Submersible electric pump well group management system and method based on digital twinning

Abstract

The application belongs to the technical field of oil well production management, and discloses a submersible electric pump well group management system and method based on digital twinning. The method collects the operation data of the submersible electric pump unit and the environmental data in the wellbore in real time, constructs a multi-scale digital twinning model for the received data, obtains the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics in the wellbore, and the geological pressure field data, performs trend prediction on the multi-scale digital twinning model operation parameters, energy efficiency indicators and health status of each well, correlates the trend prediction data with the historical health database through a machine learning algorithm and an expert knowledge base, completes early warning and intelligent diagnosis of equipment failure, and generates a maintenance strategy. The application can assist managers in decision-making based on the historical health database. The application can realize unified management and intelligent well inspection of multiple well sites, and greatly improves the efficiency of oil well production management.

Description

Technical Field

[0001] This invention belongs to the field of oil well production management technology, and in particular relates to a digital twin-based submersible electric pump well group management system and method. Background Technology

[0002] Submersible electric pumps (SAPs) are core equipment for efficient oil production in oilfields, but traditional management methods rely on manual inspections and single-well data monitoring, which suffer from problems such as delayed response and weak fault prediction capabilities. There is also a lack of well group collaborative optimization: existing systems mostly perform threshold alarms based on single-well operating parameters (current, temperature, pressure), lacking dynamic correlation analysis between well groups. Existing digital twin models are mostly based on high-fidelity simulations (such as ANSYS simulations), resulting in high computational complexity and difficulty in meeting the real-time requirements of well group management. Summary of the Invention

[0003] To overcome the problems existing in related technologies, the present invention discloses an embodiment of a digital twin-based submersible electric pump well group management system and method.

[0004] The technical solution is as follows: A management method for submersible electric pump well groups based on digital twins, the method comprising the following steps:

[0005] S1 collects real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, processes the collected data, and transmits it to the cloud server.

[0006] S2, the cloud server uses an integrated digital twin model building module to construct a multi-scale digital twin model from the received data, and obtains the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics in the wellbore, and geological pressure field data; the multi-scale digital twin model includes: equipment-level model, wellbore-level model and reservoir-level model;

[0007] The cloud server utilizes an integrated digital twin model building module to construct multi-scale digital twin models from the received data, including:

[0008] A device-level model is constructed based on the geometric characteristic parameters of the submersible electric pump to simulate its mechanical structure and electrical characteristics. The construction process includes:

[0009] (1) Based on the design drawings or 3D scanning data of the submersible electric pump, establish a geometric model, including the dimensions, shape and assembly relationship of the stator, rotor, impeller and bearing components;

[0010] (2) Assign material properties and electrical parameters to the geometric model, including density, elastic modulus, thermal conductivity, resistance, and inductance;

[0011] (3) Perform multi-physics coupling simulation, simulate the stress, deformation and fatigue life of the electric pump under alternating load through finite element analysis, and analyze the temperature field change by combining CFD method.

[0012] (4) Integrate the multiphysics simulation results into the Simulink system-level model to simulate the dynamic response of the submersible electric pump under start-up, shutdown and variable load conditions;

[0013] A wellbore-level model is constructed based on wellbore environmental data, and the multiphase flow dynamics within the wellbore are reproduced through computational fluid dynamics simulation. The construction process includes:

[0014] (i) Establish a three-dimensional wellbore geometric model based on the wellbore structural parameters;

[0015] (ii) Set multiphase flow parameters, input real-time sensor data as boundary conditions, including pressure, temperature, and sand content; and define fluid properties, including crude oil viscosity and gas solubility;

[0016] (iii) Multiphase flow CDF simulation was used to calculate the velocity distribution, pressure gradient, and sand deposition in the wellbore;

[0017] The continuous phase mathematical model for multiphase flow dynamics is the continuity equation and the Navier-Stokers equation, expressed as follows:

[0018]

[0019] In the formula, a f ρf is the fluid volume fraction, expressed as %; ρf is the fluid density, expressed as kg / m³. 3 u is the fluid velocity, in m / s; p is the pressure, in Pa; μ is the fluid viscosity, in Pa·s; f b The term represents the source term between the continuous and dispersed phases, expressed in Pa / m; g is the gravitational acceleration, expressed in m / s². 2 t represents time, in seconds (s).

