Development method of digital twinning test stand of power system and related device

Abstract

The invention discloses a development method for a digital twinning test bed of a power system and a related device, which are used for generating high-quality representative data through virtual physical twinning and solving the problem of difficulty in data acquisition and management in the prior art. Meanwhile, the test bench provides a safe and controlled test environment, the security and privacy of data are ensured, and the risk of data leakage is avoided. In addition, through a simplified model of virtual physical twinning, the requirements for high-precision hardware and computing resources are reduced, and therefore the technical cost is reduced. The invention also designs a flexible system framework, which supports integration and cooperative work between different models and platforms, and promotes standardization and interoperability. Through an adaptive process and a real-time updating mechanism, the modeling complexity is optimized, and the dynamic nature and accuracy of the digital twin model are improved. And finally, an efficient computing architecture and an algorithm are adopted, and real-time performance and performance bottlenecks are broken through in combination with virtual physical twinborn real-time data streams.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a development method and related devices for a power system digital twin test bench. Background Art

[0002] Digital twin technology is an innovative application based on the Internet of Things (IoT), big data, artificial intelligence (AI), and cloud computing. It aims to build virtual models of physical devices, systems, or processes to enable real-time monitoring, predictive maintenance, and optimized operations. Its development is driven by multiple factors, including the widespread adoption of IoT, advancements in big data, innovations in AI algorithms, and the growth of cloud computing. Digital twin technology has a wide range of applications, encompassing areas such as product lifecycle management, equipment monitoring, and urban planning.

[0003] However, existing technologies in the field of digital twins face numerous shortcomings. First, data acquisition and management face challenges. Digital twin technology relies on large amounts of real-time data input, but obtaining high-precision data is difficult. Hardware equipment may not meet requirements, and data formats between different devices are incompatible, making integration difficult. Data accuracy and real-time performance are crucial, but in practice, data loss, errors, and delays are common. Second, data security and privacy are prominent issues. Digital twin systems involve large amounts of sensitive data, and data leaks or malicious tampering can have serious consequences. Furthermore, the technology is expensive, and building and maintaining high-precision digital twin models requires extensive computing resources, placing a heavy burden on enterprises. Furthermore, the lack of unified standards and interoperability poses challenges, making integration and collaboration between different systems and platforms difficult. Modeling complexity is also a significant issue. Current modeling technologies struggle to fully account for complex physical factors, making dynamic updates and high-precision modeling difficult. Finally, real-time performance and performance bottlenecks hinder the development of digital twin technology. Achieving real-time performance requires extremely high computing power, but existing resources and algorithms face performance bottlenecks when processing large amounts of data. Summary of the Invention

[0004] The present invention provides a method for developing a power system digital twin test bench and related devices, which are used to create an efficient, safe, flexible and economical solution for power system digital twin applications.

[0005] In view of this, a first aspect of the present invention provides a method for developing a power system digital twin test bed, the method comprising:

[0006] S1. Interactive verification of virtual physical twins and power system digital twins:

[0007] Build a virtual physical twin model to generate representative data, design the system framework, and create an intermediate mapping chain;

[0008] Verifying the initialization function of the power system digital twin through online fault identification;

[0009] Achieving bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification;

[0010] S2. Deploy the power system digital twin and form a closed-loop test:

[0011] Interconnecting the power system digital twin with the virtual physical twin to simulate real events and test sampling rate, communication delay, and measurement error scenarios;

[0012] Reference to ICT data processing methods;

[0013] Modify parameters based on the actual system status to generate multiple scenarios, inject noise and adjust the sampling rate to simulate the real environment;

[0014] S3. Deployment to the actual power system:

[0015] Connecting the verified power system digital twin to the actual power system and reusing the system framework and the intermediate mapping chain for initialization;

[0016] Tracking changes in the actual power system in real time by initiating an adaptive process, and receiving actual data to optimize the digital twin of the power system;

[0017] The analysis response and service of the actual power system are enhanced based on the output of the power system digital twin.

[0018] Optionally, verifying the initialization function of the power system digital twin through online fault identification includes:

[0019] Establish a fault library based on the normalized residuals of different fault types;

[0020] A normalized residual is calculated, abnormal behavior of the physical twin is detected based on the normalized residual, and a fault alarm is triggered when the abnormal behavior is detected.

[0021] Optionally, the bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin is achieved through transient response verification, including:

[0022] The digital twin model is updated by an adaptive process that is time-driven, event-driven, or a hybrid of the time-driven and event-driven processes, and the update mode is adjusted based on the state error between the virtual physical twin and the power system digital twin.

[0023] Optionally, the adaptive process is expressed as:

[0024] ;

[0025] Where, 1 and 2 is the dynamic threshold and k is the index.

[0026] Optionally, the reference communication data processing method includes:

[0027] The communication data processing methods used include measurement compensation, measurement amplification and measurement expansion, among which:

[0028] The measurement compensation is used to replace the physical measurement with the estimated value of the power system digital twin when the physical measurement is missing;

[0029] The measurement amplification is used to generate a high sampling rate data set;

[0030] The measurement extension is used to predict non-measured variables through the power system digital twin.

[0031] Optionally, predicting non-measured variables by using the power system digital twin includes:

[0032] The electrical variables of the medium voltage side distribution transformer are predicted through the power system digital twin.

