Prediction model processing method and oil well yield calculation method

Abstract

The invention relates to the technical field of oil exploration, in particular to a prediction model processing method and an oil well yield calculation method.The method comprises the steps that well testing data of a target well and measured data of multiphase flow corresponding to the well testing data are received; when the target well does not have the corresponding flow calculation network model, the well testing data are substituted into a preset mathematical calculation model for prediction, and prediction data of the multiphase flow are obtained; when the difference value between the measured data and the predicted data of the multiphase flow does not meet the difference value condition, correcting the mathematical calculation model to obtain a first correction calculation model; generating a training set by using the first correction calculation model to train a neural network model so as to obtain a flow calculation network model of the target well; and sending the flow calculation network model of the target well to the edge calculation equipment. According to the method, real-time prediction can be carried out more accurately and timely based on well data collected on a production site, and instant feedback is provided for production decision making.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of oil exploration, and in particular to a processing method of a prediction model and a calculation method of oil well production. BACKGROUND

[0002] In oil and gas production, multiphase flow is one of the most important measurement data at the wellhead, which is crucial for production allocation, flow assurance and production monitoring. The commonly used multiphase flow monitoring method at present is to use virtual flow metering technology. As a digital measurement technology, the core principle of virtual metering is to obtain pressure and temperature information on the wellbore, Christmas tree, manifold and pipeline through ordinary communication protocols and existing measurement hardware, to comprehensively obtain characteristic parameters of the target oil and gas field fluid, to establish a multiphase flow numerical calculation model of the target oil and gas field multiphase fluid in the wellbore, Christmas tree and pipeline, and to obtain the single-phase production of each oil and gas well by solving the mass, momentum and energy equations.

[0003] The oil and gas well calculation model in the existing commercial calculation software considers fluid dynamics, thermodynamics, phase change and other aspects, and therefore has high calculation accuracy. Since the mathematical model in the commercial software is complex and the calculation amount is large, the requirement for computing power is high, and a high-performance computer or server is needed to run the software to meet the calculation requirements. Therefore, the commercial software is not suitable for direct deployment to the field.

[0004] However, multiphase flow is dynamically changing in industrial production, and real-time prediction data of multiphase flow is crucial for adjusting production parameters and optimizing process flow in oilfield exploitation and other scenarios, and can provide immediate feedback for production decisions. The existing commercial software is difficult to achieve real-time prediction of multiphase flow based on field production data, and it is difficult to meet the production requirements in terms of prediction accuracy when deploying a relatively low-computing-power-demanding prediction model such as a neural network model to the production field. SUMMARY

[0005] The present disclosure provides a processing method of a prediction model and a calculation method of oil well production to solve the problem that the existing commercial calculation model is difficult to be deployed to the production edge, and the application of a relatively low-computing-power-demanding prediction model to the edge is not accurate enough.

[0006] In a first aspect, the present disclosure provides a processing method of a prediction model, which is applied to an edge cloud system in a cloud-edge collaborative system, wherein the cloud-edge collaborative system comprises the edge cloud system and at least one edge computing device in communication with the edge cloud system, and the method comprises:

[0007] receiving well test data of a target well and measured data of multiphase flow corresponding to the well test data;

[0008] determining whether the target well has a corresponding flow calculation network model;

[0009] when the target well does not have a corresponding flow calculation network model, substituting the well testing data into a preset mathematical calculation model for prediction to obtain predicted data of the multiphase flow rate;

[0010] determining whether a difference between the measured data of the multiphase flow rate and the predicted data satisfies a preset difference condition;

[0011] when the difference does not satisfy the difference condition, correcting the mathematical calculation model so that the difference satisfies the difference condition, thereby obtaining a first corrected calculation model;

[0012] training a neural network model using the well testing data and first corrected data of the multiphase flow rate obtained according to the first corrected calculation model, to obtain a flow calculation network model of the target well;

[0013] sending the flow calculation network model of the target well to the edge computing device.

[0014] In an embodiment, after the step of determining whether the target well has a corresponding flow calculation network model, the method further comprises:

[0015] when the target well has a corresponding flow calculation network model, receiving calculation data of the multiphase flow rate sent by the edge computing device, wherein the calculation data of the multiphase flow rate is a prediction result obtained by inputting the well testing data into the flow calculation network model for prediction;

[0016] determining whether a difference between the measured data of the multiphase flow rate and the calculation data satisfies the difference condition;

[0017] when the difference does not satisfy the difference condition, correcting the first corrected calculation model so that a difference between the first corrected data of the multiphase flow rate obtained according to the first corrected calculation model and the measured data satisfies the difference condition, thereby obtaining a second corrected calculation model;

[0018] training the flow calculation network model using the well testing data and second corrected data of the multiphase flow rate obtained according to the second corrected calculation model, so that the flow calculation network model satisfies a preset model condition, thereby obtaining an updated flow calculation network model of the target well;

[0019] sending the updated flow calculation network model to the edge computing device.

[0020] In an embodiment, the step of sending the updated flow calculation network model to the edge computing device further comprises:

[0021] receiving a receipt information generated by the edge computing device replacing the flow calculation network model stored on the edge computing device with the updated flow calculation network model.

[0022] In an embodiment, the step of correcting the mathematical calculation model to make the difference satisfy the difference condition to obtain a first corrected calculation model comprises:

[0023] obtaining a preset value range corresponding to the volume content in the mathematical calculation model;

[0024] adjusting the volume content in the preset value range to obtain a temporary mathematical model;

[0025] inputting the well test data into the temporary mathematical model for prediction to obtain temporary prediction data of the multiphase flow;

[0026] When the difference between the temporary prediction data of the multiphase flow and the measured data satisfies the difference condition, the temporary mathematical model is determined as the first corrected calculation model.

[0027] In an embodiment, the step of receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data further comprises:

[0028] In response to a connection request of the edge computing device, creating a data object in a preset data model, wherein the preset data model is a standardized data structure for realizing data exchange between different edge computing devices and edge cloud systems, and the data object is configured to receive raw data and processed data from the target well,

[0029] The step of receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data comprises:

[0030] receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data, and storing the well test data and the measured data of the multiphase flow in corresponding data objects respectively.

