SVG power grid dynamic networking and simulation system combining digital twinning

Abstract

The invention relates to the technical field of power supply management systems, in particular to a digital twinning combined SVG power grid dynamic network construction and simulation system, which comprises a data acquisition module, a data processing module, a model construction module, a simulation verification module and a monitoring optimization module, the data processing module is used for carrying out data processing on power grid multi-modal data, the model building module is used for outputting a result model, the simulation verification module is used for outputting a deviation calibration strategy, and the monitoring optimization module is used for correcting the result model. According to the method, the deviation calibration strategy is output according to the multi-modal data of the power grid, deviation calibration and calculation power correction are performed on the result model so as to calibrate the result model, and the calculation power and simulation processes are balanced, so that the precision and efficiency of the result model are improved.

Description

Technical Field

[0001] This invention relates to the field of power supply management system technology, and in particular to an SVG power grid dynamic network construction and simulation system that combines digital twins. Background Technology

[0002] With the development of digital twin technology, the power grid operation mode is being upgraded from "post-event remediation" to "pre-event simulation, in-event control, and post-event iteration". In this process, there is a deviation in the time scale between the device response and system stability of the SVG model, and the operation cost of predicting new faults in the power grid field is high. Therefore, the technology of simulating the power grid topology is particularly important. However, the high accuracy of the simulation results requires the acquisition of power grid data by a large number of sensors, which leads to excessively high simulation costs. In addition, there is a lack of balance between the simulation accuracy requirements and the actual computing power.

[0003] Chinese patent CN117955122A discloses a method for modeling SVG (Static Var Generator) for new energy power plants. This invention discloses a method for modeling SVG for new energy power plants, assuming that the three-phase circuit parameters of the SVG are symmetrical and can be considered as a symmetrical three-phase circuit; the switching devices in the SVG circuit are considered as ideal switching devices; the system voltage source is a standard three-phase symmetrical power supply; for voltage-type SVG circuit topologies, the method specifically includes the following steps: Step S1, constructing a mathematical model of the SVG in the abc coordinate system; Step S2, constructing a mathematical model of the SVG in the dq rotating coordinate system, including the transformation from the abc coordinate system to the α-β coordinate system and the transformation from the α-β coordinate system to the dq coordinate system. It is evident that this scheme still suffers from insufficient SVG model accuracy, a time scale difference between the device-level model and the system-level model, high sensor costs, and an inability to balance computing power with simulation accuracy. Summary of the Invention

[0004] To address these issues, this invention provides an SVG power grid dynamic network construction and simulation system that combines digital twins, in order to overcome the problems of insufficient accuracy of SVG models, time scale differences between device-level and system-level models, high sensor costs, and the inability to balance computing power and simulation accuracy in existing technologies.

[0005] To achieve the above objectives, the present invention provides an SVG power grid dynamic network construction and simulation system combining digital twins, the system comprising: The data acquisition module is used to collect multimodal data of the power grid; The data processing module is used to process the multimodal data of the power grid to obtain processed multimodal data. The model building module is used to time-align the SVG model device-level model and system-level model based on the processed multimodal data, obtain the result model, and output the result model; The simulation verification module is used to obtain the deviation of the model results based on the result model, output the deviation calibration strategy based on the deviation of the model results, and adjust the calibration strategy based on the number of calibrations. The monitoring and optimization module is used to adjust computing power based on the available computing power during the process of obtaining model result deviations, optimize computing power based on the usage time, and correct computing power for the resulting model based on the attenuation value.

[0006] Furthermore, when the data processing module processes the multimodal data of the power grid according to the data processing method, the data processing method includes: Step A01: Denoise the power grid multimodal data to obtain denoised power grid multimodal data; Step A02: Input the denoised power grid multimodal data into the power grid data feature extraction model to obtain the power grid multimodal data feature vector output by the power grid data feature extraction model; Step A03: Perform local compression of the feature vectors of the power grid multimodal data at the edge nodes to obtain the processed multimodal data.

[0007] Furthermore, when the model building module performs time-aligned construction of the SVG model based on the processed multimodal data, the SVG model includes a device-level model and a system-level model. The device-level model is constructed using a device-level model construction method, which includes: Step B01: Input the circuit physical topology into the device dynamic equation to obtain the target device dynamic equation; Step B02: Design the control module for the dynamic equations of the target equipment to obtain the equipment-level model; The model building module also constructs a system-level model using a system-level model building method, which includes: Step C01: Perform interface unification processing on the device-level model to obtain an interface-unified device-level model; Step C02: Connect the unified device-level model of the interface according to the basic system architecture to obtain the connected device-level model; Step C03: Embed the system dynamic equations into the connected device-level model to obtain the system-level model; The model building module performs time alignment processing on the device-level model and the system-level model to obtain aligned device-level model and aligned system-level model, and outputs the aligned device-level model and aligned system-level model as the result model.