[0020] The equipment-level model, wellbore-level model, and reservoir-level model are also dynamically coupled to complete the macroscopic linkage analysis of micro-faults, including: the equipment-level model detects micro-anomalies, diagnoses them as fault modes, and the equipment-level model transmits fault parameters to the wellbore-level model, triggering the recalculation of flow velocity and pressure, which is then transmitted to the reservoir-level model to correct the pressure field in the near-wellbore zone.

[0021] S3 uses a web-based 3D visualization platform to predict trends in the operating parameters, energy efficiency indicators, and health status of multi-scale digital twin models of each well, supporting multi-dimensional data comparison and analysis and trend prediction.

[0022] The operating parameters, energy efficiency indicators, and health status of multi-scale digital twin models of each well are predicted using a web-based 3D visualization platform. This includes: simulating equipment operating conditions under different working conditions in a digital space using virtual experiment technology; actively injecting typical fault modes into the virtual model; and completing early fault location by comparing the differences between multi-dimensional sensor data and simulation data. The specific steps are as follows:

[0023] S301. Feature Extraction: Perform wavelet packet decomposition on the vibration signal, filter out most invalid frequency components, and extract energy signal features from multiple sub-bands; calculate the total harmonic distortion rate of the motor current; calculate the temperature gradient based on the thermal grid model of the temperature field.

[0024] S302. Data Comparison: Compare the multi-dimensional data features extracted from the measured data with the feature data obtained from the multi-dimensional digital twin model simulation to identify differences;

[0025] S303 Fault Location: By comparing the differences in different frequency bands of the signal, the total harmonic distortion rate of the current, and the temperature gradient, the fault is initially located in different components of the equipment, and then the specific fault type is matched according to the characteristic frequency.

[0026] S4, through machine learning algorithms and an expert knowledge base, correlates trend prediction data with a historical health database to achieve early warning and intelligent diagnosis of equipment failures, and generates maintenance strategies; including:

[0027] By analyzing the correlation between real-time monitoring data and historical health database through regression model analysis, fault signals are decomposed and features are extracted using wavelet packet decomposition, and potential fault modes are accurately identified through machine learning algorithms. Based on the risk assessment model, optimal maintenance strategy recommendations are generated. The risk assessment model includes three quantitative indicators: severity, probability, and detectability, and the risk is classified. Based on a pre-set expert knowledge base, multiple maintenance strategies are given in combination with the current risk level and fault mode.

[0028] Preferably, in step S1, real-time acquisition of operating data of the submersible electric pump unit and environmental data within the wellbore includes: adopting a distributed sensor network architecture to acquire current, voltage, vibration, and flow data through sensors deployed on the downhole motor, protector, and pump body physical structure; collecting temperature, pressure, flow rate, and sand content data within the wellbore through multiple types of sensors deployed in the wellbore and reservoir layers; and completing data preprocessing, feature extraction, and anomaly detection at the data acquisition end.

[0029] Preferably, in step S2, the reservoir-level model integrates geological data to visualize the pressure field.

[0030] Preferably, in step S2, the model accuracy level is automatically selected based on the working condition characteristics. A reduced-order model is used under normal working conditions, and a high-precision mode is switched when there are abnormalities, thereby improving the simulation speed.

[0031] Preferably, in step S2, a lightweight digital twin model is constructed locally at each well site. That is, the equipment-level model is an adaptive reduced-order finite element model, and the wellbore flow field model adopts empirical formulas to improve computational efficiency. At the same time, the model parameters are uploaded to the unified scheduling center server for analysis and calculation. The results are displayed in the visualization interface of the well group operating condition online monitoring module. The visualization interface is divided into a single-well control interface and a well group status interface. The single-well control interface can observe the operating status of key components in the digital space and use manual instruments to remotely control the electric pump. The well group status interface integrates GIS map markings of the real-time risk level of each single well.