[0033] Optionally, the bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin is achieved through transient response verification, and then further includes:

[0034] The key functions of the power system digital twin are verified so that the power system digital twin tracks the behavior of the virtual physical twin under steady-state operation and transient operation.

[0035] A second aspect of the present invention provides a development system for a power system digital twin test bench, the system comprising:

[0036] Interactive verification unit, used for interactive verification of virtual physical twins and power system digital twins, including:

[0037] Build a virtual physical twin model to generate representative data, design the system framework, and create an intermediate mapping chain;

[0038] Verifying the initialization function of the power system digital twin through online fault identification;

[0039] Achieving bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification;

[0040] A test unit, used to deploy the power system digital twin and form a closed-loop test, including:

[0041] Interconnecting the power system digital twin with the virtual physical twin to simulate real events and test sampling rate, communication delay, and measurement error scenarios;

[0042] Reference to ICT data processing methods;

[0043] Modify parameters based on the actual system status to generate multiple scenarios, inject noise and adjust the sampling rate to simulate the real environment;

[0044] Deployment unit, used for deployment to actual power systems, including:

[0045] Connecting the verified power system digital twin to the actual power system and reusing the system framework and the intermediate mapping chain for initialization;

[0046] Tracking changes in the actual power system in real time by initiating an adaptive process, and receiving actual data to optimize the digital twin of the power system;

[0047] The analysis response and service of the actual power system are enhanced based on the output of the power system digital twin.

[0048] A third aspect of the present invention provides a device for developing a digital twin test bench for a power system, the device comprising a processor and a memory:

[0049] The memory is used to store program code and transmit the program code to the processor;

[0050] The processor is used to execute the steps of the method for developing a digital twin test bench for a power system as described in the first aspect above according to the instructions in the program code.

[0051] A fourth aspect of the present invention provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the development method of the power system digital twin test bench described in the first aspect above.

[0052] It can be seen from the above technical solutions that the present invention has the following advantages:

[0053] 1) In existing technologies, obtaining high-precision data is difficult. Data formats from different devices are incompatible, making integration difficult, and data loss, errors, and delays are frequent. This invention, however, generates high-quality representative data through virtual physical twins, reducing reliance on actual system data while improving data accuracy and real-time performance, effectively resolving data acquisition and management challenges.

[0054] 2) In existing technologies, digital twin systems involve large amounts of sensitive data, and data leakage or malicious tampering can have serious consequences. This invention provides a secure and controlled testing environment, ensuring the security and privacy of sensitive data and reducing the risk of data leakage.

[0055] 3) In existing technologies, building and maintaining high-precision digital twin models requires extensive computing resources, placing a heavy burden on enterprises. This invention reduces the need for high-precision hardware and computing resources through a simplified model of virtual physical twins, thereby lowering development and maintenance costs.

[0056] 4) Existing digital twin technologies lack unified standards, resulting in poor interoperability between different systems and platforms, making system integration and collaboration difficult. This paper designs a flexible system framework that supports integration and collaboration between different models and platforms, promoting standardization and interoperability.

[0057] 5) Current modeling techniques struggle to fully account for complex physical factors, making dynamic updates and high-precision modeling difficult. This invention optimizes modeling complexity through an adaptive process and real-time update mechanism, improving the dynamics and accuracy of digital twin models and enabling them to better reflect the complexity of physical systems.

[0058] 6) In existing technologies, achieving real-time performance requires extremely high computing performance, but existing resources and algorithms face performance bottlenecks when processing large amounts of data. This invention utilizes a highly efficient computing architecture and algorithm, combined with the real-time data flow of virtual and physical twins, to overcome these real-time and performance bottlenecks and enhance the system's real-time performance and performance.

[0059] In summary, the present invention creates an efficient, safe, flexible and economical solution for the application of digital twins in power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 A schematic diagram of a process for developing a power system digital twin test bed according to an embodiment of the present invention;

[0062] Figure 2 The laboratory setup and data flow of the power system digital twin test bed provided by the embodiment of the present invention;

[0063] Figure 3 The rotor angle separation between generators at different positions under stable conditions provided by the embodiments of the present invention;

[0064] Figure 4 The embodiment of the present invention provides for the separation of rotor angles between generators at different positions in an unstable situation;

[0065] Figure 5 The stability of the real-time transient stability assessment based on the proposed test bench provided in the embodiment of the present invention;

[0066] Figure 6 The unstable situation of the real-time transient stability assessment based on the proposed test bench provided by the embodiment of the present invention;

[0067] Figure 7 A schematic structural diagram of a development system for a power system digital twin test bench provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0069] See also Figure 1 , a method for developing a power system digital twin test bed provided in an embodiment of the present invention includes:

[0070] Step 101: Interaction verification between the virtual physical twin and the power system digital twin:

[0071] Step 1011: Build a virtual physical twin model to generate representative data, design a system framework, and create an intermediate mapping chain.

[0072] It should be noted that the virtual physical twin bridges the physical and virtual worlds. Through data interaction and model mapping, it enables precise mapping of the physical system in the virtual space and real-time feedback. In the development of a power system digital twin testbed, the construction of the virtual physical twin model requires detailed modeling of various power system devices, components, and operating states to ensure that the virtual model accurately reflects the behavior of the physical system. Sensor technology and data acquisition systems are used to capture real-time operating data from the physical system and transfer it to the virtual model for analysis and processing. By continuously updating and optimizing the virtual model, it maintains a high degree of consistency with the physical system. This allows for effective verification of the accuracy and reliability of the power system digital twin model during interactive verification between the virtual physical twin and the power system digital twin, providing strong support for the safe and stable operation of the power system.

