Multi-source data fusion power grid equipment digital twin modeling method and system

Abstract

The invention discloses a multi-source data fusion power grid equipment digital twin modeling method and system. The method comprises the steps of S1, collecting original sensing data, sensing node working load power, power grid harmonic frequency, harmonic amplitude proportions of all orders and equipment equivalent load power of power grid equipment; s2, calculating a dynamic error factor of sensing node aging, and correcting the original sensing data by adopting the factor; s3, calculating a nondimensionalized harmonic accumulation interference coefficient, and correcting the data after aging correction by adopting the coefficient; and S4, calculating a temperature rise space-time coupling correction amount based on the coupling effect of the harmonic accumulation interference coefficient and the equivalent load power, introducing a reference temperature to construct a dimensionless temperature ratio to correct harmonic corrected data, obtaining real fusion data, and inputting the real fusion data into a twin model to iterate and update parameters. According to the method, the problem of model precision drift is solved through correction logic of progressive linkage, and the stability and accuracy of twin modeling are improved.

Description

Technical Field

[0001] This invention relates to the field of digital twin modeling technology for power grid equipment, specifically to a method and system for digital twin modeling of power grid equipment using multi-source data fusion. Background Technology

[0002] In the process of digital twin modeling of power grid equipment, multi-source data fusion is the core step in building a high-precision twin model. Current technologies often employ independent static compensation methods for processing sensor data, harmonic data, and temperature rise data. This involves separately correcting the fixed errors of sensor nodes, single-harmonic interference, and average temperature rise, completely ignoring the progressive and interconnected relationship between the three: aging of sensor nodes due to long-term operation reduces the accuracy of harmonic signal perception, while the continuous accumulation of multi-order harmonics in the power grid exacerbates internal power losses, leading to spatiotemporal differences in local temperature rise. The increased temperature further accelerates the aging process of sensor nodes. This fragmented approach causes the accuracy of the twin model to continuously drift over time. Even with periodic calibration, it is difficult to maintain stable modeling accuracy. For complex power grid scenarios with high loads and multiple harmonics, such as industrial parks, model errors quickly exceed the allowable range for practical applications, failing to meet the needs of power grid equipment fault early warning and condition assessment.

[0003] Based on the above problems, there is an urgent need for a multi-source data fusion method that can consider the progressive and interconnected effects of multiple factors in order to improve the accuracy and stability of digital twin modeling of power grid equipment. Summary of the Invention

[0004] This invention provides a method for digital twin modeling of power grid equipment based on multi-source data fusion, comprising:

[0005] Step S1: Collect raw sensor data of power grid equipment, working load power of sensor nodes, power grid harmonic frequency, amplitude ratio of each order of harmonics and equivalent load power of equipment;

[0006] Step S2: Calculate the dynamic error factor of sensor node aging, and use the dynamic error factor of sensor node aging to correct the aging error of the original sensor data.

[0007] Step S3: Calculate the dimensionless harmonic cumulative interference coefficient, and use the dimensionless harmonic cumulative interference coefficient to correct the harmonic interference of the data after aging error correction;

[0008] Step S4: Calculate the temperature rise spatiotemporal coupling correction based on the coupling effect of the dimensionless harmonic cumulative interference coefficient and the equivalent load power of the equipment. Introduce a reference temperature to construct a dimensionless temperature ratio. Use the dimensionless temperature ratio to perform temperature rise sensitivity correction on the harmonic interference corrected data to obtain real fused data. Input the real fused data into the digital twin model of the power grid equipment and iteratively update the model parameters.

[0009] Preferably, the raw sensing data in step S1 includes current data, voltage data, and temperature data of the power grid equipment, which are collected by sensing nodes deployed in the conductor area and shell area inside the power grid equipment.

[0010] More preferably, the dynamic error factor of sensor node aging in step S2 is calculated based on the time integral of the sensor node's workload power and the 1.5-power decay law of the running time, wherein the time integral is calculated using the trapezoidal integral method.

[0011] More preferably, the dimensionless harmonic cumulative interference coefficient in step S3 is obtained by constructing a dimensionless time ratio by introducing harmonic characteristic time, and calculating it in combination with the proportion of harmonic amplitude of each order and the total order of harmonics from the 3rd to the 12th order. The harmonic characteristic time is the characteristic duration of the continuous action of harmonics under the rated operating state of the power grid equipment.

[0012] More preferably, the dynamic error factor of the aging of the sensor node is calculated by the definition formula of the dynamic error factor of the aging of the sensor node. The definition formula of the dynamic error factor of the aging of the sensor node is based on the joint aging mechanism of electrical stress and time stress of semiconductor sensing element. The formula includes the initial error coefficient of the sensor node at the factory, the load loss aging influence coefficient, the time integral term of the working load power of the sensor node, the time aging attenuation coefficient, and the 1.5 power term of the running time.

[0013] More preferably, the dimensionless harmonic cumulative interference coefficient is calculated by the harmonic cumulative interference coefficient correction formula. The harmonic cumulative interference coefficient correction formula is constructed based on the multi-order harmonic time cumulative amplification effect. The formula includes the initial interference coefficient of the frequency f harmonic, the harmonic cumulative amplification coefficient, the amplitude ratio of each order harmonic, the dimensionless time ratio, the harmonic order, and the total harmonic order term.

[0014] More preferably, the temperature rise spatiotemporal coupling correction is calculated using a temperature rise spatiotemporal coupling correction definition formula. The temperature rise spatiotemporal coupling correction definition formula is constructed based on the thermodynamic heat conduction law and the harmonic-load coupled heat source effect. The formula includes the initial temperature of the environment where the equipment is located, the harmonic-load coupled temperature rise coefficient, the time integral term of the harmonic cumulative interference coefficient, the time integral term of the equipment's equivalent load power, the coordinates of the power grid equipment location, and the equipment's heat conduction characteristic length term.