[0008] Furthermore, when the simulation verification module acquires the model result deviation based on the result model, it inputs the processed multimodal data into the result model to obtain the power grid simulation model output by the result model. Based on the sampling quantity N, it selects actual measurement points for the power grid simulation model to obtain a set of simulated actual measurement points. It performs vector transformation on the set of simulated actual measurement points and the set of actual measurement point locations to obtain a set of simulated actual measurement point vector values ​​FZ and a set of actual measurement point location vector values ​​FX. Based on the set of simulated actual measurement point vector values ​​FZ, the set of actual measurement point location vector values ​​FX, the simulation weight w1, and the actual measurement weight w2, it calculates the model result deviation MX, setting MX = w1 × FZ + w2 × FX, to obtain the model result deviation MX.

[0009] Furthermore, when the simulation verification module outputs the deviation calibration strategy based on the model result deviation, it compares the model result deviation MX with the preset model result deviation MX0, judges the model result deviation based on the comparison result, and outputs the deviation calibration strategy based on the judgment result, wherein: When MX≤MX0, the simulation verification module determines that there is no deviation in the model result and does not output the deviation calibration strategy; When MX > MX0, the simulation verification module determines that there is a deviation in the model result and outputs a deviation calibration strategy. The deviation calibration strategy is to replace the sensor cluster position with the actual measurement point position and re-acquire the multi-mode data of the power grid based on the sensor cluster position.

[0010] Furthermore, when the simulation verification module adjusts the calibration strategy based on the number of calibrations, it compares the number of calibrations CL with the preset number of calibrations CL0, determines the status of the number of calibrations based on the comparison result, and adjusts the calibration strategy based on the determination result, wherein: When CL≤CL0, the simulation verification module determines that the calibration count is normal and does not adjust the calibration strategy. When CL > CL0, the simulation verification module determines that the calibration count is abnormal and adjusts the calibration strategy: the sensor cluster location is replaced with the highest frequency calibration point, and the power grid multimodal data is reacquired based on the sensor cluster location.

[0011] Furthermore, the monitoring optimization module includes: The computing power monitoring unit is used to adjust the computing power based on the process of obtaining deviations in model results from available computing power. The attenuation monitoring unit is used to optimize computing power during the process of adjusting computing power based on usage time, and to correct the computing power of the resulting model based on the attenuation level.

[0012] Furthermore, when the computing power monitoring unit adjusts the computing power based on the acquisition process of the model result deviation according to the available computing power, it compares the available computing power SL with the preset available computing power SL0, judges the status of available computing power based on the comparison result, and adjusts the sampling quantity N according to the judgment result, wherein: When SL≥SL0, the monitoring and optimization module determines that the available computing power is sufficient and does not adjust the computing power for the sampling quantity N. When SL < SL0, the monitoring and optimization module determines that the available computing power is insufficient, and adjusts the sampling quantity N according to the computing power adjustment coefficient hg, setting hg = 0.66 + 0.22 × e -(SL0-SL) Where e is the base of the natural logarithm, the adjusted sampling number N` is obtained, and N` is set to N×hg. The adjusted sampling number N` is rounded to a positive integer. The sampling number N is replaced with the adjusted sampling number N`, and the actual measurement points of the power grid simulation model are reselected according to the sampling number N.

[0013] Furthermore, when the attenuation monitoring unit optimizes computing power during the process of adjusting computing power based on usage duration, it compares the usage duration TD with the first preset usage duration TD1 and the second preset usage duration TD2, determines the status of the usage duration based on the comparison result, and optimizes the preset available computing power SL0 based on the determination result, wherein: When TD≤TD1, the attenuation monitoring unit determines the usage time as short and does not perform computing power optimization on the preset available computing power SL0; When TD1 < TD ≤ TD2, the attenuation monitoring unit determines the usage time as medium and optimizes the preset available computing power SL0 according to the optimization coefficient yh, setting yh = 1.46 - 0.28 × e -（TD-TD1） Where e is the base of the natural logarithm, the optimized preset available computing power SL0` is obtained, SL0` is set to SL0×yh, the preset available computing power SL0 is replaced with the optimized preset available computing power SL0`, and the available computing power SL is re-compared with the preset available computing power SL0; When TD > TD2, the attenuation monitoring unit determines that the usage time is long, optimizes the preset available computing power SL0 according to the optimization coefficient yh, and corrects the computing power of the result model according to the attenuation value.

