A current transformer metering error online identification method, device, equipment and medium

Abstract

The present application belongs to the technical field of mutual inductor measurement error identification, and particularly relates to a current transformer measurement error online identification method, device, equipment and medium, wherein the method comprises decomposing the three-phase current signal at the secondary side of the current transformer by VMD, and screening out the residual signal after decomposition; performing wavelet decomposition on the residual signal to obtain an approximate coefficient set; performing nonlinear dimension reduction processing on the approximate coefficient set by a t-distributed stochastic neighbor embedding algorithm to obtain low-dimensional feature data; estimating the probability density distribution of the low-dimensional feature data, and combining the preset confidence to determine the measurement error overrun threshold; comparing the low-dimensional feature data with the overrun threshold, if the low-dimensional feature data is less than the overrun threshold, it indicates that the current transformer is in a normal operating state at this time; otherwise, it is considered that the current transformer may be in an abnormal operating state at this time, solving the problem of inaccurate online identification of current transformer measurement error in the background technology.

Description

Technical Field

[0001] This invention belongs to the field of instrument transformer metering error identification technology, specifically relating to an online identification method, device, equipment, and medium for current transformer metering errors. Background Technology

[0002] As a core device in power systems for energy metering, relay protection, and condition monitoring, the accuracy of current transformers determines the fairness of electricity trade settlement, the reliability of power grid operation control, and the effectiveness of equipment fault diagnosis. With the development of smart grids towards digitalization and refinement, factors such as large-scale grid connection of new energy sources, flexible load fluctuations, and complex electromagnetic environments can lead to problems in current transformers during long-term operation, including core permeability degradation, increased secondary circuit contact resistance, and insulation aging. These issues can cause metering errors to exceed limits, resulting not only in energy metering deviations but also potential relay protection malfunctions, threatening the safe and stable operation of the power grid.

[0003] Currently, the detection of metering errors in current transformers mainly relies on offline verification. Offline verification requires removing the transformer from the power grid and sending it to the laboratory. The error value is obtained by comparing it with a standard transformer under standard operating conditions such as rated load and rated current. However, it has the disadvantages of long power outage time, complicated operation, and inability to reflect the operating status in real time. It is not suitable for transformers at critical nodes of ultra-high voltage and high load, and it is difficult to meet the needs of smart grids for real-time perception of equipment status. Summary of the Invention

[0004] The purpose of this invention is to provide a method, apparatus, device, and medium for online identification of metering errors in current transformers, thereby solving the problem of inaccurate online identification of metering errors in current transformers in the prior art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for online identification of metering errors in current transformers, comprising: VMD is used to decompose the three-phase current signal on the secondary side of the current transformer, and the residual signal is then screened out after decomposition. Perform wavelet decomposition on the residual term signal to obtain the approximate coefficient set; The approximation coefficient set is nonlinearly reduced using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data. Estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining it with a pre-set reliability. The low-dimensional feature data is compared with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

[0006] Preferably, the over-limit threshold is obtained through the following steps: Multi-source data from current transformers were collected as training data. The training data is subjected to outlier detection, and outliers are removed to obtain accurate data. Perform wavelet decomposition on the accurate data to obtain a training set of approximation coefficients; The approximation coefficient training set is subjected to nonlinear dimensionality reduction using the t-distributed random neighbor embedding algorithm to obtain a low-dimensional feature data training set that can distinguish different error states; For a low-dimensional feature data training set under normal operating conditions, a density estimation method combined with a pre-set reliability is used to determine the threshold for exceeding the measurement error limit.

[0007] Preferably, the multi-source data includes real-time error monitoring data from the online monitoring module, transformer attribute data, archive data, calibration data, and operating condition data.

[0008] Preferably, the step of detecting outliers and deleting outliers from the training data to obtain accurate data includes: The training data is outlier removal using a quartile box plot outlier detection algorithm and a DBSCAN clustering algorithm. Based on the constant impedance characteristic of the secondary circuit of the current transformer and combined with the core BH curve, the full-range error value is fitted using the error estimation formula and the secondary circuit impedance calculation formula to expand the training data.

