A method for constructing and consistency checking of power dispatch grid digital twin model

Abstract

The application relates to the fields of power dispatching and digital twin technology, and discloses a method for constructing and consistency checking of a power dispatching power grid digital twin model. The method comprises the following steps: obtaining real-time dispatching data of a target power grid, automatically constructing a power grid digital twin model based on the real-time dispatching data, wherein the digital twin model comprises a topology model, an operation state model and a time sequence model, performing consistency comparison between a virtual scene corresponding to the digital twin model and real dispatching business, calculating topology consistency indexes, state consistency indexes and time sequence consistency indexes, and generating a consistency checking report, and when any consistency index exceeds a corresponding preset threshold, correcting and modifying the virtual scene according to the checking report. The application automatically constructs a digital twin model with a unified data source, realizes three-dimensional quantitative checking and closed-loop calibration, and solves the problems of model updating lag and lack of consistency quantitative checking means in the prior art.

Description

Technical Field

[0001] This invention relates to the fields of power dispatching and digital twin technology, specifically to a method for constructing and verifying the consistency of a power dispatching grid digital twin model. Background Technology

[0002] Power grid dispatching is a core component ensuring the safe, stable, and economical operation of the power system. The operational skills, fault diagnosis, and accident handling capabilities of dispatchers directly impact the safe operation of the power grid. With the rapid development of ultra-high-voltage AC / DC hybrid power grids and the large-scale integration of new energy sources, power grid operation modes are becoming increasingly complex and dynamic, placing higher demands on dispatchers' real-time decision-making capabilities and emergency response levels. Dispatcher training simulation systems are an important technical means to improve dispatchers' practical and emergency response capabilities. By constructing a virtual power grid environment, they enable dispatchers to conduct normal operations, accident handling, and anti-accident drills without affecting the real power grid. In recent years, digital twin technology, by constructing a high-fidelity dynamic mapping of the physical power grid in virtual space, has provided a more realistic and immersive operating environment for dispatch training, and has become one of the key technological directions for the digital transformation of the power grid.

[0003] However, existing digital twin modeling and dispatch training simulation technologies for power grids still have the following shortcomings: First, the construction methods for virtual power grid models are outdated, and updates are severely lagging. Currently, most power grid models in dispatch training simulation systems are still manually constructed based on simplified power grid wiring diagrams and typical parameters entered offline. The topology and equipment parameters used differ significantly from those of actual operating power grids. When the power grid undergoes maintenance, expansion, or operational mode adjustments, updates to the virtual model rely on manual maintenance, often lagging by hours or even days. This prevents the model from keeping up with real-time changes in the actual power grid, resulting in training scenarios that are detached from reality.

[0004] Second, there is a lack of quantitative verification methods for the consistency between virtual and real-world training scenarios. Current technologies mostly rely on qualitative descriptions to evaluate the consistency between digital twin models and the real power grid, lacking a quantitative indicator system that can be implemented in an engineering context. Training managers cannot objectively determine the extent to which the virtual scenario approximates the real power grid, nor can they quickly pinpoint the specific sources of deviation, thus failing to effectively guarantee model fidelity.

[0005] Third, the inconsistent sources of modeling data make it difficult to achieve automatic closed-loop processing. In existing methods, topology data, measurement data, and time data often come from multiple heterogeneous systems with inconsistent data formats and update frequencies. The lack of a unified and reliable data source and an automated mapping mechanism makes it difficult for digital twin models to be automatically and quickly updated in response to changes in the real power grid, and thus prevents the formation of a closed-loop process from modeling to verification to correction. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for constructing and verifying the consistency of a digital twin model of a power grid for power dispatching. This method has advantages such as automatically constructing a digital twin model from a unified data source, achieving quantitative verification and closed-loop calibration in three dimensions (topology, state, and time), and ensuring the model's fidelity by rolling with the real power grid at the second level. This solves the aforementioned technical problems.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for constructing and verifying the consistency of a digital twin model of a power dispatching grid, comprising the following steps: S1: Obtain real-time dispatch data of the target power grid, wherein the real-time dispatch data is provided by the dispatch automation system; S2: Automatically construct a digital twin model of the power grid based on the real-time scheduling data obtained in S1. The digital twin model includes a topology model with power grid equipment as nodes and electrical connections as edges, an operating status model with telemetry and teleindication as sources, and a time series model aligned with timestamps. S3: Based on the digital twin model built on S2, the virtual scenario corresponding to the actual scheduling business is compared and analyzed for consistency. The topology consistency index, state consistency index and timing consistency index are calculated, and a consistency verification report is generated. S4: When any consistency index calculated in S3 exceeds the corresponding preset threshold, the virtual scene is corrected according to the consistency verification report generated in S3.