[0031] In a second aspect, the present disclosure provides an oil well production calculation method, comprising:

[0032] obtaining an updated neural network model by using the processing method of the prediction model according to any one of the above embodiments;

[0033] The wellhead real-time measurement data of the target well is input into the updated neural network model on the edge computing device for prediction to obtain real-time prediction data of the multiphase flow rate;

[0034] The oil well production of the target well is calculated by using the real-time prediction data of the multiphase flow rate.

[0035] In a third aspect, the present disclosure provides an oil well production calculation system, comprising a data service module, a business flow module, an edge computing module and a visualization module, wherein:

[0036] The data service module is configured to receive well testing data of a target well and measured data of a multiphase flow rate corresponding to the well testing data;

[0037] The business flow module is configured to detect whether there is a corresponding flow rate calculation network model for the target well; when there is no corresponding flow rate calculation network model for the target well, the well testing data is substituted into a preset mathematical calculation model for prediction to obtain prediction data of the multiphase flow rate; it is judged whether the difference between the measured data and the prediction data of the multiphase flow rate satisfies a preset difference condition; when the difference does not satisfy the difference condition, the mathematical calculation model is corrected to make the difference satisfy the difference condition, thereby obtaining a first corrected calculation model; a neural network model is trained by using the well testing data and corrected data of the multiphase flow rate obtained according to the first corrected calculation model to obtain a flow rate calculation network model for the target well; and the flow rate calculation network model for the target well is sent to the edge computing device;

[0038] The edge computing module is configured to receive and install the flow rate calculation network model, receive wellhead real-time measurement data of the target well and input the wellhead real-time measurement data into the flow rate calculation network model for prediction to obtain real-time prediction data of the multiphase flow rate;

[0039] The visualization module is configured to display the wellhead real-time measurement data and the real-time prediction data of the multiphase flow rate.

[0040] In a fourth aspect, the present disclosure provides a computer device, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method of the above aspect.

[0041] In a fifth aspect, the present disclosure provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of the method of the above aspect.

[0042] In a sixth aspect, the present disclosure provides a computer program product, comprising a computer program / instruction, wherein the computer program is executed by a processor to implement the steps of the method of the above aspect.

[0043] The processing method of the prediction model and the oil well production calculation method provided by the present disclosure can find out whether the mathematical calculation model deviates by comparing the first predicted multiphase flow rate obtained by prediction with the reference multiphase flow rate, correct the mathematical calculation model if deviation occurs, train the neural network model by using the corrected mathematical model and the measured data, and finally make the updated neural network model have a significant improvement in prediction accuracy. The updated neural network is installed in the edge computing device, so that real-time prediction based on well data collected in the production site can be more accurate and timely, and immediate feedback can be provided for production decision-making. In addition, model training and updating are performed in the edge cloud system, which can significantly reduce the delay and cost of data transmission, improve response speed, make the upgrade and iteration of the neural network model more timely, and improve the reliability of the real-time prediction result of the multiphase flow rate.

[0044] The method of the present disclosure uses a mathematical calculation model (an oil and gas well calculation model in existing commercial calculation software) to detect whether the prediction accuracy of the neural network model on the current edge computing device meets the standard, so that the neural network model can be tested and corrected for accuracy every time the test production is performed. The cooperation of the edge cloud system and the edge computing device can timely predict the multiphase flow rate based on the well data collected during the production process, which can provide immediate feedback for production decision-making and improve the accuracy and reliability of the real-time prediction result. BRIEF DESCRIPTION OF DRAWINGS

[0045] The present disclosure will be described in more detail below based on the embodiments and with reference to the accompanying drawings:

[0046] Figure 1 A flowchart of a processing method of a prediction model provided by an embodiment of the present disclosure;

[0047] Figure 2 An application scenario diagram of a cloud-edge collaborative system provided by an embodiment of the present disclosure;

[0048] Figure 3 A flowchart of another processing method of a prediction model provided by an embodiment of the present disclosure;

[0049] Figure 4 An internal block diagram of a processing device of a prediction model provided by an embodiment of the present disclosure;

[0050] Figure 5 An internal module diagram of a system for calculating oil well production provided by an embodiment of the present disclosure.

[0051] In the drawings, the same components are designated by the same reference numerals, and the drawings are not drawn according to the actual scale. Detailed Implementation

[0052] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0053] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0054] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0055] Example One

[0056] Figure 1 This is a flowchart illustrating a method for processing a prediction model provided in an embodiment of this disclosure. Figure 1 As shown, a method for processing a prediction model is applied to an edge cloud system in a cloud-edge collaborative system, wherein the cloud-edge collaborative system includes an edge cloud system and at least one edge computing device communicating with the edge cloud system. The method includes:

[0057] Step 110: Receive the well test data of the target well and the measured data of the multiphase flow rate corresponding to the well test data;

[0058] Well testing is typically conducted during the early exploration and development stages or specific assessment phases of oil or gas wells. Well test data usually includes parameters such as the relative changes in pressure, oil production, gas production, and water cut, as well as temperature of the target well. Analyzing well test data serves purposes such as monitoring the well's production status, determining the hydrodynamic parameters of the production layer, analyzing reservoir dynamics, and making production predictions.

[0059] Multiphase flow rate refers to the individual flow rates of oil, gas, and water in an oil well. By acquiring changes in multiphase flow rate, changes in oil well production can be detected, providing strong support for production decisions. The reference multiphase flow rate can be measured using a test separator that separates oil, water, and gas, and compared with the multiphase flow rate predicted by the predictive model. Test separators can provide the volume of each fluid within 24 hours, but due to their often large volume, heavy weight, and the time and resources required for each test, they are not suitable for routine monitoring. Since well testing is relatively infrequent, the reference multiphase flow rate can be obtained during well testing using a test separator. This reference flow rate can then be compared with the multiphase flow rate prediction values ​​from the subsequent multiphase flow prediction model to determine if model correction is needed.

[0060] In this embodiment, as Figure 2 The diagram illustrates an application scenario of a cloud-edge collaborative system for measuring oil well production. The system includes an edge cloud system 210 and at least one edge computing device 220 communicating with it. The edge cloud system collects well test data and / or oil, gas, and water metering data from the target well, updates the prediction model, and sends the prediction model to the corresponding edge computing devices. One edge computing device is deployed for each well. The edge computing device collects data from periodic measurements by various instruments and inputs relevant data for predicting multiphase flow rate into the prediction model for prediction.