[0009] Preferably, the error estimation formula is: ; Where ε is the error value. This refers to the number of turns in the secondary winding. The length of the magnetic circuit of the iron core. For secondary loop magnetic flux, This refers to the secondary circuit current. The formula for calculating the impedance of the secondary circuit is as follows: ; in, S For the core area, f For operating frequency, k For the core lamination factor, B Z represents the magnetic flux density, and Z represents the impedance of the secondary circuit.

[0010] Preferably, the step of using the t-distributed random neighbor embedding algorithm to perform nonlinear dimensionality reduction processing on the approximation coefficient set to obtain low-dimensional feature data includes: Calculate the probability distribution between any two points in the original high-dimensional space. : ; in, Forx i When centered, x j The probability of it being a neighboring point; x i For the first in the original high-dimensional space i One sample, x j For the first in the original high-dimensional space j One sample, N The number of samples; In a low-dimensional space, under the condition of a t-distribution with 1 degree of freedom, calculate the probability density of the sample. : ; in, Represents the first in low-dimensional space i One sample, Represents the first in low-dimensional space j One sample, Represents the first in low-dimensional space k One sample, Represents the first in low-dimensional space l One sample.

[0011] Preferred options also include: Based on KL divergence, the similarity between the spatial distribution probability P of the high-dimensional space and the spatial distribution probability Q of the low-dimensional space before and after mapping is quantified to obtain the quantization result C. ; In the formula, Let KL divergence be the probability distribution between any two points in the high-dimensional space before and after the mapping. The quantization results were optimized using gradient descent. .

[0012] In a second aspect, the present invention provides an online identification device for metering errors of current transformers, comprising: The filtering module is used to decompose the three-phase current signal on the secondary side of the current transformer using VMD, and then filter out the residual signal after decomposition. The decomposition module is used to perform wavelet decomposition on the residual term signal to obtain an approximate coefficient set. The processing module is used to perform nonlinear dimensionality reduction processing on the approximation coefficient set using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data; The estimation module is used to estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining the preset confidence level. The identification module is used to compare the low-dimensional feature data with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

[0013] In a third aspect, the present invention provides an electronic device including a processor and a memory, the processor being configured to execute a computer program stored in the memory to implement the online identification method for current transformer metering errors.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium storing at least one instruction that, when executed by a processor, implements the online identification method for current transformer metering errors.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention utilizes a combination of quartile box plot outlier detection and the DBSCAN clustering algorithm to eliminate outliers in real-time monitoring data caused by communication interference and sensor drift. Simultaneously, by combining the constant impedance characteristics of the current transformer's secondary circuit with the core's BH curve, it fits the full-range error value using error estimation formulas and impedance calculation models, supplementing missing data under non-rated conditions such as light load and overload. After data cleaning and expansion, the dataset completeness is improved, and the accuracy of outlier removal is high, providing high-quality data support for subsequent error feature extraction and over-limit identification, avoiding identification biases caused by data quality issues.

[0016] This invention utilizes the adaptive frequency band division characteristic of VMD technology to decompose the three-phase current signal on the secondary side of a current transformer into intrinsic mode signals of different frequency bands, filtering out the residual term signal characterizing the measurement error and removing high-frequency noise interference. Then, through the multi-resolution analysis capability of wavelet decomposition, low-frequency approximation coefficients reflecting the overall error trend are extracted from the residual term signal, further compressing redundant information and highlighting error characteristics. The improved signal-to-noise ratio of the extracted error characteristics enables the identification of weak measurement errors, providing clear characteristic basis for subsequent identification.