[0008] As a preferred technical solution of the present invention, the automatic construction of the power grid digital twin model in S2 specifically includes parsing the public information model to obtain the equipment objects and their connection relationships to generate a topology model, mapping real-time measurement data to the corresponding equipment nodes according to the unique identifier of the equipment to generate an operating status model, and aligning and interpolating the status of each equipment according to the data acquisition timestamp to generate a time series model.

[0009] As a preferred embodiment of the present invention, the expression for the S3 topology consistency index is as follows: ; in, This is a topology consistency indicator; This represents the number of device edges whose connection relationships are consistent between the virtual and real topologies. This represents the number of edges in the actual power grid topology.

[0010] As a preferred embodiment of the present invention, the expression for the S3 state consistency index is as follows: ; in, It serves as a state consistency indicator; This represents the total number of device status pairs participating in the comparison. Indexed for device status pairs; For summation operations; For the first The numerical values ​​of the corresponding state variables in the digital twin virtual scene for each device state variable pair; For the first The measured values ​​of the corresponding status variables of each device status variable in actual scheduling operations; For the first Each device status variable corresponds to the rated parameters of its respective device.

[0011] As a preferred embodiment of the present invention, the expression for the S3 timing consistency index is as follows: ; in, As a time-series consistency indicator; For the first The timestamp corresponding to each sampling point in the digital twin virtual scene; For the first The timestamp corresponding to each sampling point in the actual scheduling operation; It is the absolute value; To obtain the maximum value; These are the sampling points.

[0012] Compared with existing technologies, this invention provides a method for constructing and verifying the consistency of a digital twin model of a power grid for power dispatching, which has the following beneficial effects: 1. This invention uses the public information CIM / E model and its real-time library of the scheduling automation system as the only trusted data source. The processor automatically executes model construction and adopts an incremental reconstruction mechanism to only locally update the equipment sub-graphs affected by changes. This realizes the automatic construction and second-level rolling update of the digital twin model, which changes the traditional method of relying on manual offline modeling and greatly reduces the model update overhead.

[0013] 2. This invention defines three quantifiable evaluation criteria: topology consistency index, state consistency index, and timing consistency index. These criteria quantify the consistency between the virtual scenario and the real power grid from three dimensions: device connection relationship, operational measurement value, and event sampling time. This overcomes the deficiency of existing technologies where consistency evaluation is only at the qualitative level, and provides an objective and engineering-implementable quantitative basis for model fidelity.