[0032] Preferably, in step S3, a high-fidelity digital twin is created for each submersible electric pump well in the cloud server, integrating equipment-level, wellbore-level, and reservoir-level models. The model is synchronously updated through real-time data streams to ensure that the virtual environment is consistent with the physical equipment status.

[0033] Preferably, in step S3, the failure modes include bearing wear, motor overheating, oil pipe leakage, pump leakage, and pump jamming; the risk level is classified as high risk, medium risk, and low risk; and multiple maintenance strategies include: immediate shutdown, planned maintenance, and continuous monitoring.

[0034] Preferably, in step S3, a configurable virtual experiment module is constructed using a multiphysics coupling simulation platform, which supports user-defined input parameters or automatic import of historical operating condition data.

[0035] Preferably, in step S4, a historical health database is provided, which integrates expert experience, historical maintenance records, equipment parameters, and health condition data to provide a dataset for machine learning algorithms and accurately identify potential failure modes.

[0036] Another object of the present invention is to provide a digital twin-based submersible electric pump well cluster management system, which implements the digital twin-based submersible electric pump well cluster management method, and the system includes:

[0037] The downhole data acquisition and processing module is used to collect real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, and then transmit the collected data to the cloud server after processing.

[0038] The digital twin model construction module, integrated on a cloud server, constructs multi-scale digital twin models from received data to obtain the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics within the wellbore, and geological pressure field data; the multi-scale digital twin model includes equipment-level models, wellbore-level models, and reservoir-level models.

[0039] The well group online monitoring module is used to predict the trends of operating parameters, energy efficiency indicators and health status of multi-scale digital twin models of each well through a web-based 3D visualization platform, so as to realize remote control of single wells and monitoring of well group status.

[0040] The risk warning and decision support module uses machine learning algorithms and expert knowledge bases to correlate trend prediction data with historical health databases, enabling early warning and intelligent diagnosis of equipment failures, and generating maintenance strategies.

[0041] Combining all the above technical solutions, the beneficial effects of this invention are as follows:

[0042] First, the system provided by this invention includes a downhole data acquisition and processing module, a digital twin model construction module, a well group online monitoring module, and a risk warning and auxiliary decision-making module. The downhole data acquisition and processing module uses a distributed sensor network to collect multi-source heterogeneous data from the electric submersible pump unit, wellbore, and reservoir in real time, and performs preprocessing at the acquisition end. The digital twin model construction module establishes a multi-scale dynamic coupling model at the equipment, wellbore, and reservoir levels, supporting adaptive accuracy switching. The well group online monitoring module enables remote control of individual wells and monitoring of the well group status through a 3D visualization interface. The risk warning and auxiliary decision-making module integrates machine learning and an expert knowledge base to achieve intelligent diagnosis and graded early warning based on historical health data. This system solves the problems of data silos, delayed fault prediction, and independent operation of individual wells in existing electric submersible pump management systems, achieving collaborative optimization and intelligent management of the well group.

[0043] Secondly, the system constructs a multi-dimensional digital twin model of the electric pump well group, which, combined with real-time downhole operating data, performs virtual-real verification to achieve online monitoring and can assist managers in decision-making based on a historical health database. This invention enables unified management and intelligent well inspection of multiple well sites, significantly improving the efficiency of oil well production management. Compared with existing technologies, the advantages of this invention also include: existing technologies typically only monitor electric pump operating parameters, while this application combines reservoir pressure, fluid characteristics, geological data, and other multi-dimensional information to construct a more comprehensive monitoring system, improving the accuracy of fault diagnosis; the digital twin model proposed in this invention includes three scales: equipment-level model, wellbore-level model, and well group-level model, which can optimize well group production scheduling and improve overall oil production efficiency; this invention uses a well group operating condition online detection module to display the real-time status of the well group through a three-dimensional digital twin interface, achieving unified visual monitoring of multiple well sites; the risk warning and auxiliary decision-making module proposed in this invention introduces a historical health database, improves fault prediction capabilities through big data models, and the established expert system can provide fault cause analysis and maintenance strategies.