[0015] More preferably, the iterative update of model parameters in step S4 includes updating the equivalent impedance parameters of the power grid equipment and the characteristic parameters of the equipment insulation material. The iterative update is implemented based on the finite element analysis algorithm, and the mesh division of the finite element analysis algorithm is performed according to the different accuracy requirements of the critical and non-critical areas of the power grid equipment.

[0016] A multi-source data fusion-based digital twin modeling system for power grid equipment, applied to any of the aforementioned multi-source data fusion-based digital twin modeling methods for power grid equipment, includes a multi-source data acquisition module, a dynamic error correction module, a harmonic cumulative interference correction module, a temperature rise spatiotemporal coupling correction module, and a twin model iterative calibration module. The multi-source data acquisition module is used to acquire the raw sensor data, sensor node operating load power, power grid harmonic frequencies, the amplitude ratio of each order of harmonics, and the equivalent load power of the equipment mentioned in step S1, and transmit them to the dynamic error correction module. The dynamic error correction module is used to execute... The calculation and correction operation in step S2 transmits the data after aging error correction to the harmonic cumulative interference correction module; the harmonic cumulative interference correction module is used to perform the calculation and correction operation in step S3, transmitting the data after harmonic interference correction to the temperature rise spatiotemporal coupling correction module; the temperature rise spatiotemporal coupling correction module is used to perform the calculation and correction operation in step S4 of claim 1, transmitting the obtained real fusion data to the twin model iterative calibration module; the twin model iterative calibration module is used to input the real fusion data into the digital twin model of the power grid equipment and iteratively update the model parameters.

[0017] More preferably, the multi-source data acquisition module includes a current sensing unit, a voltage sensing unit, a temperature sensing unit, a harmonic sensing unit, and a data synchronization acquisition unit. The current sensing unit is used to acquire current data of the power grid equipment, the voltage sensing unit is used to acquire voltage data of the power grid equipment, the temperature sensing unit is used to acquire temperature data of the power grid equipment, the harmonic sensing unit is used to acquire the harmonic frequency of the power grid and the proportion of each order harmonic amplitude, and the data synchronization acquisition unit is used to realize the synchronous acquisition and transmission of data from each sensing unit. The dynamic error correction module has a built-in trapezoidal integral algorithm, the harmonic cumulative interference correction module has a built-in harmonic frequency identification algorithm and harmonic amplitude proportion extraction logic, the temperature rise spatiotemporal coupling correction module has a built-in heat conduction equation solving logic, and the twin model iterative calibration module has a built-in finite element analysis algorithm.

[0018] The present invention has the following beneficial effects:

[0019] This invention overcomes the limitations of existing technologies that fragmentedly process multi-source data by constructing a progressive linkage correction logic of sensor aging, harmonic accumulation, and temperature rise coupling. Among them, the calculation of dynamic error factors based on load integral and time power decay, the construction of dimensionless cumulative interference coefficient by introducing harmonic characteristic time, and the design of temperature rise spatiotemporal correction amount that integrates harmonic-load coupling are all creative technical points for solving model accuracy drift. It effectively considers the mutual influence of multiple factors, suppresses the continuous drift of twin model accuracy from the root, and solves the core problem of insufficient model stability in complex power grid scenarios in the background technology. Attached Figure Description

[0020] Figure 1 This application provides a flowchart of a digital twin modeling method for power grid equipment based on multi-source data fusion;

[0021] Figure 2 This application provides a connection block diagram for a digital twin modeling system for power grid equipment that integrates multi-source data. Detailed Implementation

[0022] 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.

[0023] Traditional multi-source data fusion schemes for digital twin modeling of power grid equipment do not consider the progressive linkages of sensor aging, harmonic accumulation, and temperature rise coupling, leading to model accuracy drift over time. Therefore, please refer to [link to relevant documentation]. Figure 1-2This embodiment provides a method for digital twin modeling of power grid equipment based on multi-source data fusion, including: Step S1: Collecting original sensor data of power grid equipment, working load power of sensor nodes, harmonic frequencies of the power grid, amplitude ratio of each order of harmonics, and equivalent load power of the equipment; Step S2: Calculating the dynamic error factor of sensor node aging, and using the dynamic error factor of sensor node aging to correct the aging error of the original sensor data; Step S3: Calculating the dimensionless harmonic cumulative interference coefficient, and using the dimensionless harmonic cumulative interference coefficient to correct the harmonic interference of the data after aging error correction; Step S4: Calculating the temperature rise spatiotemporal coupling correction amount based on the coupling effect of the dimensionless harmonic cumulative interference coefficient and the equivalent load power of the equipment, introducing a reference temperature to construct a dimensionless temperature ratio, and using the dimensionless temperature ratio to correct the temperature rise sensitivity of the data after harmonic interference correction to obtain real fused data, inputting the real fused data into the digital twin model of the power grid equipment, and iteratively updating the model parameters.

[0024] The core technical logic of this solution lies in constructing a progressive and interconnected multi-factor correction system to achieve accurate fusion of multi-source data. Firstly, the data acquisition step S1 forms the foundation for subsequent corrections, requiring the integrity and temporal consistency of the multi-source data. Specifically, this involves synchronously collecting raw sensor data, sensor node operating load power, grid harmonic frequencies, the proportion of each order harmonic amplitude, and the equivalent load power of the equipment through sensor units deployed in different areas of the power grid. The raw sensor data represents the core electrical parameters of the power grid equipment operation; the sensor node operating load power reflects the operating load status of the sensor unit itself; the grid harmonic frequencies and the proportion of each order harmonic amplitude characterize the harmonic interference characteristics of the power grid; and the equivalent load power of the equipment comprehensively reflects the actual operating load of the power grid equipment. This synchronous acquisition of data is achieved through a data synchronization triggering mechanism, ensuring complete consistency in the timestamps of different data types, providing a time-matched data source for subsequent coupled corrections.