[0073] It can be understood that the present invention generates high-quality representative data through virtual physical twins, reducing dependence on actual system data, while improving the accuracy and real-time nature of the data, and effectively solving the problems of data acquisition and management; specifically, the representative data generated by virtual physical twins will undergo multiple tests, such as load change tests, fault simulation tests, extreme condition tests, etc.

[0074] It should be noted that simplifying the process from virtual physical twin to digital twin and then to physical system requires the design of a repeatable system framework in which different digital twin instances can be initialized at any time. The system framework uses unique identifiers to replace the initial states of the parameters in the system framework, and then populates these initial states with variables obtained from different instances of the virtual physical twin to maintain consistency with their initialization.

[0075] It is important to note that an intermediate mapping link is created to update and track the initial states of different digital twin instances. A common method for initializing digital twins in power system applications is to perform a power flow calculation. The digital twin's behavior is then compared to its virtual physical twin in terms of both power flow calculations and steady-state operation. If the behavior between the virtual physical twin and the digital twin differs significantly after aligning the framework and mapping files with the virtual physical twin, it is necessary to confirm that all relevant parameters have been updated for digital twin initialization. This requires the use of online fault identification, as described in step 1012 below.

[0076] Step 1012: Verify the initialization function of the power system digital twin through online fault identification, including: establishing a fault library based on the normalized residuals of different fault types; calculating the normalized residuals, detecting abnormal behavior of the physical twin based on the normalized residuals, and triggering a fault alarm when abnormal behavior is detected.

[0077] It should be noted that the corrective measures applied to the physical twin by online fault identification improve the system operation, thereby minimizing the difference with the digital device. Online identification of physical twin faults based on digital twin measurements uses a digital twin with high parallelism to generate reference behavior with the physical system operation. When unexpected behavior is detected in the physical twin based on the measurement results of the digital twin, measures can be taken to remedy the physical twin. Remedial measures may be taken in the face of complex control strategies and time-varying conditions. By evaluating the residual error between the physical measurement and the digital twin output, the erroneous behavior of any asset, such as power electronics-based systems and their subsystems, can be detected and identified. When an error corresponding to the physical system is identified, an alarm is issued and a maintenance plan is generated. Fault Library F i ={F1, F2, ..., F N}, can be established based on the normalized residuals of different fault types Γ: ε[k]. Their sum is expressed as: <ε[k], F i >=∑k k-Δkε T [kj]F i [j], where Δk is the sampling interval and j is the summation index, indicating that the fault type Γ is:

[0078] (1)

[0079] Where, is a threshold value that depends on the system configuration, and i is the fault type.

[0080] It can be understood that when <ε[k], Fi> is greater than the threshold , a fault of type Fi can be determined to have occurred. This fault identification method, based on normalized residual and threshold comparison, can accurately capture anomalies in physical systems. Once a fault is identified, the system can quickly respond according to pre-defined policies. Therefore, this invention provides a secure and controlled testing environment, ensuring the security and privacy of sensitive data and reducing the risk of data leakage.

[0081] Step 1013: Realize bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification, including: updating the digital twin model using an adaptive process driven by time, event, or a hybrid of time and event, and adjusting the update mode based on the state error between the virtual physical twin and the power system digital twin.

[0082] It should be noted that transient response verification is provided by establishing bidirectional communication (e.g., via Ethernet) to interconnect the virtual physical twin and the digital twin in real time, enabling online data flow. Synchronization and adaptation capabilities are also tested interactively by making predefined changes in the virtual physical twin and confirming the state of the digital twin. Various adaptation processes can be used, such as time-driven, event-driven, and time-event-driven.

[0083] The adaptive process refers to the ability of the digital twin to learn and improve its accuracy and performance by continuously updating and improving models and algorithms based on real-world data. Adaptation requires attention to the physical twin data (Y PT ) and digital twin data (Y DT ), identify the source of error, and then compensate and correct it. The digital twin deviates from its physical counterpart but is then able to self-improve and behave like a physical twin. Adaptation is typically achieved in real time or through event-triggered big data techniques with minimal or no user input. This function analyzes Y PT [k]={Y PT1 ,…,Y PTN}, and Y DT [k]={Y DT1 ,…,Y DTN}, and obtain their differences in real time by: ε[k]=|Y DT [k]−Y PT [k]|, where index k usually corresponds to a time sample value. The mixing mode of the adaptive process is determined by the following factors:

[0084] (2)

[0085] in, 1 and 2 is the dynamic threshold ( 1 2) Depending on the system configuration, time-driven, event-driven, or a combination of these can be used for digital twin self-adaptation. Appropriate thresholds balance the efficiency of digital twin implementation and the accuracy of behavior. The thresholds should be determined based on: (i) the error tolerance level, (ii) the errors introduced by the modeling tool, and (iii) the accuracy of measurement and data acquisition.

[0086] It can be understood that the present invention optimizes the modeling complexity and improves the dynamics and accuracy of the digital twin model through adaptive processes and real-time update mechanisms, so that it can better reflect the complexity of the physical system; and adopts an efficient computing architecture and algorithm, combined with the real-time data flow of the virtual physical twin, breaking through the real-time and performance bottlenecks, and improving the real-time and performance of the system.