[0017] This invention employs t-SNE nonlinear dimensionality reduction technology. By minimizing the KL divergence between the high-dimensional and low-dimensional spaces, it maps the high-dimensional approximation coefficients obtained from wavelet decomposition to the low-dimensional space, while preserving the local structure and nonlinear relationships of the data to the greatest extent possible. After t-SNE dimensionality reduction, the data points under normal operating conditions and those under error-exceeding conditions show high separation in the low-dimensional space, intuitively presenting the differences in feature distributions of different error degradation levels. This avoids feature distortion caused by dimensionality reduction and provides highly discriminative low-dimensional feature data for subsequent threshold identification.

[0018] This invention employs a kernel density estimation nonparametric method. Based on low-dimensional feature data under normal operating conditions, it adaptively estimates the probability density distribution and determines the personalized over-limit threshold for each phase current transformer with a 99% high confidence level, replacing the traditional method that relies on empirical values ​​or fixed thresholds. This threshold setting method can dynamically adapt to the error characteristics of current transformers of different models and operating conditions. Using the threshold identification method of this invention, the false alarm rate and false negative rate of metering error over-limit identification are low, and it can capture abnormal states where the ratio difference exceeds 0.2%, providing a reliable criterion for power grid metering anomaly early warning and equipment condition assessment. Attached Figure Description

[0019] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is an outlier detection map of the DBSCAN algorithm in Embodiment 1 of the present invention; Figure 2 This is a BH curve diagram of a certain type of iron core at 10kHz in Embodiment 1 of the present invention; Figure 3 This is a graph showing 100 sets of secondary circuit currents and real-time error values ​​of the monitoring module in Embodiment 1 of the present invention; Figure 4 This is a graph showing the calculation results of 100 sets of data in Embodiment 1 of the present invention; Figure 5 This is the result of the out-of-limit phase feature extraction in Embodiment 1 of the present invention; Figure 6 This is the CT error exceeding the limit monitoring result of Embodiment 1 of the present invention; Figure 7 This is a structural block diagram of the online identification device for current transformer metering error according to Embodiment 2 of the present invention; Figure 8 This is a structural block diagram of an electronic device according to Embodiment 3 of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0021] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this invention is for describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0022] Example 1 A method for online identification of metering errors in current transformers includes: S1. The VMD decomposes the three-phase current signal on the secondary side of the current transformer and then filters out the residual signal. Preferably, multi-source data of the current transformer is collected, including real-time error monitoring data from the online monitoring module, transformer attribute data, archive data, calibration data, and operating condition data. Based on the multi-source data, outlier detection algorithms such as quartile box plots and DBSCAN clustering algorithms are used to remove outliers. Based on the constant impedance characteristic of the secondary circuit of the current transformer and combined with the core BH curve, the full-range error value is fitted using error estimation formulas and secondary circuit impedance calculation formulas to expand the training data.

[0023] Specifically, the multi-source data includes real-time error monitoring data from the online monitoring module, instrument transformer attribute data, archive data, calibration data, and operating condition data.

[0024] Current transformer data comes from numerous sources, including real-time error monitoring data from online monitoring modules, transformer attribute data, archive data, calibration data, and operating condition data. Performance evaluation relies on condition indicator data; therefore, data cleaning is necessary to ensure the system receives clean, complete, accurate, and non-duplicative data for subsequent indicator classification.

[0025] Based on multi-source data from current transformers, we analyze the influencing factors of abnormal data and determine the impact of abnormal data on subsequent processing. Based on the results of data mining and statistical analysis, we formulate cleaning rules for the data in advance. For the one-dimensional data of current transformers in the real-time monitoring data provided by the hardware module, we use data cleaning algorithms and technologies such as outlier detection of quartile box plots and outlier detection of density to improve data quality.

[0026] Outlier detection and removal are performed using quartile box plots. For error values ​​provided by the hardware module, density-based DBSCAN clustering algorithm is used for outlier detection and removal; for multiple error values ​​under a certain current value, quartile anomaly detection is used to remove outliers.