[0014] Furthermore, when the attenuation monitoring unit performs computational power correction on the result model based on the attenuation degree value, it inputs the grid voltage, grid current, and equipment temperature from the grid multimodal data into the attenuation prediction model to obtain the attenuation degree value SU output by the attenuation prediction model. The attenuation degree value SU is then compared with a preset attenuation degree value SU0. Based on the comparison result, the attenuation impact status is judged, and the computational power is corrected on the result model based on the judgment result. Wherein: When SU≤SU0, the attenuation monitoring unit determines that the attenuation effect is no effect and does not perform computational power correction on the result model; When SU>SU0, the attenuation monitoring unit determines that the attenuation effect is present and performs computational correction on the result model: attenuation correction is performed on the result model to obtain the corrected result model, the result model is replaced with the corrected result model, and the result model is output.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: The system collects multimodal power grid data through a data acquisition module, facilitating subsequent output and correction of the resulting model based on the multimodal power grid data. The system also performs denoising, feature extraction, and compression on the multimodal power grid data through a data processing module, removing interference signals and reducing data volume to increase the effectiveness of the multimodal power grid data. Furthermore, the system outputs the resulting model through a model building module, performing time alignment on the SVG model to obtain device-level and system-level models with unified time steps and timestamps, thus unifying the time scale of the SVG model and improving the efficiency of power grid simulation. The system also outputs a deviation calibration strategy through a simulation verification module, updating the sensor cluster position in real time to save probe costs, thereby improving simulation efficiency and resource utilization. Finally, the system corrects the resulting model based on available computing power and usage time through a monitoring and optimization module, saving computing power and improving simulation accuracy, thereby improving the balance between computing power and simulation. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the SVG power grid dynamic network construction and simulation system that combines digital twins in this embodiment; Figure 2 This is a schematic diagram of the monitoring and optimization module in this embodiment. Detailed Implementation

[0017] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0018] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0019] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0020] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0021] Please see Figure 1 As shown, this is a schematic diagram of the SVG power grid dynamic network construction and simulation system combining digital twins in this embodiment. The system includes: The data acquisition module is used to collect multimodal data of the power grid; A data processing module is used to process the multimodal data of the power grid to obtain processed multimodal data. The data processing module is connected to the data acquisition module. The model building module is used to construct the SVG model device-level model and system-level model in time alignment based on the processed multimodal data, obtain the result model, and output the result model. The model building module is connected to the data processing module. The simulation verification module is used to obtain the deviation of the model result based on the result model, output the deviation calibration strategy based on the deviation of the model result, and adjust the calibration strategy based on the number of calibrations. The simulation verification module is connected to the model construction module. The monitoring and optimization module is used to adjust the computing power based on the available computing power during the process of obtaining the deviation of the model results, optimize the computing power adjustment process based on the usage time, and correct the computing power of the result model based on the attenuation value. The monitoring and optimization module is connected to the simulation verification module.

[0022] Specifically, the SVG power grid dynamic network construction and simulation system combined with digital twins is applied in the power grid simulation equipment terminal. The system outputs deviation calibration strategies based on multimodal power grid data, and performs deviation calibration and computing power correction on the resulting model to calibrate the model and balance computing power and simulation processes, thereby improving the accuracy and efficiency of the resulting model. The system acquires multimodal power grid data through a data acquisition module to facilitate subsequent output and correction of the resulting model based on this data. The system also uses a data processing module to denoise, extract features, and compress the multimodal power grid data, removing interference signals and reducing the number of... To increase the effectiveness of multimodal power grid data, the system also outputs the resulting model through a model building module, performs time-aligned construction on the SVG model, and obtains device-level and system-level models with unified time steps and timestamps. This unifies the time scale of the SVG model, thereby improving the efficiency of power grid simulation. The system also outputs deviation calibration strategies through a simulation verification module and updates the sensor cluster positions in real time to save probe costs, thereby improving simulation efficiency and resource utilization. Furthermore, the system uses a monitoring and optimization module to correct the resulting model based on available computing power and usage time, in order to save computing power and improve simulation accuracy, thereby improving the balance between computing power and simulation.

[0023] Specifically, the data acquisition module collects multimodal data of the power grid, including power grid voltage, power grid current, equipment temperature, circuit physical topology, sensor cluster location, available computing power, and usage time.

[0024] Specifically, the grid voltage refers to the potential difference between the transmission, distribution, and consumption stages of the power grid; the grid current refers to the current formed by the directional movement of charges in the power grid; and the equipment temperature refers to the temperature of a single piece of equipment in the power grid. The data acquisition module acquires the grid voltage through a voltage sensor in the sensor cluster, the grid current through a current sensor in the sensor cluster, and the equipment temperature through a temperature sensor in the sensor cluster. The sensor cluster includes voltage sensors, current sensors, and temperature sensors. The voltage sensor is a device that acquires grid voltage; the current sensor is a device that acquires grid current; and the temperature sensor is a device that acquires equipment temperature. The sensor cluster location refers to the specific location of the sensor cluster within the power grid. This embodiment does not specify the location. The representation method is limited, such as constructing a three-dimensional coordinate system with the power grid simulation equipment as the origin, and representing the position in coordinate form. The circuit physical topology refers to the specific structure of the power electronic circuits, control loops and circuit connection relationships inside the power grid. This embodiment does not limit the specific method of obtaining the circuit physical topology. Those skilled in the art can freely choose according to actual needs, such as obtaining the circuit physical topology through the PCB design file of the power grid. The available computing power refers to the remaining available computing power in the power grid simulation equipment. The usage time refers to the length of time the power grid simulation equipment runs continuously. This embodiment does not limit the specific method of obtaining the available computing power and usage time. Those skilled in the art can freely choose according to actual needs, such as obtaining the available computing power and usage time through the system monitoring tool of the power grid simulation equipment.