[0087] Step 1014: Verify the key functions of the power system digital twin so that the power system digital twin tracks the behavior of the virtual physical twin under steady-state operation and transient operation.

[0088] It should be noted that, specifically, the key functions of the digital twin, such as initialization and adaptation, are verified by Test1 and Test2 in the following formula, and the difference is expected to be less than ε T1 and ε T2 ,|(Y DT [k+1]-Y DT [k]) / Δk|<σ defines the state of the power system, where σ represents the threshold, so:

[0089] (3)

[0090] The two-step validation (Test 1 and Test 2) ensures that the developed digital twin instance can track the behavior of the virtual-physical twin in both steady-state and transient operations.

[0091] Step 102: Deploy the power system digital twin and perform closed-loop testing:

[0092] Step 1021: Interconnect the power system digital twin with the virtual physical twin to simulate real events and test the sampling rate, communication delay, and measurement error scenarios.

[0093] It's important to note that by interconnecting the virtual physical twin and the digital twin, simulating the implementation of real-world events, validating and optimizing digital twin-related solutions, and completing the cycle of evaluating digital twin-based solutions on the virtual physical twin without jeopardizing the operation of any actual systems, the first step from establishing the virtual physical twin is completed. Because the virtual physical twin provides online data streaming alongside the digital twin, validated services are expected to perform as expected when deployed on the actual system. The closed-loop solution based on the virtual physical twin simulates the actual implementation, including measurement, analysis, communication, and actuation. Multiple changes introduced to the virtual physical twin to simulate real-world events drive the responses of the digital twin-based services. Furthermore, the controlled environment based on the virtual physical twin allows testing of various settings that may occur in actual deployments, such as sampling rates, communication delays, and measurement errors, thereby reducing testing risks.

[0094] Taking into account various devices (such as phasor measurement devices, remote terminal units), communication protocols and operating conditions, different configurations will be considered. Therefore, it is necessary to utilize the measurement compensation, measurement amplification, and measurement extension methods in the communication data processing method of the power system digital twin, as specifically described in step 1022.

[0095] Step 1022: citing the communication data processing method, including:

[0096] 1) Measurement compensation, which is used to replace physical measurements with estimated values ​​from the power system digital twin when physical measurements are missing;

[0097] It's important to note that power system measurements are critical for ensuring reliable and efficient grid operation and planning. These measurements are used to monitor, control, and protect the network. Inaccurate or missing data can pose significant challenges to power systems (e.g., unreliable and inefficient operation, safety risks, etc.). Digital twins can play a key role in: (i) detecting inaccurate and / or missing measurements, (ii) replacing and / or compensating for such data, and (iii) using combined data streams in power system applications to maintain operability.

[0098] Different types of measurement degradation (e.g., signal loss, noise, sensor aging, communication failures) have unique characteristics that can be identified by the digital twin, allowing compensation for such errors and mitigating any service / application downtime. The digital twin's estimated value is used to replace the missing physical measurement. The general compensation logic for a specific measurement is:

[0099] (4)

[0100] Where, is the typical threshold designed for each measurement.

[0101] It can be understood that the problem of data loss is solved by measuring compensation.

[0102] 2) measurement amplification, used to generate high sampling rate data sets;

[0103] It should be noted that relevant information from power system measurements (such as remote terminal units and / or phasor measurement units (PMU)) and other sources (such as the National Weather Service, the Energy Bureau) is collected from a large number of sensors and devices. This information is used to create highly detailed models in digital twins, update such models, and provide information for advanced digital twin applications. The performance, effectiveness, and timeliness of these applications are closely related to the input data, especially its resolution (or sampling rate). Digital twins can generate and provide different data sets with the required resolution, thereby enhancing a wide range of applications, especially those that require higher resolution to illustrate the rapid dynamics of modern power systems. When power system measurements are amplified by digital twins, more data points are provided, which can be used to predict the behavior of the system or implement online stability analysis. The series of data Y[k] sampled at different time intervals from the physical twin and the digital twin can be described as:

[0104] (5)

[0105] Where, T i is the sampling interval, T1=nT2, and ( )

[0106] It is understandable that data resolution is improved by measuring magnification.

[0107] 3) Measurement extension, used to predict non-measured variables through power system digital twins.

[0108] It should be noted that digital twins provide an alternative solution to sensors, leveraging the observability and visualization of the grid by estimating non-measured variables. With the support of digital twins, additional data can be generated without deploying any additional physical instruments. Some measurements are obtained from the physical twin, while several other measurements are provided by the digital twin, for example, the electrical variables of the medium-voltage side distribution transformer can be obtained through the digital twin by measuring the low-voltage values. Digital twins can expand the measurements of the distribution network to support near real-time operations and services. Digital twins can also predict the renewable energy generation of the distribution network while reducing the demand for real-time communication infrastructure in the control room. Therefore, the observable output can be expanded and expressed as:

[0109] (6)

[0110] It is understandable that scaling by measurement avoids the deployment of additional hardware.

[0111] Step 1023: Modify parameters based on the actual system state to generate multiple scenarios, inject noise, and adjust the sampling rate to simulate the real environment.