[0014] 3. This invention automatically performs corresponding calibration correction based on the calibration report when any consistency index exceeds the threshold after the calibration is completed. After correction, the calibration is repeated until all indicators meet the preset threshold, forming a complete closed-loop control of modeling, calibration, and verification. This ensures that the digital twin model remains consistent with the real power grid when the power grid operation mode is adjusted, the load fluctuates, or the measurement is updated, fundamentally solving the technical problem that the fidelity of traditional models cannot be continuously guaranteed. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 A method for constructing and verifying the consistency of a digital twin model of a power grid for power dispatching includes the following steps: S1: Obtain real-time dispatch data of the target power grid. The real-time dispatch data includes at least power grid topology data, equipment parameters, real-time measurement data, and current operating mode. The real-time dispatch data is provided by the dispatch automation system in the Common Information Model (CIM / E) format and its real-time library, and serves as the only trusted data source for the digital twin model. The system is interconnected with the dispatch automation system (EMS / D5000) via a dispatch data network to periodically acquire real-time dispatch data of the target power grid. The power grid topology data is expressed in the Common Information Model (CIM / E) format, covering equipment objects and connection relationships of substations, buses, transformers, transmission lines, circuit breakers, disconnect switches, and loads. Equipment parameters include the rated parameters and limits of each device. Real-time measurement data is transmitted via telemetry and telesignaling, and includes at least active power, reactive power, voltage amplitude, current, and switch position. The current power grid operation mode includes the wiring method, load level, and power generation output plan. All data carries a timestamp of acquisition and uses the dispatch automation system's time clock as a unified time reference for subsequent time alignment and consistency determination. S2: Based on the real-time scheduling data obtained from S1, an automatic digital twin model of the power grid is constructed. The digital twin model includes a topology model with power grid equipment as nodes and electrical connections as edges, an operation status model with telemetry and teleindication as sources, and a time series model aligned with timestamps. The steps for automatically constructing a digital twin model are as follows: First, the public information CIM / E model is parsed to extract equipment objects and their electrical connection relationships, generating a topology model with equipment as nodes and connections as edges. Then, real-time measurement data is mapped to the corresponding equipment nodes based on the unique identifier of the equipment, generating an operating status model reflecting the current operating status. Finally, the state sequences of each equipment are aligned and linearly interpolated according to the data acquisition timestamp to eliminate the differences in refresh cycles of different data sources, generating a time-series model. This forms a digital twin model containing the three elements of topology, state, and time-series, serving as the basis for the virtual training scenario. When the CIM / E model changes, this embodiment only performs incremental reconstruction on the equipment subgraphs affected by the change, rather than a full reconstruction of the entire network model. S3: Based on the digital twin model built on S2, the virtual scenario corresponding to the actual scheduling business is compared and analyzed for consistency. The topology consistency index, state consistency index and timing consistency index are calculated, and a consistency verification report is generated. The topology consistency index is based on the total number of device connection edges in the real power grid topology. Connection edge consistency means that the two endpoints of the edge have the same device identifier and represent the same electrical connection relationship; for example, the busbars or line devices connected to both sides of a circuit breaker are identical. The number of topology-consistent connection edges in the virtual training scenario is counted, and the topology consistency index is calculated using the following formula: ; in, This is a topology consistency indicator; This represents the number of device edges whose connection relationships are consistent between the virtual and real topologies. This represents the actual number of edges in the power grid topology. State consistency metrics refer to the virtual state variables of the same device at the same time. Compared with the actual measured state quantity Grouped by state variable type, and based on the equipment rated parameters corresponding to each type of state variable. Normalize each item separately to obtain the normalized deviation; use the root mean square of the normalized deviation of all participating equipment state variables as the state consistency index, which is calculated using the following formula: ; in, It serves as a state consistency indicator; This represents the total number of device status pairs participating in the comparison. Indexed for device status pairs; For summation operations, which involve summing all device state pairs participating in the comparison, the index... The range of values ​​is =1,2,…,m; For the first The state quantity value corresponding to each device state quantity pair in the digital twin virtual scene, that is, the virtual value of the state quantity of the device obtained by simulation calculation of the virtual scene. For the first The measured value of the status quantity corresponding to each device status quantity pair in the actual scheduling operation, that is, the actual value of the same status quantity of the same device sent by the scheduling automation system in the form of telemetry. For the first The rated parameters of each device state quantity are used as the normalization reference for that state quantity, wherein the rated parameters are selected according to the type of state quantity, including rated capacity, rated voltage and rated current; State variables include at least active power, reactive power, voltage amplitude, and current. State variables with different dimensions are first grouped and normalized according to their respective rated parameters, and then merged to calculate the root mean square, so as to avoid the deviation caused by the difference in dimensions being dominated by a single dimension. The timing consistency index uses the time synchronization clock of the scheduling automation system as a unified time reference. For the same event or the same measurement sampling point, the absolute value of the difference between the timestamp in the virtual training scenario and the corresponding timestamp in the real scheduling business is calculated. The maximum timing deviation among all sampling points is taken as the timing consistency index, which is calculated using the following formula: ; in, As a time-series consistency indicator; For the first The timestamp corresponding to each sampling point in the digital twin virtual scene; For the first The timestamp corresponding to each sampling point in actual scheduling operations is based on the time synchronization clock of the scheduling automation system; the above subscripts The range of values ​​is =1,2,…,n, where n is the total number of sampling points participating in the comparison within the verification period; It is the absolute value; To obtain the maximum value; For sampling points; The consistency verification report should record at least: the values ​​of each consistency indicator, the comparison results with the preset thresholds, the list of devices corresponding to the indicators exceeding the thresholds, and the type of deviation (topology deviation, state deviation, and time deviation); the preset thresholds can be configured according to the training accuracy requirements, and typical values ​​for specific implementation are: No less than 99%, No more than 1%, No more than 1 second; S4: When any consistency index calculated in S3 exceeds the corresponding preset threshold, the virtual scene is corrected according to the consistency verification report generated in S3. Specifically, when the topology consistency index exceeds the threshold, the topology model is reconstructed based on the latest CIM / E model; when the state consistency index exceeds the threshold, the virtual device state is resampled and corrected based on the latest real-time measurement data; when the timing consistency index exceeds the threshold, the virtual scene timestamp is aligned and the data acquisition and transmission delay is compensated; after the correction is completed, S3 is re-executed until all indicators meet the threshold for N consecutive verification cycles (N is an integer not less than 1, and N=3 is taken in this embodiment), then it is confirmed that the virtual training scene is consistent with the real scheduling business, thus forming a closed loop of modeling, verification, and calibration. In practice, steps S1 to S4 are executed cyclically according to a preset verification cycle (e.g., 5 seconds) to ensure that the virtual training scenario maintains consistency with real dispatching operations when the power grid operation mode is adjusted, load fluctuates, or measurement is updated.