[0061] Step 120: Detect whether the target well already has a corresponding flow calculation network model;

[0062] In this embodiment, the traffic calculation network model is a big data prediction model trained for the target well. Its computing power requirements are relatively lower than commercial prediction models, making it suitable for deployment on edge computing devices for real-time prediction after training. Different wells correspond to different traffic calculation network models.

[0063] Step 130: When the target well does not have a corresponding flow calculation network model, the well test data is substituted into a preset mathematical calculation model for prediction to obtain the predicted data of multiphase flow rate.

[0064] The mathematical calculation model in the embodiment can be an oil and gas well calculation model in an existing commercial calculation software, which considers fluid dynamics, thermodynamics, phase change and other aspects, and can obtain the prediction of the flow rate of each of oil, gas and water according to the input parameters such as the density, viscosity and phase content of each phase fluid. The mathematical calculation model is arranged in the edge cloud system, and is used to output the first predicted multiphase flow rate in the periodic well testing link to compare with the measured reference multiphase flow rate, and confirm whether the mathematical calculation model deviates.

[0065] In step 140, it is judged whether the difference between the measured data and the predicted data of the multiphase flow rate satisfies a preset difference condition.

[0066] In the embodiment, the first preset range can be ±3% of the reference multiphase flow rate. When the difference between the first predicted multiphase flow rate and the reference multiphase flow rate is within the range of 3% of the reference multiphase flow rate, it indicates that the prediction result of the first predicted multiphase flow rate is relatively accurate, and can be used to guide production decision.

[0067] The first preset range is specified according to the production requirements of each well, and is not limited here. For example, in an embodiment, each well corresponds to a preset mathematical calculation model to predict the multiphase flow rate. The first preset range corresponding to well A is ±1% of the reference multiphase flow rate, and the first preset range corresponding to well B is ±5% of the reference multiphase flow rate, so as to improve the reliability of the mathematical calculation model precision test.

[0068] In step 150, when the difference does not satisfy the difference condition, the mathematical calculation model is corrected to make the difference satisfy the difference condition, so as to obtain a first corrected calculation model.

[0069] Generally, oil and gas wells under different geological conditions have different production characteristics and laws, and the mathematical calculation model is often difficult to adapt to such differences. Therefore, calibration is performed through the measured data, a more accurate calculation model can be established for oil and gas wells under different geological conditions, and the applicability and reliability of the model are improved. When the difference between the first predicted multiphase flow rate and the reference multiphase flow rate is greater than or equal to the first preset range, it indicates that the prediction result of the mathematical calculation model for the target well is not accurate enough, and the mathematical calculation model needs to be corrected. When correcting, the key parameters such as volume content in the mathematical calculation model can be adjusted for multiple times until the deviation between the prediction value of the adjusted model and the measured reference multiphase flow rate is within the expected range, then it is considered that the model is corrected, fits the production characteristics and laws of the target well, and obtains the first corrected calculation model.

[0070] Step 160, training a neural network model using the well test data and the first corrected data of the multiphase flow calculated according to the first correction calculation model to obtain a flow calculation network model of the target well.

[0071] In this embodiment, the process of training the preset neural network model using the well test data input into the first correction calculation model and the first corrected data output therefrom as a training set can be to input different groups of input parameters of the target well, such as tubing pressure and casing pressure in the well test data, into the first correction calculation model for multiple runs, thereby obtaining the multiphase flow prediction value corresponding to each group of input parameters. Then, the training set is generated according to different groups of input parameters and the corresponding set of multiphase flow prediction values to train the preset neural network model until the prediction result of the preset neural network model meets the preset standard, and then it is considered that the preset neural network model has been calibrated and can be more suitable for the production characteristics and laws of the target well.

[0072] In the process of generating the training set, relevant well site dynamic / static data can also be extracted and cleaned, verified and processed (such as moving average line, statistical parameters, etc.), and the initial neural network model built in the system is automatically trained and tested until the prediction error of the neural network model meets the error standard, that is, the construction of the flow calculation network model is completed.

[0073] Step 170, sending the flow calculation network model of the target well to the edge computing device.

[0074] In this embodiment, the flow calculation network model is sent to the edge computing device, and the edge computing device can use the updated neural network model to make predictions based on real-time well data to obtain real-time prediction data of the multiphase flow, thereby more timely guiding production decisions.

[0075] When sending the updated neural network model, the edge cloud system can package the updated neural network model and all its dependencies (such as libraries, environment variables, configuration files, etc.) into an independent, lightweight, and portable container. Since the neural network model is encapsulated in the container, it can be deployed in any environment that supports container running, thereby improving the consistency and reliability of the neural network model in different computing environments. In addition, since each container is independent, it can be updated or rolled back to a certain historical version without affecting other containers or the entire system, greatly improving the flexibility and reliability of the system.

[0076] In summary, the processing method of the prediction model provided in the embodiment can find whether the mathematical calculation model deviates by comparing the first predicted multiphase flow rate obtained by the mathematical calculation model with the reference multiphase flow rate, correct the mathematical calculation model if the deviation occurs, train the neural network model by using the corrected mathematical model and the measured data, and finally make the updated neural network model have a significant improvement in prediction accuracy. The updated neural network is installed in the edge computing device, so that the real-time prediction based on the well data collected in the production site can be more accurate and timely, and immediate feedback can be provided for production decision-making. In addition, the model training and updating in the edge cloud system can significantly reduce the delay and cost of data transmission, and improve the response speed, so that the upgrading and iteration of the neural network model is more timely, and the reliability of the real-time prediction result of the multiphase flow rate is improved.

[0077] The method of the embodiment can perform the accuracy test and correction of the prediction model every time the test production is performed, and the cooperation of the edge cloud system and the edge computing device can perform the multiphase flow rate prediction based on the well data collected in the production process in time, which can provide immediate feedback for production decision-making and improve the accuracy and reliability of the real-time prediction result.

[0078] Embodiment Two

[0079] As shown in Figure 3 On the basis of the above-mentioned embodiment, after the step of detecting whether the target well has a corresponding flow rate calculation network model, the method further comprises: when the target well has a corresponding flow rate calculation network model, receiving the calculation data of the multiphase flow rate sent by the edge computing device, wherein the calculation data of the multiphase flow rate is a prediction result obtained by inputting the well test data into the flow rate calculation network model for prediction; judging whether the difference between the measured data and the calculation data of the multiphase flow rate satisfies the difference condition; when the difference does not satisfy the difference condition, correcting the first correction calculation model so that the difference between the first correction data of the multiphase flow rate obtained according to the first correction calculation model and the measured data satisfies the difference condition, thereby obtaining a second correction calculation model; training the flow rate calculation network model by using the well test data and the second correction data of the multiphase flow rate obtained according to the second correction calculation model, so that the flow rate calculation network model satisfies a preset model condition, thereby obtaining an updated flow rate calculation network model of the target well; and sending the updated flow rate calculation network model to the edge computing device.