[0025] Specifically, the data acquisition module collects multimodal data of the power grid so that the resulting model can be output and corrected based on the multimodal data of the power grid, thereby improving the simulation accuracy and efficiency of the resulting model.

[0026] Specifically, when the data processing module processes the multimodal data of the power grid according to the data processing method, the data processing method includes: Step A01: Denoise the power grid multimodal data to obtain denoised power grid multimodal data; Step A02: Input the denoised power grid multimodal data into the power grid data feature extraction model to obtain the power grid multimodal data feature vector output by the power grid data feature extraction model; Step A03: Perform local compression of the feature vectors of the power grid multimodal data at the edge nodes to obtain the processed multimodal data.

[0027] Specifically, the denoising process refers to the process of eliminating noise in the multimodal power grid data. This embodiment does not limit the specific method of denoising; those skilled in the art can freely choose according to actual needs, such as using Kalman filtering for denoising. The power grid data feature extraction model refers to a recurrent neural network model that takes the denoised multimodal power grid data as input data and the feature vectors of the multimodal power grid data as output data. This embodiment does not limit the specific construction method of the power grid data feature extraction model; those skilled in the art can freely choose according to actual needs, such as using historical denoised multimodal power grid data and its... The corresponding power grid multimodal data feature vectors are used as training datasets to train the recurrent neural network model, resulting in a power grid data feature extraction model. The power grid multimodal data feature vectors refer to the vector values ​​representing the features of the power grid multimodal data after denoising, obtained from the power grid data feature extraction model. The edge node local compression refers to the operation of retaining key features of the power grid multimodal data feature vectors using edge computing devices. The key features refer to the vector values ​​that can reflect the main features of the power grid multimodal data feature vectors. The edge computing devices refer to devices deployed in the power system close to the data source to collect power grid multimodal data, such as substations, distribution cabinets, and smart meters.

[0028] Specifically, the data processing module removes interference signals from the multimodal power grid data by denoising, extracting features, and compressing the multimodal power grid data, thereby reducing the amount of data and increasing the effectiveness of the multimodal power grid data, thus improving the efficiency of SVG power grid dynamic network simulation.

[0029] Specifically, when the model building module performs time-aligned construction of the SVG model based on the processed multimodal data, the SVG model includes a device-level model and a system-level model. The device-level model is constructed using a device-level model construction method, which includes: Step B01: Input the circuit physical topology into the device dynamic equation to obtain the target device dynamic equation; Step B02: Design the control module for the dynamic equations of the target equipment to obtain the equipment-level model; The model building module also constructs a system-level model using a system-level model building method, which includes: Step C01: Perform interface unification processing on the device-level model to obtain an interface-unified device-level model; Step C02: Connect the unified device-level model of the interface according to the basic system architecture to obtain the connected device-level model; Step C03: Embed the system dynamic equations into the connected device-level model to obtain the system-level model; The model building module performs time alignment processing on the device-level model and the system-level model to obtain aligned device-level model and aligned system-level model, and outputs the aligned device-level model and aligned system-level model as the result model.

[0030] Specifically, the device dynamic equations refer to differential equations describing grid interactions. These grid interactions refer to the coordinated operation of different components in a power system, such as power generation equipment and energy storage devices, through the flow of electrical energy and control commands. This embodiment does not limit the specific construction method of the device dynamic equations; those skilled in the art can freely choose according to actual needs, such as outputting the device dynamic equations through power system simulation software. The control module design refers to the process of designing the way the device dynamic equations control grid interactions. This embodiment does not limit the specific method of control module design; those skilled in the art can freely choose according to actual needs, such as implementing the control module design through expert design. Expert design refers to the method where an expert with expertise in control module design implements the control module design. The unified interface processing refers to the standardization of the interface of the device-level model. The process of unifying the system architecture and system dynamic equations involves, for example, unifying the current unit to A and setting the direction of charge flow in the current to positive. The basic system architecture refers to the structural pattern of connecting the unified device-level models of the interface. The system dynamic equations refer to the mathematical framework for integrating and unifying the dispersed device dynamic equations. This embodiment does not limit the specific methods of obtaining the basic system architecture and system dynamic equations. Those skilled in the art can freely choose according to actual needs, such as obtaining the basic system architecture and system dynamic equations through expert construction. Expert construction refers to the method of constructing the basic system architecture and system dynamic equations by experts with the ability to construct the basic system architecture and system dynamic equations. The time alignment process refers to the process of unifying the time step and timestamp of the device-level model and the system-level model. This embodiment does not limit the specific methods of time alignment process, such as performing time alignment process through the interface variable synchronization mechanism.

[0031] Specifically, the model building module constructs the SVG model by performing time alignment, resulting in device-level and system-level models with unified time steps and timestamps, in order to unify the time scale of the SVG model and thus improve the efficiency of power grid simulation.

