time. The VLP model is updated, and the gas production and liquid production are calculated and output. The gas production trend prediction model is used to dynamically predict the gas production change characteristics and cross-validate with the metering results.
[0046] Rotation verification: Based on well condition classification and error assessment, the testing frequency is increased for flowing wells with high errors, while low-frequency sampling is used for wells with stable processes and low errors; such as Figure 3 As shown in Table 4, by rotating the model multiple times and combining the measured data to simultaneously optimize the liquid volume prediction model and the gas production trend prediction model, the measurement error of the four wells was gradually controlled to within 5%.
[0047] Table 4. Gas production and calculation error calculated by the model before and after algorithm optimization following multiple rounds of iteration.
[0048] System operation and maintenance: Real-time monitoring of metering data, model status and gas production trend prediction results through cloud platform, and regular input of actual pressure measurement data to correct the model, ensuring long-term stable operation.
[0049] Example 3: Based on Example 1, this embodiment of the invention provides a shale gas single-well AI metering system, which can be used to implement the shale gas single-well AI metering method as described in the foregoing embodiments. The system includes: The first module is used to cut a single well into the test separator, obtain measured data of gas production and liquid production, use the measured data of gas production and liquid production to determine the liquid production law of a single well and correct the parameters of the oil and gas reservoir mechanism model, and use the corrected oil and gas reservoir mechanism model to calculate the bottom hole flowing pressure and theoretical gas production. The second module is used to construct a gas production trend prediction model based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters. The third module is used to switch a single well to the production separator for mixed transportation production, collect wellhead operating parameters in real time, calculate real-time gas production using the corrected oil and gas reservoir mechanism model, calculate real-time liquid production based on the liquid production law and real-time gas production, and then use the gas production trend prediction model to predict the production trend, thereby obtaining the predicted values of single-well gas production and liquid production. The fourth module is used to determine the production status of a single well and to calculate the error level between the predicted gas production and liquid production of a single well and the measured values. The fifth module is used to automatically generate a test rotation plan based on the production status and error level of a single well, adjust the test priority and frequency of a single well, and use the new measured data obtained from the rotation test to iteratively update the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model, thus completing the AI measurement of shale gas single wells.
[0050] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.
[0051] In an exemplary embodiment, the electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the shale gas single-well AI metering method as described in Embodiment 1 above.
[0052] In an exemplary embodiment, the readable storage medium may be a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the shale gas single-well AI metering method according to Embodiment 1 above.
[0053] In an exemplary embodiment, the computer program product includes a computer program that, when executed by a processor, implements the shale gas single-well AI metering method according to Embodiment 1 above.
[0054] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0055] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0056] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0057] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.
[0058] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.
[0059] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.
Claims
1. A single-well AI metering method for shale gas, characterized in that, Includes the following steps: A single well is inserted into the test separator to obtain measured data on gas production and liquid production. The measured data on gas production and liquid production are used to determine the liquid production pattern of the single well and to correct the parameters of the oil and gas reservoir mechanism model. The corrected oil and gas reservoir mechanism model is then used to calculate the bottom hole flowing pressure and theoretical gas production. A gas production trend prediction model is constructed based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters. The single well is switched to the production separator for mixed transportation production. The wellhead operating parameters are collected in real time, and the real-time gas production is calculated using the corrected oil and gas reservoir mechanism model. The real-time liquid production is calculated based on the liquid production law and the real-time gas production. Then, the production trend is predicted using the gas production trend prediction model to obtain the predicted values of gas production and liquid production of a single well. Determine the production status of a single well and calculate the error level between the predicted gas production and liquid production of a single well and the measured values; The test rotation plan is automatically generated based on the production status and error level of each well, and the test priority and frequency of each well are adjusted accordingly. Furthermore, by utilizing the new measured data obtained from rotation testing, the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model are iteratively updated to complete the AI measurement of shale gas single wells.
2. The shale gas single-well AI metering method according to claim 1, characterized in that, The reservoir mechanism model includes the vertical pipe flow curve model and the inflow dynamic curve model.