[0025] The core of step S2 is to overcome the limitations of traditional fixed error coefficients and achieve dynamic quantitative correction of sensor node aging errors. In specific implementation, it is necessary to calculate the dynamic error factor of sensor node aging based on the combined aging mechanism of electrical stress and time stress in semiconductor sensing elements. The calculation formula is as follows: .in, The initial error coefficient of the sensor node at the time of shipment is a dimensionless proportionality coefficient determined by the manufacturing process of the sensor element. It is obtained through calibration on a standard testing platform before shipment. Different models of sensor elements have different initial error coefficients. Different values, for example, the current sensing element of a certain model. The measured value was 0.02. The load loss aging influence coefficient has the following dimensions: This coefficient was determined through a load aging experiment on the sensing element. During the experiment, different levels of load power were applied to the sensing element, and the error change data under different loads were recorded. The error increment corresponding to unit energy loss was obtained through linear fitting. The value of ; The time integral of the sensor node's workload power from time τ to time t is given. Let be the working load power at time τ, with dimensions . The dimensions after integration are The unit of energy is used in the calculation, employing the trapezoidal integral method. The running time t is divided into N equally spaced time segments, each segment having a duration of... Collect the start time of each segment and end time Working load power and The integral value of that segment is... The total integral is obtained by summing the integral values ​​of all the smaller segments. The time aging decay coefficient has the following dimensions: Through long-term natural aging experiments of the sensing element, the error changes of the sensing element were continuously monitored under no-load conditions. Error data corresponding to different operating times were recorded, and the variation law of error over time was obtained by power function fitting, thereby determining the error. The value of ; The term is a power of 1.5 for the running time, with dimensions of... The choice of this power is based on statistical analysis of aging data from a large number of different types of semiconductor sensing elements. The results show that 1.5 powers most accurately match the aging rate variation of semiconductor materials over time, avoiding the biases caused by linear or quadratic power fitting. The entire formula dynamically quantifies the aging error of the sensing node through the superposition of electrical stress and time stress terms, ultimately yielding... For a dimensionless quantity, divide the original sensing data by This will allow for the correction of aging errors.

[0026] The core of step S3 is to resolve the conflict between the dimensions of time accumulation of multiple harmonics in traditional harmonic interference correction, thereby achieving accurate quantification of harmonic cumulative interference. Specifically, the dimensionless harmonic cumulative interference coefficient is calculated based on the time accumulation amplification effect of multiple harmonics. The calculation formula is as follows: .in, Let f be the initial interference coefficient of the harmonic of frequency f. It is dimensionless and determined by the matching characteristics between the harmonic frequency f and the impedance of the power grid equipment. An impedance analyzer is used to test the impedance changes of the equipment under the influence of different harmonic frequencies, thereby determining the initial interference level of different harmonic frequencies on data acquisition. For example, the harmonic of frequency f corresponds to... The measured value was 0.03. The harmonic cumulative amplification factor is dimensionless and reflects the degree of amplification of the interference effect under continuous harmonic action. It is determined by experiments on the impedance nonlinear characteristics of power grid equipment. In the experiment, harmonic signals of different durations are applied, and the changes in the interference level are monitored. The amplification factor is obtained through nonlinear fitting. The value of ; The value of the nth harmonic is a dimensionless proportion of the amplitude of the nth harmonic. It is obtained by collecting the total harmonic amplitude and the amplitude of each order of harmonics in the power grid using a harmonic analyzer, and then calculating the ratio of the nth harmonic amplitude to the total harmonic amplitude. The harmonic duration is expressed in T. The harmonic characteristic time, measured in T, is the shortest duration under rated operating conditions of power grid equipment where harmonics continuously act and significantly interfere with the equipment's data acquisition. It is determined through harmonic interference experiments under rated operating conditions, for example, for a certain type of transformer. The experiment determined the time to be 10 minutes. The time ratio is a dimensionless time ratio. By introducing this time ratio, the duration of harmonics of different lengths is transformed into a dimensionless quantity, thus avoiding the problem of conflicting dimensions of time terms of different orders of harmonics. n is the harmonic order, which ranges from 1 to 12. The determination of this range is based on the actual distribution characteristics of harmonics in the power grid. Statistical data shows that more than 95% of the harmonic energy in the power grid is concentrated in the 1st to 12th order, and the amplitude of higher order harmonics accounts for a very low proportion, so their impact on interference can be ignored. This is a dimensionless normalization coefficient representing the total harmonic order. It is used to balance the weights of different order harmonics in the accumulated interference, avoiding excessive influence of high-order or low-order harmonics on the overall interference coefficient. During calculation, the amplitude proportion of each order harmonic is first acquired through a harmonic sensing unit. Determine the harmonic duration t, and then calculate the dimensionless time ratio. Then, substitute the values ​​into the formula to calculate the harmonic cumulative interference coefficient. This coefficient is a dimensionless quantity; it is the result of multiplying the data after aging error correction by... Harmonic interference correction can then be completed, where k is the harmonic interference weighting coefficient, which is determined experimentally.