[0112] It's important to note that digital twins are powerful platforms for generating additional data and conducting scenario testing (e.g., what-if scenarios). Generating additional data creates large datasets for various power system scenarios and conditions without impacting or endangering the actual grid. This data in the digital twin can then be used to construct new datasets, supplement (historical) measurements, and / or continue to expand the database for training and validating AI-based applications. Scenario generation is considered the parallel simulation of different power system conditions at faster than or real-time speeds to predict power system behavior, address operational uncertainty, improve system security, and / or prevent severe losses during critical events. Multiple predetermined scenarios are executed in parallel under realistic operating conditions to tailor actions to the system's response. Once a specific scenario is realized, a series of actions can be directly and automatically executed to reduce response time. Estimating stability margins after a severe incident while minimizing load shedding is one use case for using digital twins as a platform for generating additional data and conducting scenario testing. When starting with the actual state of the grid, the actual state of the grid can be modeled as:

[0113] (7)

[0114] Multiple scenarios run in the digital twin are created by changing certain parameters and inputs as follows:

[0115] (8)

[0116] Where ẑ refers to {A, B, C, D, u}, and the generated scene is defined as S i =[S1,S2,…,S N ] T , where S i It means through z i 'The output of the example generated by the digital twin defined by the i-th one. For example, the actual output of the virtual physical twin can be modified to cover different levels of measurement noise and / or sampling rates for real-world scenarios as follows:

[0117] (9)

[0118] Where μ is the measurement noise and n is the sampling rate.

[0119] Step 103: Deploy to the actual power system:

[0120] Step 1031: Connect the verified power system digital twin to the actual power system, and reuse the system framework and intermediate mapping chain for initialization.

[0121] It's important to note that once the digital twin's functionality is verified, the virtual physical twin can be replaced with the actual power system. All configurations, data flows, and information can be adjusted based on the actual situation and use case. The digital twin is initialized using the same (reused) framework and mapping links, reducing development and maintenance costs.

[0122] Step 1032: Track the changes of the actual power system in real time by starting the adaptive process, and receive actual data to optimize the digital twin of the power system.

[0123] It should be noted that the adaptive process is also initialized to track the changes in the power system and change the system parameters accordingly. In this stage, the digital twin receives real-time data from the actual power system to perform continuous adaptation, as shown in Equation (3).

[0124] Step 1033: Enhance the analysis response and service of the actual power system based on the output of the power system digital twin.

[0125] It should be noted that the digital twin-based system is connected to the actual power system, and its test performance on the virtual physical twin is expected to be comparable, which enables the analytical response of the digital twin to enhance actual applications and services.

[0126] Furthermore, it is important to note that a unique architecture has been created by combining software, hardware, and data flows to implement the functionality of a power system digital twin. This architecture, which includes essential components such as a database, data processing modules, and communication infrastructure, is adaptable and can be easily modified to accommodate the needs of other power systems and their respective digital twins. The testbed and workflow address these issues and validate the updating, expansion, and replacement of digital twins. The processes and cycles for performing such maintenance work in the field are optimized. The testbed, integrated with multiple digital twins, provides a unique opportunity to understand the potential interactions and synergies between them, thereby achieving reliable, efficient, and secure power system digital twins.

[0127] See also Figure 2-Figure 6 The following is a verification and case study of the development method of the power system digital twin test bed provided by an embodiment of the present invention:

[0128] Following the principles established in the proposed development methodology, the following example demonstrates the use of a digital twin testbed developed for real-time transient stability assessment (RTSTA). This example demonstrates the development of a virtual physical twin, a digital twin, and a complete testbed for validation and comparison. The use of RTSTA is motivated by its importance for power system analysis, the extensive literature on offline and online approaches, and the well-documented effectiveness of various AI-based algorithms proposed for this application. All of this makes it an ideal scenario for demonstrating the value of the proposed testbed in power system applications.

[0129] 1. Hardware and network model:

[0130] The digital twin testbed was developed in the laboratory using eight real-time digital simulator racks based on the PB5 simulation processing board; four served as virtual physical twins and four served as digital twins of the power system. Measurements and external communication with databases, services, and other devices were implemented using a combination of analog / digital I / O interfaces and communication cards. Specifically, real-time data transmission from workstation to workstation was based on the IEEE 754 standard. The testbed employed three different sampling rates: (i) high-resolution data captured in an oscilloscope for local storage of variables at 50 μs, (ii) 10–50 Hz measurement frequencies based on phasor measurement devices, and (iii) 0.25–0.5 Hz for supervisory control and data acquisition (SCADA) measurements and equipment.

[0131] The AI-based algorithms were implemented in TensorFlow software and trained using a Python-based application programming interface that accessed datasets stored in a database. All computations in the remainder of this paper were performed on a desktop workstation equipped with a 2.9 GHz Intel Core i7-10700 CPU and 16 GB of RAM. This workstation connected to the virtual physical twin and the digital twin in real time via TCP / IP.