[0044] Third, this invention can significantly reduce unplanned downtime through real-time monitoring, early fault warning, and intelligent diagnosis, while also reducing the cost of manual inspection and emergency repair, thus achieving direct economic benefits. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure;

[0047] Figure 1 This is a flowchart of a digital twin-based submersible electric pump well group management method provided in an embodiment of the present invention;

[0048] Figure 2 This is a wavelet packet decomposition structure diagram provided in an embodiment of the present invention. Detailed Implementation

[0049] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0050] This invention achieves intelligent monitoring, fault early warning, and collaborative optimization of submersible electric pump well groups by constructing a multi-scale dynamically coupled digital twin model. The innovation of this invention lies in:

[0051] (1) A multi-scale digital twin coupling architecture is proposed to solve the data silo problem in traditional single-well monitoring;

[0052] (2) Integrate virtual experiments with real data and locate faults through data comparison;

[0053] (3) By leveraging machine learning and expert databases, fully utilize health data to improve the system's diagnostic capabilities.

[0054] Example 1, as Figure 1 As shown in the embodiment of the present invention, the method for managing submersible electric pump well groups based on digital twins includes:

[0055] S1 collects real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, processes the collected data, and transmits it to the cloud server.

[0056] S2, the cloud server uses an integrated digital twin model building module to construct a multi-scale digital twin model from the received data, and obtains the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics in the wellbore, and geological pressure field data; the multi-scale digital twin model includes: equipment-level model, wellbore-level model and reservoir-level model;

[0057] S3 uses a web-based 3D visualization platform to predict trends in the operating parameters, energy efficiency indicators, and health status of multi-scale digital twin models of each well.

[0058] S4 uses machine learning algorithms and an expert knowledge base to correlate trend prediction data with a historical health database, enabling early warning and intelligent diagnosis of equipment failures and generating maintenance strategies.

[0059] For example, in step S1, the downhole includes a variety of key equipment and components (motor, protector, pump body), and sensors are arranged on its physical structure to collect real-time operating status data of the physical objects. The collected data includes data such as current, voltage, vibration, and flow rate; and environmental data inside the wellbore is collected, including data such as temperature, pressure, flow rate, and sand content.

[0060] For example, in step S1, the downhole data acquisition and processing module adopts a distributed sensor network architecture, which uses multiple types of sensors deployed on the electric pump unit, wellbore and reservoir to collect multi-source heterogeneous data in real time, including current, voltage, vibration, temperature, pressure and flow rate.

[0061] The downhole data acquisition and processing module completes data preprocessing, feature extraction, and anomaly detection at the data acquisition end, improving the real-time performance and reliability of data processing.

[0062] For example, in step S2, when constructing a multi-scale digital twin model through the digital twin model construction module, an equipment-level model is established based on the geometric feature parameters of the submersible electric pump to accurately simulate the mechanical structure and electrical characteristics of the submersible electric pump. The construction process is as follows:

[0063] (1) Based on the design drawings or three-dimensional scanning data of the submersible electric pump, establish a geometric model, including the size, shape and assembly relationship of key components such as stator, rotor, impeller and bearing.

[0064] (2) Assign material properties and electrical parameters to the geometric model, including density, elastic modulus, thermal conductivity, resistance, inductance, etc.

[0065] (3) Perform multi-physics coupling simulation, simulate the stress, deformation and fatigue life of the electric pump under alternating load through finite element analysis, and analyze the temperature field change by combining CFD method.

[0066] (4) Integrate the multiphysics simulation results into the Simulink system-level model to simulate the dynamic response of the submersible electric pump under start-up, shutdown, and load changes.

[0067] For example, in step S2, a wellbore-level model is established based on wellbore environmental data, and the multiphase flow dynamics within the wellbore are reproduced through computational fluid dynamics simulation. The construction process is as follows:

[0068] (i) Establish a three-dimensional wellbore geometric model based on wellbore structural parameters (such as casing diameter and inclination of the deviated section).

[0069] (ii) Set multiphase flow parameters, input real-time sensor data as boundary conditions, including pressure, temperature, sand content, etc., and define fluid properties (such as crude oil viscosity, gas solubility).

[0070] (iii) Multiphase flow CFD simulation was used to calculate the velocity distribution, pressure gradient, and sand deposition in the wellbore.