[0080] In this embodiment, after the initial flow calculation network model is trained, the flow calculation network model on the edge computing device calculates the corresponding predicted value according to the well testing data during well testing, that is, the calculation data of the multiphase flow rate, and returns the calculation data of the multiphase flow rate, the measured data and the well testing data to the edge cloud system. When the edge cloud system determines that the difference between the measured data and the calculation data of the multiphase flow rate does not satisfy the difference condition, it means that the current flow calculation model deviates greatly from the production characteristics of the target well, and needs to be retrained. When retraining the flow calculation network model, the current mathematical calculation model (first correction calculation model) is first corrected, because the training data of the flow calculation network model all come from the mathematical calculation model, and a more accurate training set is obtained after correcting the mathematical calculation model, and the flow calculation network model is trained using the training set, so that the accuracy of the flow calculation network model is higher.

[0081] In this embodiment, the flow calculation network model on the edge computing device feeds back the predicted data to the edge cloud system, the edge cloud system corrects the flow calculation network model again, and sends the updated flow calculation network model to the edge computing device, so as to realize the self-loop accuracy verification and update of the system without manual intervention, and the operation of the whole system is more efficient.

[0082] In one embodiment, there is a flow calculation network model on the edge computing device, and when well testing, the edge cloud system receives the well testing data of the target well and the measured data of the multiphase flow rate corresponding to the well testing data, first substitutes the well testing data into the first correction calculation model to obtain the predicted result, compares the predicted result with the measured data, and when it is found that the predicted result deviates greatly from the measured data, corrects the first correction calculation model to obtain the second correction calculation model, and then trains the flow calculation network model using the well testing data and the second correction data of the multiphase flow rate obtained according to the second correction calculation model to obtain the updated flow calculation network model.

[0083] In this embodiment, the deviation result of the predicted value of the first correction calculation model and the measured value is used to determine, which can make the determination result of the model accuracy more reliable. The reason is that the mathematical calculation model is an algorithm based on physical principles, and the predicted result obtained in most cases is more accurate than the predicted result obtained by the neural network model, so when the predicted result of the mathematical calculation model deviates greatly from the measured result, it means that the predicted value of the neural network model also deviates greatly, so the mathematical calculation model can be directly corrected, and the neural network model can be iteratively trained.

[0084] Example Three

[0085] On the basis of the above-mentioned embodiments, after the step of sending the updated neural network model to the edge computing device, the method further comprises receiving acknowledgement information generated by the edge computing device when replacing the neural network model stored on the edge computing device with the updated neural network model.

[0086] In this embodiment, after the model replacement is completed, the edge device generates acknowledgement information to notify the system or other related components that the update has been completed. For example, the edge computing device replaces the preset neural network model stored on the edge computing device with the updated neural network model, and in response thereto, generates acknowledgement information and sends the acknowledgement information to the edge cloud system. For example, the edge computing device replaces the last updated neural network model stored on the edge computing device with the updated neural network model, and in response thereto, generates acknowledgement information and sends the acknowledgement information to the edge cloud system.

[0087] The acknowledgement information can include the following contents:

[0088] Update timestamp: records the specific time when the model replacement is completed.

[0089] Model version: version number or identifier of the new model for tracking and verification.

[0090] Status information: indicates whether the update is successful, and whether there are any errors or exceptions.

[0091] Device identifier: unique identifier of the edge device for distinguishing among multiple devices.

[0092] After the cloud server receives the acknowledgement information, further processing is performed, including:

[0093] Log recording: records the acknowledgement information into a log system for subsequent analysis and auditing.

[0094] Status update: updates the system status to reflect that the model on the edge device has been updated to the latest version.

[0095] Notification and alarm: if the acknowledgement information indicates that the update has failed or there are exceptions, the system can trigger a notification or alarm mechanism so that the administrator can intervene in a timely manner.

[0096] Example Four

[0097] On the basis of the above-mentioned embodiments, the mathematical calculation model is corrected so that the difference between the first correction data and the reference multiphase flow rate is within a first preset range, and the step of obtaining the first corrected calculation model includes: obtaining a preset value range corresponding to the volume fraction in the mathematical calculation model; adjusting the volume fraction within the preset value range to obtain a temporary mathematical model; inputting the well test data into the temporary mathematical model for prediction to obtain temporary prediction data of the multiphase flow rate; and when the temporary prediction data of the multiphase flow rate and the measured data are within the first preset range, determining that the temporary mathematical model is the first corrected calculation model.

[0098] The volume fraction is one of the key parameters affecting the accuracy of multiphase flow rate prediction. In the embodiment, the preset value range corresponding to the volume fraction in the mathematical calculation model is obtained, which ensures that the subsequent adjustment of the volume fraction is within a reasonable range, avoiding meaningless or possibly unstable parameter settings. By adjusting the volume fraction, a plurality of temporary mathematical models are generated to test the effectiveness of these models through actual data.

[0099] Specifically, the adjustment of the volume fraction can be performed by gradually increasing, decreasing, or automatically adjusting based on a preset optimization algorithm such as genetic algorithm, particle swarm optimization, etc. For example, the preset value range of the volume fraction is [1, 5], and the value in the current model is 3. When adjusting, the value can be adjusted by 0.5 steps, i.e. increasing or decreasing 0.5 each time based on 3. According to the adjustment strategy, the value of the volume fraction is changed one by one, and a new temporary mathematical model is generated each time the value is changed. The well test data is input into each temporary mathematical model, the model is run for prediction, and a plurality of temporary prediction multiphase flow rates are obtained. The prediction results (temporary prediction multiphase flow rates) of each temporary mathematical model are compared with the reference multiphase flow rate (which can be obtained by actual measurement or more accurate model calculation), and the temporary mathematical model whose difference is within the first preset range is determined as the first corrected calculation model.