3. The shale gas single-well AI metering method according to claim 2, characterized in that, The gas production trend prediction model adopts a fully connected feedforward neural network, with bottom hole flowing pressure, theoretical gas production and real-time wellhead operating parameters as input features, and the measured gas production of the test separator as the training target. The feedforward neural network is iteratively updated through an optimization mechanism that minimizes errors.
4. The shale gas single-well AI metering method according to claim 1, characterized in that, The production status of a single well is determined by a three-pressure parameter fusion method, based on casing pressure, oil pressure, transmission pressure, and oil-casing pressure difference. The production status of a single well includes four types: flowing well, potential liquid accumulation well, liquid accumulation well, and abnormal fluctuation well.
5. The shale gas single-well AI metering method according to claim 4, characterized in that, The specific method for determining the production status of a single well is as follows: Flowing wells: The fluctuation range of casing pressure, oil pressure and oil-casing pressure difference is ≤±5% / h, and there is no continuous trend of increase or decrease; Potential liquid accumulation wells: oil pressure drop rate ≥ 5% / h, oil-casing pressure difference rise rate ≥ 5% / h, and casing pressure fluctuation ≤ ±3% / h; For wells with accumulating liquid: the rate of decrease in oil pressure is ≥10% / h, the rate of increase in oil-casing pressure difference is ≥10% / h, and the casing pressure fluctuation is >±3% / h; Abnormal fluctuation well: Any pressure parameter or oil casing pressure differential change rate ≥ ±15% / h.
6. The shale gas single-well AI metering method according to claim 5, characterized in that, The specific rules for classifying error levels are as follows: Low error: E_avg < ±5%, where E_avg represents the average relative error between the most recent N measured data and the predicted values of the gas production trend prediction model; Mean error: ±5%≤E_avg≤±10%; High error: E_avg > ±10%.
7. The shale gas single-well AI metering method according to claim 6, characterized in that, The method for adjusting the testing priority and frequency of a single well is as follows: Wells with accumulating fluid and wells with abnormal fluctuations are classified as Level 1, the highest priority, and the basic testing frequency is once every 24 hours. Potentially liquid-accumulating wells with a high error level are classified as Level 1 priority, and the basic testing frequency is once every 36 hours. Potentially liquid-accumulating wells with an error level of medium or low are classified as Level 2 priority, and the basic testing frequency is once every 48 hours. For self-flowing wells with a high error level (priority level 2), the basic testing frequency is once every 72 hours. For wells that are self-flowing and have an error level of medium error and a priority of level 3, the basic testing frequency is once every 96 hours; For wells that are self-flowing and have a low error level (level 4, lowest priority), the basic testing frequency is once every 168 hours. When testing resources are limited, time windows are allocated according to priority level 1 > level 2 > level 3 > level 4. For wells with the same priority, the wells with higher error levels are prioritized. For level 1 priority wells, the scheduling order is: wells with accumulated fluid > wells with potential accumulated fluid > wells with abnormal fluctuations.
8. A shale gas single-well AI metering system, characterized in that, include: The first module is used to cut a single well into the test separator, obtain measured data of gas production and liquid production, use the measured data of gas production and liquid production to determine the liquid production law of a single well and correct the parameters of the oil and gas reservoir mechanism model, and use the corrected oil and gas reservoir mechanism model to calculate the bottom hole flowing pressure and theoretical gas production. The second module is used to construct a gas production trend prediction model based on bottom hole flowing pressure, theoretical gas production, and real-time collected wellhead operating parameters. The third module is used to switch a single well to the production separator for mixed transportation production, collect wellhead operating parameters in real time, calculate real-time gas production using the corrected oil and gas reservoir mechanism model, calculate real-time liquid production based on the liquid production law and real-time gas production, and then use the gas production trend prediction model to predict the production trend, thereby obtaining the predicted values of single-well gas production and liquid production. The fourth module is used to determine the production status of a single well and to calculate the error level between the predicted gas production and liquid production of a single well and the measured values. The fifth module is used to automatically generate a test rotation plan based on the production status and error level of a single well, and adjust the test priority and frequency of a single well. Furthermore, by utilizing the new measured data obtained from rotation testing, the parameters of the oil and gas reservoir mechanism model and the gas production trend prediction model are iteratively updated to complete the AI measurement of shale gas single wells.
9. An electronic device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the method according to any one of claims 1-7.
10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-7.