[0032] Specifically, when the simulation verification module obtains the model result deviation based on the result model, it inputs the processed multimodal data into the result model to obtain the power grid simulation model output by the result model. Based on the sampling quantity N, it selects actual measurement points for the power grid simulation model to obtain a set of simulated actual measurement points. It performs vector transformation on the set of simulated actual measurement points and the set of actual measurement point locations to obtain a set of simulated actual measurement point vector values ​​FZ and a set of actual measurement point location vector values ​​FX. Based on the set of simulated actual measurement point vector values ​​FZ, the set of actual measurement point location vector values ​​FX, the simulation weight w1, and the actual measurement weight w2, it calculates the model result deviation MX, setting MX = w1 × FZ + w2 × FX, to obtain the model result deviation MX.

[0033] Specifically, the power grid simulation model refers to a power grid model that simulates the real power grid environment based on the result deviation evaluation model. The sampling quantity refers to the total number of simulated measurement points selected. This embodiment does not limit the sampling quantity; for example, the sampling quantity N=300. The measurement point selection refers to the process of selecting simulated measurement points of the power grid simulation model according to the sampling quantity to calculate the model result deviation. This embodiment does not limit the specific method of measurement point selection; for example, random selection is allowed. The simulated measurement point set refers to the set of preset location points used to calculate the model result deviation. The measurement point location set refers to the set of location points in the real power grid environment corresponding to the simulated measurement point set. The vector transformation refers to the ability to describe the simulated measurement point set and the measurement point location set. The vector values ​​of the features are not limited in this embodiment to the specific method of vector transformation. Those skilled in the art can freely choose according to actual needs, such as using Paython for vector transformation. The set of simulated measured point vector values ​​refers to the set of vector values ​​describing the features of the simulated measured points after vector transformation. The set of measured point position vector values ​​refers to the set of vector values ​​describing the features of the measured point position after vector transformation. The simulation weight refers to the coefficient that measures the importance of the set of simulated measured point vector values ​​in the model result deviation. The measured weight refers to the coefficient that measures the importance of the set of measured point position vector values ​​in the model result deviation. The model result deviation refers to the numerical value that reflects the degree of difference between the set of simulated measured points and the set of measured point positions.

[0034] Specifically, the simulation verification module acquires the deviation of the model results to measure the degree of difference between the power grid simulation model and the real power grid environment, which facilitates the subsequent calibration of the result model based on the deviation of the model results.

[0035] Specifically, when the simulation verification module outputs the deviation calibration strategy based on the model result deviation, it compares the model result deviation MX with the preset model result deviation MX0, judges the model result deviation based on the comparison result, and outputs the deviation calibration strategy based on the judgment result, wherein: When MX≤MX0, the simulation verification module determines that there is no deviation in the model result and does not output the deviation calibration strategy; When MX > MX0, the simulation verification module determines that there is a deviation in the model result and outputs a deviation calibration strategy. The deviation calibration strategy is to replace the sensor cluster position with the actual measurement point position and re-acquire the multi-mode data of the power grid based on the sensor cluster position.

[0036] Specifically, the preset model result deviation refers to a preset value for judging the deviation of the model result. This embodiment does not limit the specific value of the preset model result deviation. Those skilled in the art can freely choose according to actual needs. For example, in this embodiment, the preset model result deviation MX0 is set to 0.2. The condition of the model result deviation refers to the presence or absence of model result deviation judged based on the difference between the model result deviation and the preset model result deviation. The condition of the model result deviation includes no deviation and deviation.

[0037] Specifically, the simulation verification module judges the deviation of the model results. When the model results show a deviation, it performs deviation calibration on the result model to calibrate the result model in real time, thereby improving the accuracy of the result model and improving the efficiency of SVG power grid dynamic network simulation.

[0038] Specifically, when the simulation verification module adjusts the calibration strategy based on the number of calibrations, it compares the number of calibrations CL with the preset number of calibrations CL0, determines the status of the number of calibrations based on the comparison result, and adjusts the calibration strategy based on the determination result, wherein: When CL≤CL0, the simulation verification module determines that the calibration count is normal and does not adjust the calibration strategy. When CL > CL0, the simulation verification module determines that the calibration count is abnormal and adjusts the calibration strategy: the sensor cluster location is replaced with the highest frequency calibration point, and the power grid multimodal data is reacquired based on the sensor cluster location.

[0039] Specifically, the calibration count refers to the number of times the deviation calibration strategy is performed within a preset time. This embodiment does not limit the specific method for obtaining the number of deviation calibration strategies; those skilled in the art can freely choose according to actual needs, such as obtaining the calibration count through the system log of the power grid monitoring device. This embodiment does not limit the preset time; for example, the preset time is set to 10 seconds. The preset calibration count refers to a preset value for judging the state of the calibration count. This embodiment does not limit the specific value of the preset calibration count; those skilled in the art can freely choose according to actual needs. For example, this embodiment sets the preset calibration count CL0 = 10 times based on the average calibration count. The state of the calibration count refers to the degree of abnormality of the calibration count judged based on the calibration count and the preset calibration count. The state of the calibration count includes normal and abnormal. The highest frequency calibration point refers to the location of the actual measurement point corresponding to the simulation measurement point with the most calibration count in the simulation measurement point set. This embodiment does not limit the specific method for obtaining the highest frequency calibration point; for example, the calibration count of all simulation measurement points in the simulation measurement point set can be input into Paython to obtain the highest frequency calibration point.