[0027] The core of step S4 is to achieve spatiotemporal correction of temperature rise under the coupling effect of harmonic accumulation and load power, and to complete the final data fusion and model update. Specifically, the spatiotemporal coupling correction amount of temperature rise is first calculated based on the thermodynamic heat conduction laws and the harmonic-load coupled heat source effect. The calculation formula is as follows: .in, The initial temperature of the environment in which the equipment is located, with dimensions of The data is collected in real time by temperature sensing units deployed around the equipment. The harmonic-load coupling temperature rise coefficient has the following dimensions: This reflects the degree of temperature increase caused by unit harmonic-load coupling energy. It is determined through thermal characteristic experiments on the equipment. Different levels of harmonics and load power are applied in the experiments, and the temperature change data is monitored. The results are obtained through linear fitting. The value of ; Let be the harmonic cumulative interference coefficient at time τ, which is dimensionless; Let be the equivalent load power of the equipment at time τ, with dimensions . The dimensions of the product of the two are: The dimensions after integration are , i.e., energy unit, this integral term is the harmonic-load coupled heat source term, which is calculated using the trapezoidal integration method; and These are the location coordinates of the power grid equipment, with the dimension L. They are determined by the structural design drawings of the equipment and are used to characterize the spatial information of different locations of the equipment. The characteristic length of thermal conduction of the equipment, with dimensions L, is determined by the structural dimensions and thermal conductivity of the equipment and is obtained through thermal conduction experiments. This is an exponential term, dimensionless, reflecting the spatial attenuation characteristics of heat conduction; the farther away from the center of the heat source, the greater the attenuation. The larger the value, the weaker the temperature rise. The calculated spatiotemporal coupling correction for temperature rise is... Then, a reference temperature is introduced to construct a dimensionless temperature ratio. The reference temperature is the rated operating temperature of the power grid equipment, determined by the equipment's technical parameters. The dimensionless temperature ratio is... Multiply the harmonic interference corrected data by Temperature rise sensitivity correction can then be completed, among which The temperature-sensing sensitivity coefficient is dimensionless and determined experimentally. After completing the above three progressive correction steps, the actual fused data is obtained. This data is input into the digital twin model of the power grid equipment, and the model parameters are iteratively updated through the finite element analysis algorithm. This includes the equivalent impedance parameters of the power grid equipment and the characteristic parameters of the equipment insulation materials. During the iteration process, the goal is to minimize the deviation between the actual fused data and the model output data until the deviation meets the set convergence condition, thus completing the model parameter update.

[0028] Traditional multi-source data acquisition focuses only on surface data of equipment, lacking monitoring of critical internal areas, resulting in insufficient data integrity. Therefore, according to the method described above, the raw sensor data in step S1 includes current, voltage, and temperature data of the power grid equipment. This data is collected by sensor nodes deployed in the conductor and casing areas inside the equipment. Specifically, the deployment locations of the sensor nodes are determined based on the structural design of the power grid equipment. The internal conductor area is the core current-carrying component area of ​​the equipment; the current and voltage data in this area directly reflect the core operating status of the equipment. Therefore, Hall effect current sensing units and voltage divider voltage sensing units are deployed in this area. The Hall effect current sensing unit collects current data through electromagnetic induction, while the voltage divider voltage sensing unit collects voltage data through resistive voltage division. The casing area is the external protection area of ​​the equipment; the temperature data in this area reflects the external heat dissipation status of the equipment. Therefore, platinum resistance temperature sensing units are deployed in this area; these units collect temperature data based on the change in resistance value with temperature. All sensor nodes are secured using standardized mounting brackets, ensuring robust installation without affecting normal equipment operation. The acquisition frequency of each sensor node is uniformly set to 100Hz to ensure data timeliness. A data synchronization acquisition unit enables synchronized data acquisition from all sensor nodes. This unit generates a unified clock trigger signal to control the acquisition time of each sensor node, ensuring that data acquired from different sensor nodes have the same timestamp and avoiding the impact of timing discrepancies on subsequent data fusion. The acquired current, voltage, and temperature data are transmitted to the data processing center via wired transmission, providing a comprehensive and accurate data source for subsequent multi-factor correction.

[0029] Traditional sensing error correction uses fixed coefficients, which cannot match the dynamic aging process of sensing nodes. Therefore, according to the above method, the dynamic error factor for sensing node aging in step S2 is calculated based on the time integral of the sensing node's workload power and the 1.5-power decay law of the running time. The time integral is calculated using the trapezoidal integration method. Specifically, the trapezoidal integration method is executed as follows: First, the integration time interval Δτ is determined, and Δτ is set to 0.01 seconds based on the sensing node's sampling frequency of 100Hz; then, the workload power data for all sampling times from time τ to time t are acquired to form a power sequence. ,in , , Next, calculate each time segment. The integral value, the integral value of each small segment is Finally, the integral values ​​of all time segments are summed to obtain the time integral value of the sensor node's workload power from time τ to time t. The implementation of the 1.5-power decay law is based on an aging characteristic experiment of the sensor node. The experiment uses the same model of sensor node as in actual applications, running continuously for 10,000 hours in an environment without external interference. Error data of the sensor node is recorded every 10 hours, obtaining the error variation curve over time. A power function is used to fit this curve; the fitting function is... The value of β, i.e., the time aging decay coefficient, is determined by the least squares method. In actual calculation, the time integral value of the working load power is first calculated by the trapezoidal integral method, then the 1.5th power term of the running time is calculated, and then the integral value is multiplied by α, the 1.5th power term is multiplied by β, the two are added together and then multiplied by 1, and then multiplied by the initial error coefficient ε0 to obtain the dynamic error factor of sensor node aging. Finally, the original sensor data is divided by 1 and the sum of the dynamic error factor is added to complete the aging error correction.