[0100] In one embodiment, if there are multiple temporary mathematical models that meet the conditions, the prediction accuracy, stability or calculation efficiency of each model can be further compared, and finally one is selected as the first corrected calculation model.

[0101] Example Five

[0102] On the basis of the above-mentioned embodiments, before the step of receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data, the method further comprises: in response to a connection request of the edge computing device, creating a data object in a preset data model, wherein the preset data model is a standardized data structure for realizing data exchange between different edge computing devices and edge cloud systems, and the data object is configured to receive raw data and processed data from the target well; and the step of receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data comprises: receiving the well test data of the target well and the measured data of the multiphase flow corresponding to the well test data, and storing the well test data and the measured data of the multiphase flow in corresponding data objects, respectively.

[0103] In this embodiment, the preset data model is a standardized data structure for realizing data exchange between edge computing devices and edge cloud systems. The preset data model defines the format, type, structure and possible metadata of data, ensuring that data can be seamlessly exchanged and understood between different devices and systems. After verifying the connection request of the edge computing device, the edge cloud system creates a corresponding data object in the internal database according to the preset data model. The data object is configured to receive and process data from a specific edge computing device, including raw data and processed data. Each data object is configured according to its purpose, such as setting the interface for data reception, the format for data storage, and the rules for data processing. These configurations ensure the consistency and accuracy of data during reception, storage and processing.

[0104] When each edge device requests to connect to the edge cloud system, the system automatically discovers and brings it online, and creates a data object in the unified data model. The data object completely matches the data format exposed by the edge device itself, achieving on-site configuration only once, avoiding configuration errors during debugging, and reducing the setup time required to enable on-site devices.

[0105] When the edge computing device starts sending well test data of the target well and measured reference multiphase flow, the edge cloud system receives these data through the preset interface. These data may include raw well test data such as wellhead pressure, temperature, production, water cut, and reference multiphase flow measured by a specific method. The received data is stored in the previously created data object according to its type and purpose. For well test data, it can be stored in an object specifically for storing raw wellsite data; and for measured reference multiphase flow, it can be stored in an object for storing key performance indicators or reference data.

[0106] Example Six

[0107] On the basis of the above-mentioned embodiments, the oil well production calculation method provided by the present disclosure comprises:

[0108] The processing method of the prediction model in any of the above-mentioned embodiments is used to obtain an updated neural network model; the wellhead real-time measurement data of the target well are input into the updated neural network model on the edge computing device for prediction to obtain real-time prediction data of the multiphase flow rate; and the oil well production of the target well is calculated by using the real-time prediction data of the multiphase flow rate.

[0109] The oil well production calculation method of the present embodiment can rapidly process the wellhead real-time measurement data and calculate the oil well production due to the use of the real-time processing capability of the edge computing device, thereby providing timely data support for oilfield production management. By continuously updating the neural network model to adapt to the data characteristics of the target well, the prediction accuracy is improved, thereby ensuring the accuracy of the oil well production calculation. In addition, the real-time prediction process is completed on the edge computing device, thereby reducing the delay of data transmission and the calculation pressure of the cloud, and improving the overall efficiency and scalability of the system. The present method combines the prediction model processing method and the real-time data processing technology to realize the rapid and accurate calculation of the oil well production, thereby providing strong support for oilfield production management.

[0110] Example Seven

[0111] On the basis of the above-mentioned embodiments, the present embodiment provides an application example. The technical method is tested and applied in a certain development block of a petrochemical company. The specific implementation process is as follows:

[0112] (1) The system for automatically measuring the oil well production based on edge computing is deployed in the regional server and well site edge terminal of the target oilfield, the related data of the oilfield is extracted, and the system is activated to run.

[0113] (2) A new well test of X well is received, and the workflow is activated. The new well test data is stored in the system database of X well.

[0114] When the system receives the well test information of X well, the workflow starts. This process can be automatically received through the existing automated well test system, or manually input by the user. If more well tests from multiple wells are received, multiple instances of this workflow can be started. As a prerequisite, each well must have a model (mathematical calculation model) generated by a nodal analysis well modeling and simulation software package, and each mathematical calculation model is stored in the well model library of the edge cloud system.

[0115] (3) The mathematical calculation model corresponding to X well is obtained from the well model library.

[0116] (4) The new well test data is compared to the results predicted by the mathematical model. If the results match, i.e. the calculated flow rate matches the well test measured flow rate, no further action is performed.

[0117] Specifically, the well test data is input into the mathematical model for prediction to obtain a first predicted multiphase flow rate; it is determined whether a difference between the measured data and the predicted data of the multiphase flow rate satisfies a preset difference condition; when the difference between the first predicted multiphase flow rate and the reference multiphase flow rate is less than a first preset range, it is indicated that the mathematical model matches the well test measured flow rate, and the existing mathematical model is kept unchanged.

[0118] (5) The mathematical model of X well is automatically updated according to the new well test data. For example, the workflow opens the mathematical model and adjusts its internal parameters to match the latest well test data.

[0119] When the difference between the first corrected data and the reference multiphase flow rate is greater than or equal to the first preset range, the mathematical model is corrected so that the difference between the first corrected data and the reference multiphase flow rate is within the first preset range, to obtain a first corrected calculation model.

[0120] When correcting, the workflow obtains a preset value range corresponding to the volume fraction in the mathematical model; the volume fraction is adjusted within the preset value range to obtain a temporary mathematical model; the well test data is input into the temporary mathematical model for prediction to obtain temporary prediction data of the multiphase flow rate; when the temporary prediction data of the multiphase flow rate and the measured data are less than the first preset range, the temporary mathematical model is determined as the first corrected calculation model.

[0121] Adjusting the volume fraction can include gradually increasing, decreasing, or automatically adjusting based on a preset optimization algorithm such as a genetic algorithm, particle swarm optimization, etc. According to the adjustment strategy, the value of the volume fraction is changed one by one, and a new temporary mathematical model is generated after each change. The well test data is input into each temporary mathematical model, the model is run for prediction to obtain a plurality of temporary prediction multiphase flow rates. The prediction results (temporary prediction multiphase flow rates) of each temporary mathematical model are compared with the reference multiphase flow rate (which can be obtained by actual measurement or more accurate model calculation) to determine the temporary mathematical model whose difference is less than the first preset range as the first corrected calculation model.

[0122] (6) If the mathematical model of X well cannot be automatically adjusted using the well test data, an event will be generated. The event is sent to the team of engineers, who will study the case and finally adjust the mathematical model.