[0132] The power system network is based on the IEEE benchmark model, which has been used in different studies. The model consists of 5 strongly interconnected areas, 14 generators, 5 static var compensators (SVCs), 59 buses and 104 transmission lines. The model includes meshed areas and long, relatively weak and loosely interconnected sections, which can be flexibly modified to include inverter-based generation to simulate the growth of renewable energy in the power system and can be replicated twice in a hardware simulator available in the laboratory. Area 4 was selected as the area of ​​interest for measurements at selected buses (i.e. buses 405, 407, 408, 410 and 414). As an alternative to simplify the testbed setup, two identical models can be used as virtual physical twin and digital twin to ensure comparability in terms of accuracy, fidelity and behavior. This approach eliminates uncertainties due to differences between the models. The virtual physical twin provides online measurements to test and verify the performance of the digital twin. The final structure and setup of the virtual testbed are shown in Figure 4. Figure 2 The workflow for developing a real-time transient stability assessment service based on the proposed power system digital twin testbed is shown below.

[0133] Step 1: Establishment of the power system digital twin test bed, which includes the following sub-steps: Sub-step 1: Determine available data and models; Sub-step 2: Check resources and their functions; Sub-step 3: Build digital twins (such as mapping files and frameworks); Sub-step 4: Establish virtual physical twins using necessary data flows; Sub-step 5: Customize measurements, communications, etc.; Sub-step 6: The virtual test bed includes virtual physical twins and digital twins.

[0134] Step 2: Import the measurement values ​​and input values ​​from the virtual physical twin into the mapping file and framework in the digital twin for initialization.

[0135] Step 3: Synchronize the interaction between the digital twin and the virtual physical twin.

[0136] Step 4: The digital twin generates a data set according to the preset script.

[0137] Step 5: Use artificial intelligence algorithms for offline training, with the objective function being to minimize the cost function loss and maximize the accuracy until an acceptable accuracy result is output.

[0138] Step 6: Import input values, different operating conditions, and interaction events into the virtual physical twin for real-time online application testing until valid values ​​are output.

[0139] Step 7: Input the effective value into the actual power system to check whether it operates stably.

[0140] 2. Test bench initialization:

[0141] Initialization of the proposed testbed's digital twin portion is performed using a reusable "framework" based on the network model and a mapping file that maps all unique model identifiers to their measured values ​​from the virtual physical twin once the virtual physical twin reaches its steady state. Variables prioritized when defining the system framework include circuit breaker states, the modes of the various controllers in the model's components, and the voltage and active power setpoints of the generators. Because a single virtual physical twin can support multiple digital twin instances, each digital twin instance has a unique mapping file to store its initial state. (In our example, the unique identifiers follow the format $AAABBBCCD', where "$" is the title, "AA" defines the model number, "BBB" defines the component type, "CC" defines the individual component number, and "D" is used for the name of each variable. For example, the active power output of generator 9 in model 1 is assigned the identifier "$01GEN09P.")

[0142] Verification of correct digital twin initialization is also performed at this stage. If the difference between the measured virtual physical twin values ​​and the initialized digital twin values ​​is within a predefined acceptable range (e.g., active and reactive power output of each generator, voltage and angle of each bus), the initialization process is complete. However, if a large mismatch is found between the initial state of the digital twin and the measured values ​​of the virtual physical twin, additional verification and initialization steps should be taken. The initial parameters of the system framework and mapping files are updated to maintain consistency with the measured values ​​of the virtual physical twin. This step can also be used for the adaptation process, however, this is beyond the scope of the current example of this invention.

[0143] 3. Synchronization:

[0144] The digital twin is updated using a combination of time-based and event-based approaches. Under steady-state conditions, the digital twin is updated periodically and the state of the system is evaluated; in the proposed example, this occurs every 0.5 seconds. Changes in certain system variables in the virtual physical twin trigger an event-based update of the digital twin to reflect the observed changes. Other updates may be event-driven (e.g., faults, circuit breaker operations, etc.). These variables can be selected based on the needs of each application to minimize the computational and communication overhead of the testbed (in the case of a real-time transient stability assessment, the selected variables include (i) the state of the 14 circuit breakers in region 4 (on / off), (ii) the voltage and active power setpoints (actual values ​​of the setpoints) for all 14 generators, (iii) the control mode settings for each of the 14 generators (e.g., locked / free speed mode), (iv) the setpoints for the 14 governors, (v) the control settings for the five SVCs, (vi) the regulator modes for the five SVCs, and vii) the settings for each application fault during training (e.g., location and duration).

[0145] 4. Data generation and offline training:

[0146] A key step in the development of a real-time transient stability assessment application is data generation and offline training of the algorithm. Based on a predefined set of events and scenarios, measurements from five locations and the corresponding rotor angle intervals between any two generators in the system are recorded in a database. Six load levels are considered, corresponding to conditions ranging from peak to light load. Furthermore, each base load level has been increased by 5% from 90% to 110%. At various locations throughout the network (F loc. ) Consider a three-phase ground fault; at three locations (e.g., 0%, 25%, 50%, 75%, 100%) on each transmission line and its terminal, with a duration of 5 cycles (F t Therefore, multiple scenarios (S VPT ),like:

[0147] (9)

[0148] The power of the load and the generator are: P L , Q L , P G and Q G ∈[90%, 95%, 100%, 105%, 110%], and F loc. ∈[0%, 25%, 50%, 75%, 100% of the transmission line length].