[0040] Specifically, the simulation verification module judges the status of the calibration count. When the status of the calibration count is abnormal, the calibration strategy is adjusted to update the sensor cluster position in real time, thereby saving probe costs and improving simulation efficiency and resource utilization.

[0041] Please see Figure 2 As shown, this is a schematic diagram of the monitoring and optimization module in this embodiment. The monitoring and optimization module includes: The computing power monitoring unit is used to adjust the computing power based on the process of obtaining deviations in model results from available computing power. The attenuation monitoring unit is used to optimize computing power during the process of adjusting computing power based on usage time, and to correct the computing power of the resulting model based on the attenuation level.

[0042] Specifically, when the computing power monitoring unit adjusts the computing power based on the acquisition process of the model result deviation according to the available computing power, it compares the available computing power SL with the preset available computing power SL0, judges the status of available computing power based on the comparison result, and adjusts the sampling quantity N according to the judgment result, wherein: When SL≥SL0, the monitoring and optimization module determines that the available computing power is sufficient and does not adjust the computing power for the sampling quantity N. When SL < SL0, the monitoring and optimization module determines that the available computing power is insufficient, and adjusts the sampling quantity N according to the computing power adjustment coefficient hg, setting hg = 0.66 + 0.22 × e -(SL0-SL)Where e is the base of the natural logarithm, the adjusted sampling number N` is obtained, and N` is set to N×hg. The adjusted sampling number N` is rounded to a positive integer. The sampling number N is replaced with the adjusted sampling number N`, and the actual measurement points of the power grid simulation model are reselected according to the sampling number N.

[0043] Specifically, the preset available computing power refers to a preset value for judging the availability of computing power. This embodiment does not limit the specific value of the preset available computing power. Those skilled in the art can freely choose according to actual needs. For example, in this embodiment, the preset available computing power SL0 is set to 0.6. The availability of computing power refers to the sufficiency of available computing power judged based on the available computing power and the preset available computing power. The availability of computing power includes sufficient computing power and insufficient computing power.

[0044] Specifically, the computing power monitoring unit determines the availability of computing power. When the available computing power is insufficient, it reduces the number of samples according to the computing power adjustment coefficient. The constant term of the computing power adjustment coefficient, 0.66, is the minimum value that the computing power adjustment coefficient can reach, and the constant coefficient, 1.22, is the change amplitude of the computing power adjustment coefficient. This ensures that the change trend of the computing power adjustment coefficient is from 0.88 to 0.66 and remains constant, so as to steadily reduce the number of samples. This allows for timely reduction of the scale of selected measurement points when computing power is insufficient, thereby saving computing power and improving the balance between computing power and modeling simulation.

[0045] Specifically, when the attenuation monitoring unit optimizes computing power by adjusting computing power based on usage duration, it compares the usage duration TD with a first preset usage duration TD1 and a second preset usage duration TD2. Based on the comparison result, it determines the status of the usage duration and optimizes the preset available computing power SL0 based on the determination result, wherein: When TD≤TD1, the attenuation monitoring unit determines the usage time as short and does not perform computing power optimization on the preset available computing power SL0; When TD1 < TD ≤ TD2, the attenuation monitoring unit determines the usage time as medium and optimizes the preset available computing power SL0 according to the optimization coefficient yh, setting yh = 1.46 - 0.28 × e -（TD-TD1） Where e is the base of the natural logarithm, the optimized preset available computing power SL0` is obtained, SL0` is set to SL0×yh, the preset available computing power SL0 is replaced with the optimized preset available computing power SL0`, and the available computing power SL is re-compared with the preset available computing power SL0; When TD > TD2, the attenuation monitoring unit determines that the usage time is long, optimizes the preset available computing power SL0 according to the optimization coefficient yh, and corrects the computing power of the result model according to the attenuation value.

[0046] Specifically, the first preset usage duration refers to a preset lower limit value for judging the state of usage duration, and the second preset usage duration refers to a preset upper limit value for judging the state of usage duration. This embodiment does not limit the specific numerical settings of the first preset usage duration and the second preset usage duration. Those skilled in the art can set them freely according to actual needs. For example, in this embodiment, the first preset usage duration TD1 is set to 30 days and the second preset usage duration TD2 is set to 35 days. The state of usage duration refers to the length of time of usage duration judged based on the usage duration and the first preset usage duration and the second preset usage duration. The state of usage duration includes short time, medium time and long time.

[0047] Specifically, the attenuation monitoring unit judges the usage time status. When the usage time status is medium or long, it steadily increases the preset available computing power according to the optimization coefficient which increases from 1.18 to 1.46. The constant term of the optimization coefficient, 1.46, is the maximum value that the optimization coefficient can reach, and the constant coefficient, 0.28, represents the change amplitude of the optimization coefficient. This is to reasonably reduce the impact of usage time on the evaluation of available computing power, thereby improving the balance between computing power and simulation.