[0123] (7) The newly updated mathematical model is stored in the well model repository.

[0124] (8) The updated mathematical model is run multiple times using changes in the well input parameters (tubing pressure, casing pressure, etc.) and obtains a calculated result in each case, such as liquid flow rate. A large dataset, or "synthetic" data, is generated, with each record including specific input parameter values and expected results.

[0125] (9) Using the generated dataset, the neural network is trained and tested, and once the error is below a set threshold of 2%, the new neural network model (flow rate calculation model) is saved.

[0126] (10) The newly trained neural network model is placed in a container to generate a new neural network predictor. The new neural network predictor is then installed on the edge device of Well X oilfield and waits for the edge to successfully confirm receipt of the new neural network predictor.

[0127] If the new neural network predictor container is received in the edge device, the system will replace the current predictor container and send a confirmation receipt to the edge cloud system module.

[0128] (11) The edge device operates in the field, collects real-time data, and feeds it into the updated neural network predictor running in the new container, which estimates the liquid flow rate using the received neural network model. The edge device will continue to use this predictor until a new one is received.

[0129] (12) The estimated flow rate calculated in real time by the neural network model is sent to the system in the edge cloud along with all other real-time data and presented to the user.

[0130] The method and system for automatically metering oil well production of the present embodiment have the following advantages:

[0131] (1) Compared with the actual measurement method currently used in oilfields, the present application uses the existing measurement data at the wellhead and a data-driven model deployed on an edge device to achieve low-cost (low investment cost, low operation and maintenance cost) real-time automatic calculation of oil and gas production.

[0132] (2) Compared with the virtual metering method currently used in oilfields, the present application uses neural networks, edge computing technology, and an automatically started workflow to achieve real-time automatic calculation of oil and gas production using a continuously updated data-driven model, reducing the need for manual intervention to set up models during oil and gas production, eliminating the need for manual operation of commercial software for production calculation, saving time and cost.

[0133] (3) The real-time processing capability of the edge computing device is utilized to quickly process the wellhead real-time measurement data and calculate the oil well production, thereby providing timely data support for oilfield production management. By continuously updating the neural network model to adapt to the data characteristics of the target well, the prediction accuracy is improved, thereby ensuring the accuracy of the oil well production calculation.

[0134] (4) The real-time prediction process is completed on the edge computing device, reducing the data transmission delay and the computing pressure of the cloud, and improving the overall efficiency and scalability of the system. The method combines the prediction model processing method and the real-time data processing technology to realize the rapid and accurate calculation of the oil well production, thereby providing strong support for oilfield production management.

[0135] Example Eight

[0136] On the basis of the above embodiments, as shown in Figure 4 The prediction model processing device provided by the embodiment includes:

[0137] The receiving module 410 is configured to receive the well test data of a target well and measured data of multiphase flow corresponding to the well test data.

[0138] The model detection module 420 is configured to detect whether there is a corresponding flow calculation network model for the target well.

[0139] The prediction module 430 is configured to, when there is no corresponding flow calculation network model for the target well, substitute the well test data into a preset mathematical calculation model to obtain predicted data of the multiphase flow.

[0140] The deviation detection module 440 is configured to determine whether a difference between the measured data of the multiphase flow and the predicted data satisfies a preset difference condition.

[0141] The correction module 450 is configured to, when the difference does not satisfy the difference condition, correct the mathematical calculation model to make the difference satisfy the difference condition, thereby obtaining a first corrected calculation model.

[0142] The training module 460 is configured to train a neural network model by using the well test data and first corrected data of the multiphase flow obtained according to the first corrected calculation model, to obtain a flow calculation network model for the target well.

[0143] The sending module 470 is configured to send the flow calculation network model for the target well to the edge computing device.

[0144] In one embodiment,

[0145] The receiving module is further configured to receive the calculation data of the multiphase flow rate sent by the edge computing device when the target well has a corresponding flow rate calculation network model, wherein the calculation data of the multiphase flow rate is a prediction result obtained by inputting the well test data into the flow rate calculation network model for prediction.

[0146] The deviation detection module is further configured to determine whether the difference between the measured data and the calculation data of the multiphase flow rate satisfies the difference condition.

[0147] The correction module is further configured to correct the first correction calculation model when the difference does not satisfy the difference condition, so that the first correction data of the multiphase flow rate obtained according to the first correction calculation model satisfies the difference condition with the measured data, thereby obtaining a second correction calculation model.

[0148] The training module is further configured to train the flow rate calculation network model by using the well test data and the second correction data of the multiphase flow rate obtained according to the second correction calculation model, so that the flow rate calculation network model satisfies a preset model condition, thereby obtaining an updated flow rate calculation network model of the target well.

[0149] The sending module is further configured to send the updated flow rate calculation network model to the edge computing device.

[0150] In an embodiment, the method further includes:

[0151] The receipt receiving module is configured to receive receipt information generated by the edge computing device when the edge computing device replaces the flow rate calculation network model stored on the edge computing device with the updated flow rate calculation network model.

[0152] In an embodiment, the training module includes:

[0153] The value range obtaining unit is configured to obtain a preset value range corresponding to the volume fraction in the mathematical calculation model.

[0154] The parameter adjusting unit is configured to adjust the volume fraction within the preset value range to obtain a temporary mathematical model.

[0155] The temporary prediction unit is configured to input the well test data into the temporary mathematical model for prediction to obtain temporary prediction data of the multiphase flow rate.

[0156] The correction result determining unit is configured to determine that the temporary mathematical model is a first correction calculation model when the temporary prediction data of the multiphase flow rate is less than a first preset range from the measured data.

[0157] In an embodiment, the method further includes:

[0158] a data object creation module, configured to create a data object in a preset data model in response to the connection request of the edge computing device, wherein the preset data model is a standardized data structure for realizing data exchange between different edge computing devices and edge cloud systems, and the data object is configured to receive raw data and processed data from the target well,

[0159] The receiving module comprises:

[0160] a storage unit, configured to receive well testing data of a target well and measured data of multiphase flow corresponding to the well testing data, and store the well testing data and the measured data of multiphase flow in corresponding data objects respectively.

[0161] The specific limitations of the processing device of the prediction model can refer to the limitations of the processing method of the prediction model described above, which will not be repeated here. Each unit in the above prediction model processing device can be realized by software, hardware and their combinations in whole or in part. The above units can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each unit.