[0030] Traditional harmonic interference correction does not consider the time accumulation effect of multi-order harmonics and suffers from dimensional conflicts. Therefore, according to the method described above, the dimensionless harmonic cumulative interference coefficient in step S3 is constructed by introducing the harmonic characteristic time to create a dimensionless time ratio. This ratio is calculated by combining the amplitude proportions of each order of harmonics and the total order of harmonics from 3rd to 12th. The harmonic characteristic time is the characteristic duration of harmonic effects under rated operating conditions of the power grid equipment. In practice, the harmonic characteristic time is determined through harmonic interference experiments under rated operating conditions. During the experiment, the power grid equipment operates at rated voltage and rated load, and standard harmonic signals of different durations are applied. The interference level of the harmonic signals on the equipment's data acquisition is monitored. When the interference level reaches twice the initial interference level, the harmonic duration at this point is recorded; this time is the harmonic characteristic time t0. The proportion of each order harmonic amplitude is obtained through a harmonic sensing unit. The unit uses a Fourier transform algorithm to perform harmonic analysis on the acquired grid voltage signal, decomposing it to obtain the amplitudes of the 1st to 12th order harmonics and the total harmonic amplitude. The ratio of each order harmonic amplitude to the total harmonic amplitude is then calculated to obtain the proportion H of each order harmonic amplitude. n (f) The dimensionless time ratio is calculated as the ratio of the harmonic duration t to the harmonic characteristic time t0, i.e. In actual calculations, the total harmonic order is first determined to be from 3 to 12, and then the terms corresponding to each order n are calculated. The terms corresponding to the 3rd to 12th orders are summed to obtain the summation result. Then, the summation result is multiplied by the harmonic cumulative amplification factor γ, and after adding 1, it is multiplied by the initial interference factor η0(f) to obtain the dimensionless harmonic cumulative interference factor. Finally, the data after aging error correction is multiplied by 1 and the product of the harmonic cumulative interference factor and the harmonic interference weight factor k is subtracted to complete the harmonic interference correction.

[0031] Traditional calculations of sensor error factors lack clear mechanistic support, and the formula construction is highly arbitrary. Therefore, based on the aforementioned method, the dynamic error factor of sensor node aging is calculated using a defined formula for the dynamic error factor of sensor node aging. This formula is based on the joint aging mechanism of electrical stress and time stress in semiconductor sensing elements. The formula includes the initial factory error coefficient of the sensing node, the load loss aging influence coefficient, the time integral term of the sensing node's operating load power, the time aging decay coefficient, and a 1.5 power term of the operating time. In practice, the formula construction process is based on the aging mechanism of semiconductor materials. The aging of semiconductor sensing elements is mainly caused by the combined effects of electrical stress and time stress. Electrical stress refers to the impact of energy loss caused by load power on element performance during operation, while time stress refers to the impact of natural aging of the element over time on performance. To quantify the impact of electrical stress, a time integral term of the sensing node's operating load power is introduced. This integral term reflects the total electrical stress borne by the element during operation. The load loss aging influence coefficient α is used to convert electrical stress into error increment. To quantify the impact of time stress, a 1.5 power term of the running time is introduced. This power term is determined based on extensive aging experimental data of semiconductor components and can accurately match the aging rate variation of semiconductor materials over time. The time aging decay coefficient β is used to convert time stress into error increment. The initial error coefficient ε0 reflects the inherent error of the component at the time of manufacture and is the benchmark value of the error factor. The derivation process of the formula is as follows: First, the relationship between electrical stress and error increment is determined experimentally, thus obtaining the electrical stress error term. Then, the relationship between time stress and error increment was determined experimentally, and the time stress error term was obtained. Next, the electrical stress error term and the time stress error term are added together to obtain the total error increment coefficient. Adding 1 gives the error amplification coefficient. Finally, the error amplification coefficient is multiplied by the initial error coefficient to obtain the dynamic error factor of the sensor node aging. All parameters in the formula are determined experimentally to ensure that the formula can accurately reflect the actual aging process of semiconductor sensing elements and provide a scientific quantitative basis for the dynamic correction of sensing errors.

[0032] Traditional harmonic interference coefficient calculations do not consider the weighted accumulation of multiple harmonics, resulting in insufficient correction accuracy. Therefore, based on the aforementioned method, the dimensionless harmonic cumulative interference coefficient is calculated using a harmonic cumulative interference coefficient correction formula. This formula is constructed based on the time-cumulative amplification effect of multiple harmonics and includes the initial interference coefficient of the frequency f harmonic, the harmonic cumulative amplification coefficient, the amplitude proportion of each harmonic order, the dimensionless time ratio, the harmonic order, and the total harmonic order term. In practice, the formula construction process is based on the interference mechanism of multiple harmonics. Harmonics in the power grid are not of a single order; different orders of harmonics have different degrees of interference to equipment data acquisition, and the interference effect accumulates continuously over time. To quantify the interference contribution of different orders of harmonics, the amplitude proportion H of each order of harmonics is introduced. n (f) Harmonics with larger amplitude proportions contribute more to the total interference. To quantify the time-cumulative effect of harmonic interference, a power term of the dimensionless time ratio is introduced. The interference accumulation rate varies for harmonics of different orders, which is reflected in the difference in the power n; higher-order harmonics have a faster interference accumulation rate. To balance the weight of each order of harmonics, a normalization coefficient for the total order of harmonics is introduced. To avoid excessive influence of a certain order harmonic on the total interference coefficient, the harmonic cumulative amplification factor γ is used to reflect the amplification degree of the interference effect under the continuous action of harmonics, and is determined by the nonlinear characteristics of the equipment impedance. The initial interference coefficient η0(f) reflects the initial interference degree of different frequency harmonics. The derivation process of the formula is as follows: First, the interference accumulation law of different order harmonics is determined experimentally, and the interference term of each order harmonic is obtained. Then, the interference terms of each order harmonic are weighted and summed to obtain the total interference accumulation coefficient. Next, the total interference accumulation coefficient is multiplied by the harmonic accumulation amplification coefficient to obtain the total interference amplification term, and 1 is added to obtain the interference amplification coefficient. Finally, the interference amplification coefficient is multiplied by the initial interference coefficient to obtain the dimensionless harmonic accumulation interference coefficient. All parameters of the formula are determined experimentally to ensure that the formula can accurately reflect the time accumulation amplification effect of multi-order harmonics, providing a scientific quantitative basis for the accurate correction of harmonic interference.