[0048] Specifically, when the attenuation monitoring unit performs computational power correction on the result model based on the attenuation degree value, it inputs the grid voltage, grid current, and equipment temperature from the grid multimodal data into the attenuation prediction model to obtain the attenuation degree value SU output by the attenuation prediction model. The attenuation degree value SU is then compared with a preset attenuation degree value SU0. Based on the comparison result, the attenuation impact status is judged, and the computational power is corrected on the result model based on the judgment result. Wherein: When SU≤SU0, the attenuation monitoring unit determines that the attenuation effect is no effect and does not perform computational power correction on the result model; When SU>SU0, the attenuation monitoring unit determines that the attenuation effect is present and performs computational correction on the result model: attenuation correction is performed on the result model to obtain the corrected result model, the result model is replaced with the corrected result model, and the result model is output.

[0049] Specifically, the attenuation prediction model refers to a recurrent neural network model that takes grid voltage, grid current, and equipment temperature as input data and attenuation degree values ​​as output data. This embodiment does not limit the specific construction method of the attenuation prediction model; those skilled in the art can freely choose according to actual needs. For example, historical grid voltage, grid current, and equipment temperature can be used as historical grid data, and the historical grid data and their corresponding attenuation degree values ​​can be used as training datasets to train the recurrent neural network model to obtain the attenuation prediction model. The attenuation degree value refers to the numerical value obtained from the attenuation prediction model that reflects the attenuation degree of the grid simulation equipment. The grid simulation equipment refers to the physical equipment of the output grid simulation model based on the result model. The attenuation degree value refers to a preset value used to judge the attenuation effect state. This embodiment does not limit the specific value of the preset attenuation degree value. Those skilled in the art can freely choose according to actual needs. For example, in this embodiment, the preset attenuation degree value SU0=0.4 is set. The attenuation effect state refers to the state of whether the attenuation degree value, judged according to the attenuation degree value and the preset attenuation degree value, has an impact on the result model. The attenuation effect state includes having an impact and not having an impact. The attenuation correction refers to the process of correcting the accuracy of the result model after it is affected by the attenuation of the power grid simulation equipment. This embodiment does not limit the specific method of attenuation correction. Those skilled in the art can freely choose according to actual needs, such as using a PID control algorithm for attenuation correction.

[0050] Specifically, the attenuation monitoring unit judges the attenuation impact status. When the attenuation impact status is "affected", it promptly corrects the attenuation of the result model to reduce the impact of the attenuation degree of the power grid simulation equipment on the accuracy of the result model in real time, thereby improving the simulation accuracy of the result model.

[0051] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A dynamic grid construction and simulation system for SVG power grids combining digital twins, characterized in that, The system includes: The data acquisition module is used to collect multimodal data of the power grid; The data processing module is used to process the multimodal data of the power grid to obtain processed multimodal data. The model building module is used to time-align the SVG model device-level model and system-level model based on the processed multimodal data, obtain the result model, and output the result model; The simulation verification module is used to obtain the deviation of the model results based on the result model, output the deviation calibration strategy based on the deviation of the model results, and adjust the calibration strategy based on the number of calibrations. The monitoring and optimization module is used to adjust computing power based on the available computing power during the process of obtaining model result deviations, optimize computing power based on the usage time, and correct computing power for the resulting model based on the attenuation value.

2. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 1, characterized in that, When the data processing module processes the multimodal data of the power grid according to the data processing method, the data processing method includes: Step A01: Denoise the power grid multimodal data to obtain denoised power grid multimodal data; Step A02: Input the denoised power grid multimodal data into the power grid data feature extraction model to obtain the power grid multimodal data feature vector output by the power grid data feature extraction model; Step A03: Perform local compression of the feature vectors of the power grid multimodal data at the edge nodes to obtain the processed multimodal data.

3. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 2, characterized in that, When the model building module performs time-aligned construction of the SVG model based on the processed multimodal data, the SVG model includes a device-level model and a system-level model. The device-level model is constructed using a device-level model construction method, which includes: Step B01: Input the circuit physical topology into the device dynamic equation to obtain the target device dynamic equation; Step B02: Design the control module for the dynamic equations of the target equipment to obtain the equipment-level model; The model building module also constructs a system-level model using a system-level model building method, which includes: Step C01: Perform interface unification processing on the device-level model to obtain an interface-unified device-level model; Step C02: Connect the unified device-level model of the interface according to the basic system architecture to obtain the connected device-level model; Step C03: Embed the system dynamic equations into the connected device-level model to obtain the system-level model; The model building module performs time alignment processing on the device-level model and the system-level model to obtain aligned device-level model and aligned system-level model, and outputs the aligned device-level model and aligned system-level model as the result model.

4. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 3, characterized in that, When the simulation verification module obtains the model result deviation based on the result model, it inputs the processed multimodal data into the result model to obtain the power grid simulation model output by the result model. Based on the sampling quantity N, it selects actual measurement points for the power grid simulation model to obtain a set of simulated actual measurement points. It performs vector transformation on the set of simulated actual measurement points and the set of actual measurement point locations to obtain the set of simulated actual measurement point vector values ​​FZ and the set of actual measurement point location vector values ​​FX. Based on the set of simulated actual measurement point vector values ​​FZ, the set of actual measurement point location vector values ​​FX, the simulation weight w1, and the actual measurement weight w2, it calculates the model result deviation MX and sets MX = w1 × FZ + w2 × FX to obtain the model result deviation MX.

5. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 4, characterized in that, When the simulation verification module outputs the deviation calibration strategy based on the model result deviation, it compares the model result deviation MX with the preset model result deviation MX0, judges the model result deviation based on the comparison result, and outputs the deviation calibration strategy based on the judgment result, wherein: When MX≤MX0, the simulation verification module determines that there is no deviation in the model result and does not output the deviation calibration strategy; When MX > MX0, the simulation verification module determines that there is a deviation in the model result and outputs a deviation calibration strategy. The deviation calibration strategy is to replace the sensor cluster position with the actual measurement point position and re-acquire the multi-mode data of the power grid based on the sensor cluster position.

6. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 5, characterized in that, When the simulation verification module adjusts the calibration strategy based on the number of calibrations, it compares the number of calibrations CL with the preset number of calibrations CL0, determines the status of the number of calibrations based on the comparison result, and adjusts the calibration strategy based on the determination result, wherein: When CL≤CL0, the simulation verification module determines that the calibration count is normal and does not adjust the calibration strategy. When CL > CL0, the simulation verification module determines that the calibration count is abnormal and adjusts the calibration strategy: the sensor cluster location is replaced with the highest frequency calibration point, and the power grid multimodal data is reacquired based on the sensor cluster location.

7. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 6, characterized in that, The monitoring optimization module includes: The computing power monitoring unit is used to adjust the computing power based on the process of obtaining deviations in model results from available computing power. The attenuation monitoring unit is used to optimize computing power during the process of adjusting computing power based on usage time, and to correct the computing power of the resulting model based on the attenuation level.

8. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 7, characterized in that, When the computing power monitoring unit adjusts the computing power based on the process of obtaining the deviation of the model results from the available computing power, it compares the available computing power SL with the preset available computing power SL0, judges the status of the available computing power based on the comparison result, and adjusts the sampling quantity N based on the judgment result, wherein: When SL≥SL0, the monitoring and optimization module determines that the available computing power is sufficient and does not adjust the computing power for the sampling quantity N. When SL < SL0, the monitoring and optimization module determines that the available computing power is insufficient, and adjusts the sampling quantity N according to the computing power adjustment coefficient hg, setting hg = 0.66 + 0.22 × e -(SL0-SL) Where e is the base of the natural logarithm, the adjusted sampling number N` is obtained, and N` is set to N×hg. The adjusted sampling number N' is rounded to a positive integer. The sampling number N is replaced with the adjusted sampling number N`, and the actual measurement points of the power grid simulation model are reselected according to the sampling number N.

9. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 8, characterized in that, When the attenuation monitoring unit optimizes computing power by adjusting computing power based on usage duration, it compares the usage duration TD with the first preset usage duration TD1 and the second preset usage duration TD2. Based on the comparison result, it determines the status of the usage duration and optimizes the preset available computing power SL0 based on the determination result, wherein: When TD≤TD1, the attenuation monitoring unit determines the usage time as short and does not perform computing power optimization on the preset available computing power SL0; When TD1 < TD ≤ TD2, the attenuation monitoring unit determines the usage time as medium and optimizes the preset available computing power SL0 according to the optimization coefficient yh, setting yh = 1.46 - 0.28 × e -（TD-TD1） Where e is the base of the natural logarithm, the optimized preset available computing power SL0` is obtained, SL0` is set to SL0×yh, the preset available computing power SL0 is replaced with the optimized preset available computing power SL0`, and the available computing power SL is re-compared with the preset available computing power SL0; When TD > TD2, the attenuation monitoring unit determines that the usage time is long, optimizes the preset available computing power SL0 according to the optimization coefficient yh, and corrects the computing power of the result model according to the attenuation value.

10. The SVG power grid dynamic network construction and simulation system combining digital twins according to claim 9, characterized in that, When the attenuation monitoring unit performs computational power correction on the result model based on the attenuation degree value, it inputs the grid voltage, grid current, and equipment temperature from the grid multimodal data into the attenuation prediction model to obtain the attenuation degree value SU output by the attenuation prediction model. The attenuation degree value SU is then compared with a preset attenuation degree value SU0. Based on the comparison result, the attenuation impact status is judged, and the computational power is corrected on the result model based on the judgment result. Wherein: When SU≤SU0, the attenuation monitoring unit determines that the attenuation effect is no effect and does not perform computational power correction on the result model; When SU>SU0, the attenuation monitoring unit determines that the attenuation effect is present and performs computational correction on the result model: attenuation correction is performed on the result model to obtain the corrected result model, the result model is replaced with the corrected result model, and the result model is output.

Images