[0018] The specific implementation method is as follows: The example uses a typical power grid connection in a certain region as the simulation object, with the following scale: one 500kV substation, eight 220kV substations, and 23 110kV substations; 64 main transformers, 412 transmission lines, and approximately 1100 circuit breakers; a total of approximately 38,000 telemetry and tele-signaling measurement points. The dispatch automation system is D5000, with a telemetry refresh cycle of 1 second and event-driven tele-signaling. The virtual-real consistency verification cycle is set to 5 seconds. The modeling and verification process is automatically executed on the processor deployed on the dispatcher training simulation system.

[0019] In the comparative experiment, the same training scenario was constructed using the traditional method of manual offline modeling without consistency verification and the method of automatic modeling plus closed-loop consistency verification of the present invention, respectively. The simulation was run continuously for 24 hours, and key consistency indicators were recorded. The results are shown in Table 1.

[0020] Table 1. Comparison of consistency indicators between traditional methods and the present invention. As shown in Table 1, after adopting the method of this invention, the topology consistency improved from 96.2% to 99.97%, the state consistency NRMSE decreased from 3.8% to 0.42%, and the maximum deviation of timing consistency decreased from 5.0 seconds to 0.3 seconds, all meeting the preset thresholds. Simultaneously, modeling time was reduced from approximately 4 hours manually to 86 seconds automatically (this 86 seconds includes CIM / E resolution, state mapping, and a complete time pairing link), and no training interruptions due to model distortion occurred during 24-hour continuous training simulation. These results demonstrate that this invention can effectively eliminate topology, state, and timing deviations between the virtual training scenario and the real power grid, significantly improving the fidelity and usability of the training environment. This example is a simulation verification; in actual engineering, the index values ​​vary depending on the power grid scale and calibration cycle configuration.

[0021] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for constructing and verifying the consistency of a digital twin model of a power grid for power dispatching, characterized in that: Includes the following steps: S1: Obtain real-time dispatch data of the target power grid, wherein the real-time dispatch data is provided by the dispatch automation system; S2: Automatically construct a digital twin model of the power grid based on the real-time scheduling data obtained in S1. The digital twin model includes a topology model with power grid equipment as nodes and electrical connections as edges, an operating status model with telemetry and teleindication as sources, and a time series model aligned with timestamps. S3: Based on the digital twin model built on S2, the virtual scenario corresponding to the actual scheduling business is compared and analyzed for consistency. The topology consistency index, state consistency index and timing consistency index are calculated, and a consistency verification report is generated. S4: When any consistency index calculated in S3 exceeds the corresponding preset threshold, the virtual scene is corrected according to the consistency verification report generated in S3.

2. The method for constructing and verifying the consistency of a digital twin model of a power dispatching grid according to claim 1, characterized in that: The automatic construction of the power grid digital twin model in S2 specifically includes parsing the public information model to obtain equipment objects and their connection relationships to generate a topology model, mapping real-time measurement data to corresponding equipment nodes according to the unique identifier of the equipment to generate an operating status model, and aligning and interpolating the status of each equipment according to the data acquisition timestamp to generate a time series model.

3. The method for constructing and verifying the consistency of a digital twin model of a power dispatching grid according to claim 1, characterized in that: The expression for the S3 topology consistency index is as follows: ； in, This is a topology consistency indicator; This represents the number of device edges whose connection relationships are consistent between the virtual and real topologies. This represents the number of edges in the actual power grid topology.

4. The method for constructing and verifying the consistency of a digital twin model of a power dispatching grid according to claim 1, characterized in that: The expression for the S3 state consistency index is as follows: ； in, It serves as a state consistency indicator; This represents the total number of device status pairs participating in the comparison. Indexed for device status pairs; For summation operations; For the first The numerical values ​​of the corresponding state variables in the digital twin virtual scene for each device state variable pair; For the first The measured values ​​of the corresponding status variables of each device status variable in actual scheduling operations; For the first Each device status variable corresponds to the rated parameters of its respective device.

5. The method for constructing and verifying the consistency of a digital twin model of a power dispatching grid according to claim 1, characterized in that: The expression for the S3 time series consistency index is as follows: ； in, As a time-series consistency indicator; For the first The timestamp corresponding to each sampling point in the digital twin virtual scene; For the first The timestamp corresponding to each sampling point in the actual scheduling operation; It is the absolute value; To obtain the maximum value; These are the sampling points.

Images