[0033] Traditional temperature rise correction uses average temperature rise, ignoring spatiotemporal distribution differences, resulting in poor correction effectiveness. Therefore, based on the aforementioned method, the spatiotemporal coupling correction amount for temperature rise is calculated using a defined formula. This formula is constructed based on thermodynamic heat conduction laws and the harmonic-load coupling heat source effect. The formula includes the initial temperature of the equipment's environment, the harmonic-load coupling temperature rise coefficient, the time integral term of the harmonic cumulative interference coefficient, the time integral term of the equipment's equivalent load power, the coordinates of the power grid equipment, and the equipment's heat conduction characteristic length. In practice, the formula construction process is based on thermodynamic heat conduction theory and the heat source mechanism of harmonic-load coupling. The temperature rise of power grid equipment is mainly caused by the heat generated by load power loss, and the presence of harmonics exacerbates load power loss, forming an additional heat source, namely the harmonic-load coupling heat source. To quantify the heat of the harmonic-load coupling heat source, the time integral terms of the harmonic cumulative interference coefficient and the equipment's equivalent load power are introduced. This integral term reflects the total heat generated by the coupled heat sources. The harmonic-load coupling temperature rise coefficient δ is used to convert the total heat into a temperature increment. To quantify the spatial distribution differences in temperature rise, a spatial attenuation term for heat conduction is introduced. This attenuation term, derived from the heat conduction equation, reflects the temperature decay from the center of the heat source to the periphery. x and y are the position coordinates, and L is the characteristic length of heat conduction. The initial ambient temperature T0 is the baseline value for temperature rise calculation. The derivation process is as follows: First, the spatial attenuation law of temperature distribution is derived from the heat conduction equation, resulting in the spatial attenuation term. Then, the relationship between harmonic-load coupled heat and temperature increment was determined experimentally, and the temperature increment term was obtained. Next, the temperature increment term is multiplied by the spatial attenuation term to obtain the spatially distributed temperature increment; finally, the spatially distributed temperature increment is added to the initial ambient temperature to obtain the temperature rise spatiotemporal coupling correction. All parameters in the formula were determined experimentally to ensure that the formula accurately reflects the spatiotemporal distribution characteristics of temperature rise at different locations of the equipment, providing a scientific quantitative basis for precise temperature rise correction.

[0034] Traditional model parameter updates do not differentiate between the accuracy requirements of different equipment regions, resulting in an imbalance between modeling efficiency and accuracy. Therefore, according to the method described above, the iterative update of model parameters in step S4 includes updating the equivalent impedance parameters of the power grid equipment and the characteristic parameters of the equipment's insulation materials. This iterative update is implemented based on the finite element analysis (FEM) algorithm, and the mesh generation of the FEM algorithm is performed according to the different accuracy requirements of critical and non-critical regions of the power grid equipment. Specifically, critical regions of the power grid equipment include internal conductor connections and weak insulation areas. The operating status of these areas directly affects the safe and stable operation of the equipment, thus requiring higher modeling accuracy. Non-critical regions include the external areas of the equipment casing. These areas have a smaller impact on the overall operating status of the equipment, and the modeling accuracy can be appropriately reduced. The mesh generation process of the FEM algorithm is as follows: First, a three-dimensional structural model of the power grid equipment is obtained. Then, the model is divided into critical and non-critical regions. The mesh size for critical regions is set to no more than 5 mm, and the mesh size for non-critical regions is set to no more than 20 mm. Tetrahedral meshes are used to refine the critical regions, ensuring that the mesh can accurately represent the structural details of the region. Hexahedral meshes are used to divide the non-critical regions, improving modeling efficiency while ensuring modeling accuracy. The update process for the equivalent impedance parameters is as follows: The fused current and voltage data are input into the finite element model to calculate the impedance distribution in each region of the equipment. The impedance parameters are then compared with those of the initial model to obtain a correction value. Finally, the initial impedance parameters are added to the correction value to obtain the updated equivalent impedance parameters. The characteristic parameters of the equipment's insulation material include dielectric constant and breakdown field strength. Their update process involves fitting the fused temperature data with experimental data on the insulation material's characteristics to obtain a relationship model between temperature and characteristic parameters. Based on real-time temperature data, the current values ​​of the characteristic parameters are calculated using this relationship model, and the initial characteristic parameters are updated to these current values. The convergence condition for the iterative update is set to a deviation of less than 0.5% between the fused data and the model output data. After each iteration, the deviation is calculated. If the deviation is greater than or equal to 0.5%, the parameter update process is repeated until the deviation is less than 0.5%, completing the iterative update of the model parameters.