[0162] Example Nine

[0163] On the basis of the above embodiments, as Figure 5 shown, a system for calculating oil well production is provided, which mainly includes an edge computing module 510 deployed by an edge terminal and a data service module 520, a business flow module 530 and a visualization module part 540 deployed in an edge cloud, specifically:

[0164] The data service module realizes the connection of well site edge measurement data, calculation data and other real-time data by using standard Internet of Things (IoT) exchange protocols such as MQTT, AMQP, etc. through real-time data aggregation function, and also supports using other communication protocols (such as OPC / DA, OPC / UA, Modbus and Ethernet / IP, etc.) to establish data connection with existing systems or other real-time or near real-time data sources;

[0165] The data service module collects non-real-time data from static or information sources through static data connection function. For structured data files (Microsoft Excel, PDF, etc.), SQL, ODBC, JDBC or similar protocols can be used, and for the collection and / or storage of other unstructured data, available extraction, transmission and loading (ETL) mechanisms are used;

[0166] The data service module standardizes the data gathered from various data sources according to a unified data model and encapsulates the data in well-defined standardized data objects through a unified data standard function. In a single data unit, each data object collects all information (real-time, time-stamped, static, structured or unstructured) from the well site entity. These data units also include any data processing programs, engineering unit conversions, and application-specific subroutine (method) processing / computing of intermediate data formed, and many others;

[0167] The data service module provides the data units in the form of data services to the upper-layer business flow module or other systems for use through a data service interface management function, isolating the data application from the data aggregation processing; the data service module is also responsible for the storage of all aggregated data and intermediate data of processing / computing;

[0168] The business flow module adjusts the model by programming through a multiphase flow mathematical model adjustment function, so that the adjusted model matches the new well test data. If the model cannot be automatically adjusted (the probability of occurrence under the current system automation process is extremely low), it will be forwarded to the engineering team for revision or manual reconfiguration of parameter adjustment to complete the model adjustment;

[0169] The business flow module generates a large amount of "synthetic" data (including all inputs (well real-time data) and their corresponding outputs (liquid velocity)) through a multiphase flow mathematical model simulation calculation function, which repeatedly inputs simulation calculation functions according to a series of model input values through an automatic workflow;

[0170] The business flow module extracts relevant well site dynamic / static data and synthetic data and cleans, verifies and processes the data (moving average line, statistical parameters, etc.) through a data-driven model construction and training function, automatically trains and tests the initial neural network model built in the system until the model prediction error is less than the set threshold of 3%. The well site dynamic / static data include but are not limited to: surface instrument data (tubing head pressure, etc.); ESP controller information (ESP pump inlet pressure, ESP pump discharge pressure, ESP motor frequency); well completion information (vertical total depth, pump depth, pump size, pump stage, motor nominal frequency, nominal surface voltage, nominal current); PVT information (oil specific gravity, gas specific gravity, water specific gravity, bubble point pressure, solution gas-oil ratio, water cut, production gas-oil ratio); reservoir information (reservoir pressure, reservoir temperature); well test data, etc.

[0171] The service flow module transmits the data-driven model constructed and trained to the corresponding well site edge terminal through the on-site edge service management function, receives the on-site edge terminal operation data and monitors the model operation state, regularly evaluates the accuracy of the data-driven model applied by the edge terminal based on the on-site measured data in the data module, timely re-trains and improves the model and updates the data-driven model applied by the edge terminal;

[0172] The visualization module provides the user with service data visualization, system state and edge service visualization, basic data visualization and other monitoring functions through the front-end interface of the PC terminal and the mobile terminal.

[0173] The edge computing module is deployed at the well site edge terminal and runs on the basis of the edge service system infrastructure, packs all application program codes, dependencies and configurations of the edge terminal into a computing object and encapsulates them in a "container" through the service container management function, can be deployed and updated in any computing environment, thus improving the reliability of the edge system and supporting the update of individual containers (application services) without affecting other containers (application services).

[0174] The edge computing module automatically discovers each edge device when it requests to connect to the edge cloud system through the automatic discovery function, and puts it online and creates a data object in the unified data model, which is fully matched with the data model exposed by the edge device itself, realizes that the on-site configuration only needs to be completed once, does not produce configuration errors during debugging, and reduces the setting time required to enable the on-site device.

[0175] The edge computing module realizes the local storage of the data collected by the edge device and the data calculated by the data-driven model and the data communication with the data service module in the edge cloud through the edge data management and communication function.

[0176] Example Ten

[0177] On the basis of the above-mentioned embodiments, the present embodiment provides a computer device, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to realize the steps of the method of the above-mentioned embodiments.

[0178] In some embodiments of the present embodiment, a computer readable storage medium is provided, and a computer program is stored on the computer readable storage medium, wherein the computer program is executed by a processor to realize the steps of the method of the above-mentioned embodiments.

[0179] In some embodiments of the present embodiment, a computer program product is provided, comprising a computer program / instruction, wherein the computer program is executed by a processor to realize the steps of the method of the above-mentioned embodiments.

[0180] The processor can include, but is not limited to, one or more processors or microprocessors, etc. Each processor can be an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a controller, a microcontroller, a microprocessor, or other electronic elements for executing the methods in the above-described embodiments

[0181] The computer-readable storage medium can be implemented by any type of volatile or nonvolatile storage devices or a combination thereof, and can include, but is not limited to, for example, a random access memory (RAM), a read-only memory (ROM), a flash memory, an EPROM memory, an EEPROM memory, a register, a computer storage medium (such as a hard disk, a floppy disk, a solid state disk, a removable disk, a Blu-ray disk, etc.).

[0182] The computer-readable storage medium can also store at least one computer executable program / instruction, such as computer readable instructions. The computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory can include, for example, a random access memory (RAM) and / or a cache, etc. The computer-readable storage medium can include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. For example, the non-transitory computer-readable storage medium can be connected to a computing device, such as a computer, and then, in the case where the computing device executes the computer readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0183] In addition, the computer device can also include, but is not limited to, a data bus, an input / output (I / O) bus, a display, and an input / output device (such as a keyboard, a mouse, a speaker, etc.), etc.

[0184] The processor can communicate with an external device via a wired or wireless network through the I / O bus.

[0185] In one embodiment, the at least one computer-executable instruction can also be compiled or composed into a software product / computer program product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods described in the embodiments of the present technology.