[0071] The mathematical model for the continuous phase flow consists of the continuity equation and the Navier-Stokers equation:

[0072]

[0073] In the formula, a f ρf is the fluid volume fraction, expressed as %; ρf is the fluid density, expressed as kg / m³. 3 u is the fluid velocity, in m / s; p is the pressure, in Pa; μ is the fluid viscosity, in Pa·s; f b The term represents the source term between the continuous and dispersed phases, expressed in Pa / m; g is the gravitational acceleration, expressed in m / s². 2 t represents time, measured in seconds (s).

[0074] For example, in step S2, the reservoir-level model integrates geological data to visualize the pressure field.

[0075] For example, in step S2, the digital twin models at each level in the digital twin model construction module are dynamically coupled to realize the macroscopic linkage analysis of micro faults. That is, the equipment-level model detects micro anomalies and diagnoses them as fault modes. The equipment-level model transmits the fault parameters to the wellbore-level model, triggering the recalculation of flow velocity and pressure, and then transmits them to the reservoir-level model to correct the pressure field in the near-wellbore zone.

[0076] The digital twin model building module can automatically select the model accuracy level based on the characteristics of the working conditions. For example, a reduced-order model is used under normal working conditions, and a high-precision mode is switched under abnormal conditions to improve the simulation speed.

[0077] The digital twin model construction module builds a lightweight digital twin model locally at each well site (single well). That is, the equipment-level model is an adaptive reduced-order finite element model, and the wellbore flow field model can use empirical formulas to improve computational efficiency. At the same time, the model parameters are uploaded to the unified scheduling center server for analysis and calculation, and the results are displayed in the visualization interface of the well group operating condition online monitoring module.

[0078] For example, in step S3, the well group online monitoring module visually displays the operating parameters, energy efficiency indicators, and health status of each well through a 3D visualization interface, supporting multi-dimensional data comparison analysis and trend prediction. The downhole data acquisition and processing module employs virtual experiment technology, which can simulate the equipment operating status under different conditions in a digital space.

[0079] A high-fidelity digital twin is created for each submersible electric pump well in a cloud server, integrating equipment-level (motor, bearing, impeller), wellbore-level (multiphase flow dynamics), and reservoir-level (pressure field) models. Real-time data streams (such as current, vibration, temperature, and pressure) drive synchronous updates of the models, ensuring consistency between the virtual environment and the physical equipment.

[0080] A configurable virtual experiment module is built using a multiphysics coupling simulation platform (such as COMSOL or ANSYS), which supports user-defined input parameters (such as pump speed, oil viscosity, and sand content) or automatic import of historical operating data.

[0081] Typical fault modes (such as bearing wear and inter-turn short circuits in the motor) are actively injected into the virtual model. By comparing the differences between multi-dimensional sensor data and simulation data, early fault location can be achieved.

[0082] S301. Feature Extraction: Perform wavelet packet decomposition on the vibration signal, filter out most invalid frequency components, and extract energy signal features from multiple sub-bands; calculate the total harmonic distortion rate of the motor current; calculate the temperature gradient based on the thermal grid model of the temperature field.

[0083] S302. Data Comparison: Compare the multi-dimensional data features extracted from the measured data with the feature data obtained from the simulation of the multi-dimensional digital twin model to identify differences.

[0084] S303 Fault Location: By comparing the differences in different frequency bands of the signal, the total harmonic distortion rate of the current, and the temperature gradient, the fault is initially located in different components of the equipment, and then the specific fault type is matched according to the characteristic frequency.

[0085] For example, a vibration sensor captures high-frequency impact signals from a bearing (characteristic frequency > 5kHz), and the equipment-level model diagnoses this as "bearing outer ring spalling" (microscopic fault).

[0086] For example, in step S3, the visualization interface of the well group operating condition online monitoring module is divided into a single well control interface and a well group status interface. The single well control interface can observe the operating status of key components in the digital space and use manual instruments to remotely control the electric pump. The well group status interface integrates GIS map markings to indicate the real-time risk level of each single well.

[0087] For example, in step S4, the risk warning and auxiliary decision-making module integrates machine learning algorithms and expert knowledge bases to achieve early warning and intelligent diagnosis of equipment failures.