[0035] Traditional digital twin modeling systems lack modular, interconnected design, resulting in fragmented data processing workflows. To address this, a multi-source data fusion-based digital twin modeling system for power grid equipment includes a multi-source data acquisition module, a dynamic error correction module, a harmonic cumulative interference correction module, a temperature rise spatiotemporal coupling correction module, and a twin model iterative calibration module. The multi-source data acquisition module collects the raw sensor data mentioned in step S1, including sensor node operating load power, power grid harmonic frequencies, the amplitude proportion of each order of harmonics, and the equivalent load power of the equipment, and transmits this data to the dynamic error correction module. The dynamic error correction module performs the calculation and correction operations of step S2, transmitting the data after aging error correction to the harmonic cumulative interference correction module. The harmonic cumulative interference correction module performs the calculation and correction operations of step S3, transmitting the data after harmonic interference correction to the temperature rise spatiotemporal coupling correction module. The temperature rise spatiotemporal coupling correction module performs the calculation and correction operations of step S4, transmitting the obtained real fused data to the twin model iterative calibration module. The twin model iterative calibration module inputs the real fused data into the digital twin model of the power grid equipment and iteratively updates the model parameters. In practical implementation, each module transmits data via a high-speed data bus, with the data bus transmission rate set to 1Gbps to ensure real-time data transmission. The multi-source data acquisition module operates as follows: each sensing unit synchronously acquires data according to a set acquisition frequency. The data synchronization acquisition unit standardizes the acquired data, converting it to a unified binary format and adding a timestamp. Then, the standardized data is transmitted to the dynamic error correction module via the data bus. The dynamic error correction module receives data from the multi-source data acquisition module, extracts the sensor node's operating load power data, calculates the dynamic error factor for sensor node aging, and then uses this factor to correct the aging error of the original sensing data. After correction, the data is transmitted to the harmonic cumulative interference correction module. The harmonic cumulative interference correction module receives data from the dynamic error correction module, extracts the grid harmonic frequency and the proportion of each order harmonic amplitude, calculates the dimensionless harmonic cumulative interference coefficient, and then uses this coefficient to correct the harmonic interference of the aging error-corrected data. After correction, the data is transmitted to the temperature rise spatiotemporal coupling correction module. The workflow of the temperature rise spatiotemporal coupling correction module is as follows: it receives data transmitted from the harmonic cumulative interference correction module, extracts the equivalent load power data of the equipment from it, calculates the temperature rise spatiotemporal coupling correction amount, introduces a reference temperature to construct a dimensionless temperature ratio, uses this ratio to correct the temperature rise sensitivity of the harmonic interference corrected data, obtains the true fused data, and transmits it to the twin model iterative calibration module.The workflow of the twin model iterative calibration module is as follows: it receives the real fused data transmitted by the temperature rise spatiotemporal coupling correction module, inputs it into the digital twin model of the power grid equipment, iteratively updates the model parameters through the finite element analysis algorithm until the deviation between the fused data and the model output data meets the convergence condition, and completes the model iterative calibration.

[0036] Traditional multi-source data acquisition modules have limited functionality and poor data synchronization; each correction module lacks dedicated algorithm support, resulting in insufficient processing accuracy. Therefore, based on the aforementioned system, the multi-source data acquisition module includes a current sensing unit, a voltage sensing unit, a temperature sensing unit, a harmonic sensing unit, and a data synchronization acquisition unit. The current sensing unit collects current data from the power grid equipment, the voltage sensing unit collects voltage data, the temperature sensing unit collects temperature data, the harmonic sensing unit collects the harmonic frequencies and the proportion of each order harmonic amplitude, and the data synchronization acquisition unit enables synchronous acquisition and transmission of data from all sensing units. The dynamic error correction module incorporates a trapezoidal integral algorithm, the harmonic cumulative interference correction module incorporates a harmonic frequency identification algorithm and harmonic amplitude proportion extraction logic, the temperature rise spatiotemporal coupling correction module incorporates heat conduction equation solving logic, and the twin model iterative calibration module incorporates a finite element analysis algorithm. In specific implementation, the current sensing unit uses a Hall effect current sensor with a measurement range of 0 to 1000A, a measurement error of less than 0.1%, and a working power supply of 24V DC; the voltage sensing unit uses a resistive voltage divider voltage sensor with a measurement range of 0 to 10kV, a measurement error of less than 0.1%, and a working power supply of 24V DC; the temperature sensing unit uses a platinum resistance temperature sensor with a measurement range of -40℃ to 150℃, a measurement error of less than 0.1℃, and a working power supply of 24V DC; the harmonic sensing unit uses a harmonic analyzer based on Fourier transform with a harmonic analysis range of 1 to 50 orders, a frequency resolution of 1Hz, a measurement error of less than 0.5%, and a working power supply of 220V AC. The data synchronization acquisition unit includes a clock generator and a data aggregator. The clock generator produces a standard clock signal with a frequency of 100Hz, which is transmitted to each sensing unit through a synchronization signal line to control the acquisition time of each sensing unit and achieve synchronized data acquisition. The data aggregator receives the data acquired by each sensing unit, performs format conversion and adds timestamps, and then transmits it to subsequent modules through a data bus. The trapezoidal integral algorithm built into the dynamic error correction module is implemented through embedded software written in C language and running on a 32-bit ARM processor with a main frequency of 1GHz to ensure the real-time performance of the algorithm. The harmonic cumulative interference correction module's built-in harmonic frequency identification algorithm is implemented through Fast Fourier Transform, which can quickly and accurately identify the frequency of power grid harmonics. The harmonic amplitude ratio extraction logic is implemented by calculating the ratio of the amplitude of each order harmonic to the total harmonic amplitude. The temperature rise spatiotemporal coupling correction module's built-in heat conduction equation solving logic is implemented through the finite difference method, which can quickly solve the heat conduction equation and obtain the spatial distribution of temperature. The finite element analysis algorithm built into the twin model iterative calibration module is implemented using a secondary development interface of commercial finite element analysis software. The software runs on an industrial control computer with an Intel Core i7 CPU, 32GB of memory, and a 1TB hard drive, ensuring the algorithm's running efficiency and modeling accuracy.The hardware of each module adopts industrial-grade design with an IP65 protection rating, which can adapt to the harsh environment of power grid equipment operation site and ensure the stable and reliable operation of the system.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for digital twin modeling of power grid equipment using multi-source data fusion, characterized in that, include: Step S1: Collect raw sensor data of power grid equipment, working load power of sensor nodes, power grid harmonic frequency, amplitude ratio of each order of harmonics and equivalent load power of equipment; Step S2: Calculate the dynamic error factor of sensor node aging, and use the dynamic error factor of sensor node aging to correct the aging error of the original sensor data. Step S3: Calculate the dimensionless harmonic cumulative interference coefficient, and use the dimensionless harmonic cumulative interference coefficient to correct the harmonic interference of the data after aging error correction; Step S4: Calculate the temperature rise spatiotemporal coupling correction based on the coupling effect of the dimensionless harmonic cumulative interference coefficient and the equivalent load power of the equipment. Introduce a reference temperature to construct a dimensionless temperature ratio. Use the dimensionless temperature ratio to perform temperature rise sensitivity correction on the harmonic interference corrected data to obtain real fused data. Input the real fused data into the digital twin model of the power grid equipment and iteratively update the model parameters.

2. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 1, characterized in that, The raw sensing data in step S1 includes current data, voltage data, and temperature data of the power grid equipment. The current data, voltage data, and temperature data of the power grid equipment are collected by sensing nodes deployed in the conductor area and shell area inside the power grid equipment.

3. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 1, characterized in that, The dynamic error factor of sensor node aging in step S2 is calculated based on the time integral of the sensor node's workload power and the 1.5 power decay law of the running time. The time integral is calculated using the trapezoidal integral method.

4. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 1, characterized in that, The dimensionless harmonic cumulative interference coefficient in step S3 is obtained by constructing a dimensionless time ratio by introducing harmonic characteristic time, and by combining the proportion of harmonic amplitude of each order and the total order of harmonics from the 3rd to the 12th order. The harmonic characteristic time is the characteristic duration of the continuous action of harmonics under the rated operating state of the power grid equipment.

5. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 3, characterized in that, The dynamic error factor of the aging of the sensor node is calculated by the definition formula of the dynamic error factor of the aging of the sensor node. The definition formula of the dynamic error factor of the aging of the sensor node is based on the joint aging mechanism of electrical stress and time stress of semiconductor sensing element. The formula includes the initial error coefficient of the sensor node at the factory, the load loss aging influence coefficient, the time integral term of the sensor node working load power, the time aging attenuation coefficient, and the 1.5 power term of the running time.

6. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 4, characterized in that, The dimensionless harmonic cumulative interference coefficient is calculated using the harmonic cumulative interference coefficient correction formula. The harmonic cumulative interference coefficient correction formula is based on the multi-order harmonic time cumulative amplification effect. The formula includes the initial interference coefficient of the frequency f harmonic, the harmonic cumulative amplification coefficient, the amplitude ratio of each order harmonic, the dimensionless time ratio, the harmonic order, and the total harmonic order term.

7. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 1, characterized in that, The temperature rise spatiotemporal coupling correction is calculated using the temperature rise spatiotemporal coupling correction definition formula. The temperature rise spatiotemporal coupling correction definition formula is constructed based on the thermodynamic heat conduction law and the harmonic-load coupled heat source effect. The formula includes the initial temperature of the environment where the equipment is located, the harmonic-load coupled temperature rise coefficient, the time integral term of the harmonic cumulative interference coefficient, the time integral term of the equipment's equivalent load power, the grid equipment's location coordinates, and the equipment's heat conduction characteristic length term.

8. The method for digital twin modeling of power grid equipment based on multi-source data fusion according to claim 1, characterized in that, The iterative update of model parameters in step S4 includes updating the equivalent impedance parameters of the power grid equipment and the characteristic parameters of the equipment insulation material. The iterative update is implemented based on the finite element analysis algorithm, and the mesh division of the finite element analysis algorithm is performed according to the different accuracy requirements of the critical and non-critical areas of the power grid equipment.

9. A digital twin modeling system for power grid equipment based on multi-source data fusion, applied to the digital twin modeling method for power grid equipment based on multi-source data fusion as described in any one of claims 1-8, characterized in that, The system includes a multi-source data acquisition module, a dynamic error correction module, a harmonic cumulative interference correction module, a temperature rise spatiotemporal coupling correction module, and a twin model iterative calibration module. The multi-source data acquisition module is used to acquire the raw sensor data mentioned in step S1, the sensor node operating load power, the power grid harmonic frequency, the proportion of each order harmonic amplitude, and the equipment equivalent load power, and transmit them to the dynamic error correction module. The dynamic error correction module is used to perform the calculation and correction operation in step S2, and transmit the data after aging error correction to the harmonic cumulative interference correction module. The harmonic cumulative interference correction module is used to perform the calculation and correction operation in step S3, and transmit the harmonic interference corrected data to the temperature rise spatiotemporal coupling correction module. The temperature rise spatiotemporal coupling correction module is used to perform the calculation and correction operation in step S4 of claim 1, and transmit the obtained real fusion data to the twin model iterative calibration module; the twin model iterative calibration module is used to input the real fusion data into the digital twin model of the power grid equipment and iteratively update the model parameters.

10. The system according to claim 9, characterized in that, The multi-source data acquisition module includes a current sensing unit, a voltage sensing unit, a temperature sensing unit, a harmonic sensing unit, and a data synchronization acquisition unit. The current sensing unit is used to collect current data from the power grid equipment, the voltage sensing unit is used to collect voltage data from the power grid equipment, the temperature sensing unit is used to collect temperature data from the power grid equipment, the harmonic sensing unit is used to collect the harmonic frequencies of the power grid and the proportion of each order harmonic amplitude, and the data synchronization acquisition unit is used to realize the synchronous acquisition and transmission of data from each sensing unit. The dynamic error correction module has a built-in trapezoidal integral algorithm, the harmonic cumulative interference correction module has a built-in harmonic frequency identification algorithm and harmonic amplitude proportion extraction logic, the temperature rise spatiotemporal coupling correction module has a built-in heat conduction equation solving logic, and the twin model iterative calibration module has a built-in finite element analysis algorithm.

Images