[0186] In the embodiments provided in the present disclosure, it should be understood that the disclosed apparatus and method can also be implemented in other manners. The described embodiments of the apparatus are merely exemplary, and the present disclosure is not limited thereto. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architectures, functions and operation of the apparatus, method and computer program product according to the embodiments of the present disclosure. In this regard, each block in the flowcharts or block diagrams can represent a module, a segment or a portion of code which comprises one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the flowcharts. For example, two blocks shown in succession can in fact be executed substantially concurrently or in the reverse order, depending upon the functionality involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by dedicated hardware-based systems which perform the specified functions or acts, or can be implemented by a combination of dedicated hardware and computer instructions.

[0187] It should be noted that, in the present disclosure, the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusion, such that processes, methods, articles or devices that comprise a series of elements not only include those elements, but also include other elements not explicitly listed, or other elements inherent to such processes, methods, articles or devices. Without more limitations, the element limited by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0188] Although the embodiments disclosed in the present disclosure are as described above, the above description is only for the purpose of facilitating understanding of the present disclosure, and is not intended to limit the present disclosure. Any person skilled in the art without departing from the spirit and scope of the present disclosure can make any modification and change in the implementation form and details, but the patent protection scope of the present disclosure shall be subject to the scope defined by the appended claims.

Claims

1. A method for processing a prediction model, characterized in that, The method is applied to an edge cloud system in a cloud-edge collaborative system, wherein the cloud-edge collaborative system includes an edge cloud system and at least one edge computing device communicating with the edge cloud system, and the method includes: Receive well test data from the target well, as well as measured data of the multiphase flow rate corresponding to the well test data; Check whether the target well already has a corresponding flow calculation network model; When the target well does not have a corresponding flow calculation network model, the well test data is substituted into a preset mathematical calculation model for prediction to obtain the predicted data of multiphase flow. Determine whether the difference between the measured data and the predicted data of the multiphase flow rate meets the preset difference condition; When the difference does not meet the difference condition, the mathematical calculation model is corrected so that the difference meets the difference condition, thereby obtaining the first corrected calculation model; The neural network model is trained using the well test data and the first correction data of the multiphase flow rate obtained according to the first correction calculation model to obtain the flow rate calculation network model of the target well. Send the flow calculation network model of the target well to the edge computing device.

2. The method according to claim 1, characterized in that, After the step of detecting whether the target well already has a corresponding flow calculation network model, the method further includes: When the target well already has a corresponding flow calculation network model, the calculation data of the multiphase flow rate sent by the edge computing device is received, wherein the calculation data of the multiphase flow rate is the prediction result obtained by inputting the well test data into the flow calculation network model for prediction; Determine whether the difference between the measured data and the calculated data of the multiphase flow rate satisfies the difference condition; When the difference does not meet the difference condition, the first correction calculation model is corrected so that the difference between the first corrected data of the multiphase flow rate obtained according to the first correction calculation model and the measured data meets the difference condition, thereby obtaining the second correction calculation model; The flow calculation network model is trained using the well test data and the second correction data of the multiphase flow rate obtained according to the second correction calculation model, so that the flow calculation network model meets the preset model conditions, thereby obtaining the updated flow calculation network model of the target well; The updated traffic calculation network model is sent to the edge computing device.

3. The method according to claim 2, characterized in that, After the step of sending the updated traffic calculation network model to the edge computing device, the method further includes: Receive receipt information generated by the edge computing device when it replaces the traffic calculation network model stored on the edge computing device with the updated traffic calculation network model.

4. The method according to claim 1, characterized in that, The step of correcting the mathematical calculation model to make the difference satisfy the difference condition, thereby obtaining the first corrected calculation model, includes: Obtain the preset value range corresponding to the volume fraction in the mathematical calculation model; The volume fraction is adjusted within the preset value range to obtain a temporary mathematical model; The well test data is input into the temporary mathematical model for prediction to obtain temporary predicted data of the multiphase flow rate; When the difference between the temporary predicted data and the measured data of the multiphase flow rate meets the difference condition, the temporary mathematical model is determined to be the first correction calculation model.

5. The method according to claim 1, characterized in that, Before the step of receiving the well test data of the target well and the measured data of the multiphase flow rate corresponding to the well test data, the method further includes: In response to a connection request from the edge computing device, a data object is created in a preset data model, wherein the preset data model is a standardized data structure for data exchange between different edge computing devices and edge cloud systems, and the data object is configured to receive raw data and processed data from the target well. The steps of receiving well test data from the target well and the measured multiphase flow rate data corresponding to the well test data include: Receive well test data from the target well and measured multiphase flow rate data corresponding to the well test data, and store the well test data and the measured multiphase flow rate data in corresponding data objects respectively.

6. A method for calculating oil well production, characterized in that, include: An updated neural network model is obtained using the processing method of the prediction model according to any one of claims 1-5; The real-time measurement data at the wellhead of the target well is input into the updated neural network model on the edge computing device for prediction, and real-time predicted data of multiphase flow rate is obtained. The oil well production of the target well is calculated using the real-time predicted data of the multiphase flow rate.

7. An oil well production calculation system, characterized in that, It includes a data service module, a business flow module, an edge computing module, and a visualization module, among which: The data service module is used to receive well test data of the target well, as well as measured data of multiphase flow rate corresponding to the well test data; The business flow module is used to detect whether the target well already has a corresponding flow calculation network model; when the target well does not have a corresponding flow calculation network model, the well test data is substituted into a preset mathematical calculation model for prediction to obtain predicted data of multiphase flow; it is determined whether the difference between the measured data and the predicted data of the multiphase flow meets a preset difference condition; when the difference does not meet the difference condition, the mathematical calculation model is corrected to make the difference meet the difference condition, thereby obtaining a first corrected calculation model; a neural network model is trained using the well test data and the corrected data of the multiphase flow obtained according to the first corrected calculation model to obtain the flow calculation network model of the target well; the flow calculation network model of the target well is sent to the edge computing device; The edge computing module is used to receive and install the flow calculation network model, receive real-time measurement data of the wellhead of the target well and input it into the flow calculation network model for prediction, so as to obtain real-time predicted data of the multiphase flow rate. A visualization module is used to display the real-time measurement data at the wellhead and the real-time predicted data of the multiphase flow rate.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 6.

Images