[0088] For example, such as Figure 2 As shown, the correlation between real-time monitoring data and historical health database is analyzed using a regression model. Fault signals are decomposed using wavelet packet analysis to extract features, and potential fault modes (such as bearing wear, motor overheating, oil pipe leakage, pump leakage, pump jamming, etc.) are accurately identified using machine learning algorithms. Based on a risk assessment model, optimal maintenance strategy recommendations are generated. The risk assessment model includes three quantitative indicators: severity (production loss due to fault), probability (historical fault frequency and rate of deterioration), and detectability (accuracy of the diagnostic model). Risks are classified (high risk, medium risk, low risk). Based on a pre-set expert knowledge base, and combined with the current risk level and fault mode, various maintenance strategies are provided, such as immediate shutdown, planned maintenance, and continuous monitoring.

[0089] For example, in step S4, the risk warning and auxiliary decision-making module is equipped with a historical health database, which integrates expert experience, historical maintenance records, equipment parameters, and health condition data to provide a dataset for machine learning algorithms and accurately identify potential failure modes.

[0090] The risk warning and auxiliary decision-making module can provide warnings via sound or light.

[0091] Example 2: The digital twin-based submersible electric pump well group management system provided in this embodiment of the invention includes a downhole data acquisition and processing module, a digital twin model construction module, a well group operating condition online monitoring module, and a risk warning and auxiliary decision-making module.

[0092] The downhole data acquisition and processing module is used to collect real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, and then transmit the collected data to the cloud server after processing.

[0093] The digital twin model construction module, integrated on a cloud server, constructs multi-scale digital twin models from received data to obtain the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics within the wellbore, and geological pressure field data; the multi-scale digital twin model includes equipment-level models, wellbore-level models, and reservoir-level models.

[0094] The well group online monitoring module is used to predict the trends of operating parameters, energy efficiency indicators and health status of multi-scale digital twin models of each well through a web-based 3D visualization platform, so as to realize remote control of single wells and monitoring of well group status.

[0095] The risk warning and decision support module uses machine learning algorithms and an expert knowledge base to correlate trend prediction data with a historical health database, enabling early warning and intelligent diagnosis of equipment failures and generating maintenance strategies. This system addresses issues such as data silos, delayed fault prediction, and independent operation of individual wells in existing submersible pump management systems, achieving collaborative optimization and intelligent management of well groups.

[0096] For example, the downhole data acquisition and processing module adopts a distributed sensor network architecture, deploying multiple types of sensors at the electric pump unit, wellbore, and reservoir levels to collect multi-source heterogeneous data such as current, voltage, vibration, temperature, pressure, and flow rate in real time. Preprocessing is then performed at the acquisition end.

[0097] The equipment-level model, wellbore-level model, and reservoir-level model are dynamically coupled at each level to achieve linked analysis of micro-faults and macro-operating conditions.

[0098] For example, the digital twin model building module can automatically select the model accuracy level according to the working condition characteristics. For example, a reduced-order model is used to improve the calculation speed under normal working conditions, and a high-precision mode is switched when there is an abnormality.

[0099] Furthermore, the risk warning and auxiliary decision-making module is equipped with a historical health database, which integrates expert experience, historical maintenance records, equipment parameters and health condition data to provide a training set for machine learning algorithms and improve the accuracy of fault identification.

[0100] To further illustrate the effects of the embodiments of the present invention, the present invention uses "wear of motor bearings leading to sand blockage in the wellbore" as an example to illustrate the linkage analysis process:

[0101] The equipment-level model performs bearing wear detection. In the early stages of motor bearing wear, vibration sensors capture high-frequency impact signals with energy values ​​exceeding normal thresholds. Simultaneously, the digital twin model simulates the increase in bearing clearance through finite element simulation and compares it with the measured vibration spectrum. If the correlation coefficient exceeds the judgment value, bearing wear failure is confirmed. This triggers the wellbore-level model to recalculate flow velocity and sand distribution. As the simulated sediment volume increases, the reservoir-level model modifies the pressure field accordingly. The system comprehensively assesses the risk level and generates a maintenance strategy from a pre-set expert knowledge base: immediately shut down the machine, replace the bearing, and perform well flushing.