[0149] The transient stability assessment is performed based on the rotor angle interval. If any of the measured rotor angle intervals exceeds 180°, the system is marked as unstable (displayed as 1 in the test bench results in this case), otherwise it is considered stable (marked as 0 in the test bench results). Figure 3 and Figure 4 The rotor angle separations obtained from the digital twin for stable and unstable conditions are shown, respectively. A total of 1,785 cases were generated; they were divided into training and testing groups with a ratio of 7:3. To demonstrate the proposed testbed for comparative purposes, two different AI-based algorithms were selected: (i) an artificial neural network and (ii) a support vector machine. These algorithms were trained to provide the required accuracy based on previous work on developing accurate offline transient stability assessment algorithms (i.e., for the present case, an accuracy of >97.5% was sufficient).

[0150] 5. Real-time transient stability assessment:

[0151] In steady state, data is measured from the virtual physical twin, and the AI-based algorithm is executed every 0.5 seconds using a time-based approach. When an event (i.e., a fault in the system) triggers an event-driven update, the algorithm switches to event-driven execution. Five consecutive post-emergency measurements are acquired to predict transient instabilities. The evaluation results are communicated to the virtual physical twin so that appropriate measures can be taken. The output of the virtual physical twin can replicate different sampling rates and measurement noise, as shown in Equation (8). In this case study, additional measurement noise and communication delays are not considered.

[0152] Figure 5 and Figure 6 Results for both stable and unstable scenarios are shown separately. Before the rotor angle separation returns to a stable state or is detected exceeding 180°, the virtual-physical twin receives a stability alert indicating potential system behavior. Two AI-based algorithms were also compared by reproducing the test scenario in the digital twin. In both cases, the fault occurred at 2 seconds. The support vector machine-based approach was found to require less time (0.0362 seconds) than the artificial neural network-based solution to complete the real-time transient stability assessment and generate a stability alert after the unexpected event. The results of the real-time transient stability assessment are summarized in Table 1.

[0153] Table 1 Summary of real-time transient stability assessment results

[0154]

[0155] 6. Results:

[0156] Using a real-time transient stability assessment application as an example, this paper demonstrates the value of a testbed based on virtual-physical twins for the further development of power system digital twins. Specifically, a sample workflow is provided, from setting up a testbed to initializing multiple digital twin instances. The digital twin instances are initialized and synchronized based on the current operating conditions of the virtual-physical twin.

[0157] Using two machine learning algorithms as examples, the testbed's practical application is briefly illustrated, rather than focusing on the development of new algorithms. The performance of the machine learning algorithms was tested and validated offline using a dataset (test set) generated by the digital twin. They were tuned to achieve acceptable accuracy before being deployed on the virtual physical twin for online applications. Two independent three-phase ground faults of 0.1s duration were applied to test the online application. On the one hand, when a fault was applied at 5% of the length of the transmission line between buses 414 and 415 (see Figure 5 On the other hand, when a fault is applied at 5% of the length of the transmission line between buses 406 and 407, as shown in FIG. Figure 6 shown.

[0158] The transient stability assessment relies on the rotor angle separation between any two generators. If any rotor angle separation exceeds 180 degrees after a fault occurs, it is labeled "unstable"; otherwise, it is a stable condition. In both cases, the support vector machine-based solution provides faster prediction results than the artificial neural network-based method, as shown in Table 2. Furthermore, the artificial neural network requires a longer time to evaluate the post-emergency response (5.2 ms and 6.7 ms for the stable and unstable cases, respectively) compared to the pre-emergency operation. In the unstable case, the first rotor angle separation exceeding 180 degrees occurs 2.9274 s after the fault occurs. The difference in execution time between the two algorithms (0.0362 s) can be taken into account when implementing corrective measures for instability.

[0159] The real-time transient stability assessment case allows validating algorithms in a laboratory test environment, generating data for multiple events in a consistent and reproducible way, and comparing different algorithms, scenarios, settings, hardware setups, and services before deployment on the actual power system.

[0160] Table 2 Detailed results of real-time transient stability assessment

[0161]

[0162] The embodiment of the present invention provides a method for developing a digital twin test bed for a power system, which generates high-quality representative data through virtual physical twins, solving the problem of difficult data acquisition and management in the existing technology. At the same time, the test bed provides a safe and controlled testing environment, ensuring the security and privacy of the data and avoiding the risk of data leakage. In addition, the simplified model of the virtual physical twin reduces the demand for high-precision hardware and computing resources, thereby reducing the technical cost. The present invention also designs a flexible system framework to support the integration and collaboration between different models and platforms, and promotes standardization and interoperability. Through adaptive processes and real-time update mechanisms, the modeling complexity is optimized, the dynamics and accuracy of the digital twin model are improved, and it better reflects the complexity of the physical system. Finally, by adopting an efficient computing architecture and algorithm, combined with the real-time data stream of the virtual physical twin, the real-time and performance bottlenecks are broken through, and the real-time and performance of the system are improved.

[0163] The above is a development method of a power system digital twin test bench provided in an embodiment of the present invention. The following is a development system of a power system digital twin test bench provided in an embodiment of the present invention.