[0102] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention and within the spirit and principles of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A management method for submersible electric pump well groups based on digital twins, characterized in that, The method includes the following steps: S1 collects real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, processes the collected data, and transmits it to the cloud server. S2, the cloud server uses an integrated digital twin model building module to construct a multi-scale digital twin model from the received data, and obtains the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics in the wellbore, and geological pressure field data; the multi-scale digital twin model includes: equipment-level model, wellbore-level model and reservoir-level model; The cloud server utilizes an integrated digital twin model building module to construct multi-scale digital twin models from the received data, including: A device-level model is constructed based on the geometric characteristic parameters of the submersible electric pump to simulate its mechanical structure and electrical characteristics. The construction process includes: (1) Based on the design drawings or 3D scanning data of the submersible electric pump, establish a geometric model, including the dimensions, shape and assembly relationship of the stator, rotor, impeller and bearing components; (2) Assign material properties and electrical parameters to the geometric model, including density, elastic modulus, thermal conductivity, resistance, and inductance; (3) Perform multi-physics coupling simulation, simulate the stress, deformation and fatigue life of the electric pump under alternating load through finite element analysis, and analyze the temperature field change by combining CFD method. (4) Integrate the multiphysics simulation results into the Simulink system-level model to simulate the dynamic response of the submersible electric pump under start-up, shutdown and variable load conditions; A wellbore-level model is constructed based on wellbore environmental data, and the multiphase flow dynamics within the wellbore are reproduced through computational fluid dynamics simulation. The construction process includes: (i) Establish a three-dimensional wellbore geometric model based on the wellbore structural parameters; (ii) Set multiphase flow parameters, input real-time sensor data as boundary conditions, including pressure, temperature, and sand content; and define fluid properties, including crude oil viscosity and gas solubility; (iii) Multiphase flow CDF simulation was used to calculate the velocity distribution, pressure gradient, and sand deposition in the wellbore; The continuous phase mathematical model for multiphase flow dynamics is the continuity equation and the Navier-Stokers equation, expressed as follows: In the formula, a f ρ represents the fluid volume fraction, expressed as a percentage (%). f Fluid density, in kg / m³ 3 u is the fluid velocity, in m / s; p is the pressure, in Pa; μ is the fluid viscosity, in Pa·s; f b The term represents the source term between the continuous and dispersed phases, expressed in Pa / m; g is the gravitational acceleration, expressed in m / s². 2 t represents time, in seconds (s). The equipment-level model, wellbore-level model, and reservoir-level model are also dynamically coupled to complete the macroscopic linkage analysis of micro-faults, including: the equipment-level model detects micro-anomalies, diagnoses them as fault modes, and the equipment-level model transmits fault parameters to the wellbore-level model, triggering the recalculation of flow velocity and pressure, which is then transmitted to the reservoir-level model to correct the pressure field in the near-wellbore zone. S3 uses a web-based 3D visualization platform to predict trends in the operating parameters, energy efficiency indicators, and health status of multi-scale digital twin models of each well, supporting multi-dimensional data comparison and analysis and trend prediction. The operating parameters, energy efficiency indicators, and health status of multi-scale digital twin models of each well are predicted using a web-based 3D visualization platform. This includes: simulating equipment operating conditions under different working conditions in a digital space using virtual experiment technology; actively injecting typical fault modes into the virtual model; and completing early fault location by comparing the differences between multi-dimensional sensor data and simulation data. The specific steps are as follows: S301. Feature Extraction: Perform wavelet packet decomposition on the vibration signal, filter out most invalid frequency components, and extract energy signal features from multiple sub-bands; calculate the total harmonic distortion rate of the motor current; calculate the temperature gradient based on the thermal grid model of the temperature field. S302. Data Comparison: Compare the multi-dimensional data features extracted from the measured data with the feature data obtained from the multi-dimensional digital twin model simulation to identify differences; S303 Fault Location: By comparing the differences in different frequency bands of the signal, the total harmonic distortion rate of the current, and the temperature gradient, the fault is initially located in different components of the equipment, and then the specific fault type is matched according to the characteristic frequency. S4, through machine learning algorithms and an expert knowledge base, correlates trend prediction data with a historical health database to achieve early warning and intelligent diagnosis of equipment failures, and generates maintenance strategies; including: By analyzing the correlation between real-time monitoring data and historical health database through regression model analysis, fault signals are decomposed and features are extracted using wavelet packet decomposition, and potential fault modes are accurately identified through machine learning algorithms. Based on the risk assessment model, optimal maintenance strategy recommendations are generated. The risk assessment model includes three quantitative indicators: severity, probability, and detectability, and the risk is classified. Based on a pre-set expert knowledge base, multiple maintenance strategies are given in combination with the current risk level and fault mode.

2. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S1, real-time acquisition of operating data of the submersible electric pump unit and environmental data within the wellbore is performed, including: using a distributed sensor network architecture, current, voltage, vibration, and flow data are collected through sensors deployed on the downhole motor, protector, and pump body physical structure; temperature, pressure, flow rate, and sand content data within the wellbore are collected through multiple types of sensors deployed in the wellbore and reservoir layers; and data preprocessing, feature extraction, and anomaly detection are performed at the data acquisition end.

3. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S2, the reservoir-level model integrates geological data to visualize the pressure field.

4. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S2, the model accuracy level is automatically selected based on the operating condition characteristics. A reduced-order model is used under normal operating conditions, and a high-precision mode is switched when there are abnormalities, thereby improving the simulation speed.

5. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S2, a lightweight digital twin model is constructed locally at each well site. The equipment-level model is an adaptive reduced-order finite element model, and the wellbore flow field model uses empirical formulas to improve computational efficiency. At the same time, the model parameters are uploaded to the unified scheduling center server for analysis and calculation. The results are displayed in the visualization interface of the well group operating condition online monitoring module. The visualization interface is divided into a single-well control interface and a well group status interface. The single-well control interface can observe the operating status of key components in the digital space and use manual instruments to remotely control the electric pumps. The well group status interface integrates GIS map markings to indicate the real-time risk level of each single well.

6. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S3, a high-fidelity digital twin is created for each submersible electric pump well in the cloud server, integrating equipment-level, wellbore-level, and reservoir-level models. The model is updated synchronously through real-time data streams to ensure that the virtual environment is consistent with the physical equipment status.

7. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S3, the failure modes include bearing wear, motor overheating, oil pipe leakage, pump leakage, and pump jamming. The risk is classified as high risk, medium risk, and low risk, and various maintenance strategies include: immediate shutdown, planned maintenance, and continuous monitoring.

8. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S3, a configurable virtual experiment module is constructed using a multiphysics coupling simulation platform, which supports user-defined input parameters or automatic import of historical working condition data.

9. The method for managing submersible electric pump well groups based on digital twins according to claim 1, characterized in that, In step S4, a historical health database is set up, which integrates expert experience, historical maintenance records, equipment parameters, and health condition data to provide a dataset for machine learning algorithms and accurately identify potential failure modes.

10. A digital twin-based management system for submersible electric pump well clusters, characterized in that, The system implements the digital twin-based submersible electric pump well group management method as described in any one of claims 1-9, and the system includes: The downhole data acquisition and processing module is used to collect real-time operating data of the submersible electric pump unit and environmental data inside the wellbore, and then transmit the collected data to the cloud server after processing. The digital twin model construction module, integrated on a cloud server, constructs multi-scale digital twin models from received data to obtain the mechanical structure and electrical characteristics of the submersible electric pump unit, the multiphase flow dynamics within the wellbore, and geological pressure field data; the multi-scale digital twin model includes equipment-level models, wellbore-level models, and reservoir-level models. The well group online monitoring module is used to predict the trends of operating parameters, energy efficiency indicators and health status of multi-scale digital twin models of each well through a web-based 3D visualization platform, so as to realize remote control of single wells and monitoring of well group status. The risk warning and decision support module uses machine learning algorithms and expert knowledge bases to correlate trend prediction data with historical health databases, enabling early warning and intelligent diagnosis of equipment failures, and generating maintenance strategies.

Images