[0164] See also Figure 7 , a development system of a power system digital twin test bench provided in an embodiment of the present invention includes:

[0165] The interactive verification unit 201 is used for interactive verification of the virtual physical twin and the power system digital twin, including:

[0166] Build a virtual physical twin model to generate representative data, design the system framework, and create an intermediate mapping chain;

[0167] Verify the initialization function of the power system digital twin through online fault identification;

[0168] Achieve bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification;

[0169] The test unit 202 is used to deploy the power system digital twin and form a closed-loop test, including:

[0170] Interconnecting the power system digital twin with a virtual physical twin to simulate real events and test sampling rates, communication delays, and measurement error scenarios;

[0171] Reference to ICT data processing methods;

[0172] Modify parameters based on the actual system status to generate multiple scenarios, inject noise and adjust the sampling rate to simulate the real environment;

[0173] The deployment unit 203 is configured to deploy the system to an actual power system, and includes:

[0174] Connect the verified power system digital twin to the actual power system and reuse the system framework and intermediate mapping chain for initialization;

[0175] By initiating an adaptive process, it tracks changes in the actual power system in real time and receives actual data to optimize the power system digital twin;

[0176] Enhance the analysis response and services of the actual power system based on the output of the power system digital twin.

[0177] Furthermore, an embodiment of the present invention also provides a device for developing a power system digital twin test bench, the device including a processor and a memory:

[0178] The memory is used to store program code and transmit the program code to the processor;

[0179] The processor is used to execute the steps of the method for developing a digital twin test bench for a power system as described in the above method embodiment according to the instructions in the program code.

[0180] Furthermore, an embodiment of the present invention also provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the development method of the power system digital twin test bench described in the above method embodiment.

[0181] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0182] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0183] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0185] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0186] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for developing a digital twin test bed for a power system, characterized in that: include: S1. Interactive verification of virtual physical twins and power system digital twins: Build a virtual physical twin model to generate representative data, design the system framework, and create an intermediate mapping chain; Verifying the initialization function of the power system digital twin through online fault identification; Achieving bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification; S2. Deploy the power system digital twin and form a closed-loop test: Interconnecting the power system digital twin with the virtual physical twin to simulate real events and test sampling rate, communication delay, and measurement error scenarios; Reference to ICT data processing methods; Modify parameters based on the actual system status to generate multiple scenarios, inject noise and adjust the sampling rate to simulate the real environment; S3. Deployment to the actual power system: Connecting the verified power system digital twin to the actual power system and reusing the system framework and the intermediate mapping chain for initialization; Tracking changes in the actual power system in real time by initiating an adaptive process, and receiving actual data to optimize the digital twin of the power system; The analysis response and service of the actual power system are enhanced based on the output of the power system digital twin.

2. The method for developing a digital twin test bed for a power system according to claim 1, characterized in that: The verification of the initialization function of the power system digital twin through online fault identification includes: Establish a fault library based on the normalized residuals of different fault types; A normalized residual is calculated, abnormal behavior of the physical twin is detected based on the normalized residual, and a fault alarm is triggered when the abnormal behavior is detected.

3. The method for developing a digital twin test bed for a power system according to claim 1, characterized in that: The method of achieving bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification includes: The digital twin model is updated by an adaptive process that is time-driven, event-driven, or a hybrid of the time-driven and event-driven processes, and the update mode is adjusted based on the state error between the virtual physical twin and the power system digital twin.

4. The method for developing a digital twin test bed for a power system according to claim 3, characterized in that: The expression of the adaptive process is: ； Where, 1 and 2 is the dynamic threshold and k is the index.

5. The method for developing a digital twin test bed for a power system according to claim 1, characterized in that: The reference communication data processing method includes: The communication data processing methods used include measurement compensation, measurement amplification and measurement expansion, among which: The measurement compensation is used to replace the physical measurement with the estimated value of the power system digital twin when the physical measurement is missing; The measurement amplification is used to generate a high sampling rate data set; The measurement extension is used to predict non-measured variables through the power system digital twin.

6. The method for developing a digital twin test bed for a power system according to claim 5, characterized in that: Predicting non-measured variables by using the power system digital twin includes: The electrical variables of the medium voltage side distribution transformer are predicted through the power system digital twin.

7. The method for developing a digital twin test bed for a power system according to claim 1, characterized in that: The bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin is achieved through transient response verification, and further includes: The key functions of the power system digital twin are verified so that the power system digital twin tracks the behavior of the virtual physical twin under steady-state operation and transient operation.

8. A development system for a power system digital twin test bed, characterized in that: include: Interactive verification unit, used for interactive verification of virtual physical twins and power system digital twins, including: Build a virtual physical twin model to generate representative data, design the system framework, and create an intermediate mapping chain; Verifying the initialization function of the power system digital twin through online fault identification; Achieving bidirectional real-time data synchronization between the virtual physical twin and the power system digital twin through transient response verification; A test unit, used to deploy the power system digital twin and form a closed-loop test, including: Interconnecting the power system digital twin with the virtual physical twin to simulate real events and test sampling rate, communication delay, and measurement error scenarios; Reference to ICT data processing methods; Modify parameters based on the actual system status to generate multiple scenarios, inject noise and adjust the sampling rate to simulate the real environment; Deployment unit, used for deployment to actual power systems, including: Connecting the verified power system digital twin to the actual power system and reusing the system framework and the intermediate mapping chain for initialization; Tracking changes in the actual power system in real time by initiating an adaptive process, and receiving actual data to optimize the digital twin of the power system; The analysis response and service of the actual power system are enhanced based on the output of the power system digital twin.

9. A development device for a power system digital twin test bench, characterized in that: The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the method for developing a digital twin test bench for a power system according to any one of claims 1 to 7 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program code, and the program code is used to execute the method for developing a digital twin test bench for a power system according to any one of claims 1 to 7.

Images