[0027] As a specific example, such as Figure 1 As shown, for a certain current transformer, the set of error data from 96 monitoring points per day is D = {0.129, 0.054, 0.069, 0.021, -0.15, ...}. By sorting the error values ​​from smallest to largest and finding their quartiles, the quartile values ​​and interquartile ranges are calculated. Finally, using 1.5 times the interquartile range as a standard, upper and lower limits for the error data of this current transformer are set, and discrete error values ​​outside the upper and lower limits are eliminated. This method can remove invalid data caused by gross measurement errors and communication anomalies.

[0028] Curve error fitting: Assuming the secondary circuit impedance value after the current transformer is installed The size remains basically unchanged, and the number of turns of the secondary winding is known. and the length of the iron core magnetic circuit fixed, For secondary loop magnetic flux, B Given the magnetic induction intensity, the metering error value of the current transformer is calculated according to formula (1). : (1) By obtaining multiple current transformer error values ​​and corresponding secondary circuit currents Multiple magnetic fluxes were obtained. , … According to the BH curve of the specified core at a specified frequency, such as Figure 2 The corresponding magnetic flux density can be obtained. , … .

[0029] The corresponding secondary circuit impedance can be obtained from the relationship between magnetic induction intensity and secondary circuit impedance. , … : (2) in, For the core area, For operating frequency, This represents the lamination coefficient of the iron core.

[0030] Can target , … The average value is taken as the secondary circuit impedance value, and the value specified in the verification procedure is used. , , 、 Substitute the rated load into formula (2), then obtain the corresponding magnetic flux H according to the BH curve, and finally obtain the transformer error value by fitting the corresponding calibration point according to the error estimation formula.

[0031] Taking a current transformer with an accuracy class of 0.5S and a rated secondary current of 5A as an example, 100 sets of data on the collected secondary circuit current and the real-time error value of the monitoring module are listed as follows: Figure 3 As shown: Assume the number of turns of the secondary winding =1. The length of the magnetic circuit of the iron core is 0.1m. According to formula (1), we get: Calculation results Figure 4 .

[0032] The calculated result Mapped to Figure 4 The magnetic flux density value is obtained from the BH curve. Let the core area be... , , Then we have: ; Given that the magnitude of the secondary impedance is constant, Figure 3 The impedance estimate Z = 0.074Ω is obtained by least squares or direct averaging of the scattered points. The estimated magnetic flux density B is obtained by returning Z to formula (2). Figure 4 Map H, then substitute H into formula (1), and calculate the real-time estimation error or verification estimation error, i.e. the full-range error value of the current, based on the actual secondary circuit current or the current specified in the verification procedure. This error is used to determine the error level of the current transformer under the uncollected load level and serves as the data source for the current transformer. It is then used in the analysis of accuracy indicators or stability indicators to expand the training data.

[0033] VMD decomposition technology is used to adaptively decompose the three-phase current signal on the secondary side of the current transformer data source. The residual term signal is selected based on the frequency characteristics of the intrinsic mode signal after decomposition. The residual term signal characterizes the measurement error characteristics. Specifically, when analyzing the three-phase current signal on the secondary side of a current transformer, in order to separate noise and extract the weak characteristic components caused by measurement errors, the complex current signal is adaptively decomposed into mode functions, thereby obtaining residual terms that characterize the error characteristics, providing a high-quality data foundation for subsequent phase identification of measurement errors exceeding limits.

[0034] VMD (Vibration Mode Decomposition) technology uses adaptive frequency band division to decompose complex signals into intrinsic mode signals of different frequency bands, extracting information from the complex signal based on the frequency magnitude. The entire decomposition process mainly includes multiple steps such as frequency domain separation, variational optimization, and adaptive decomposition. VMD decomposition technology is used to decompose the phase current signals of the CT secondary side, and then the residual signals are selected based on the changing frequencies of the decomposed intrinsic mode signals. The residual signals extracted through VMD decomposition can remove interference factors and highlight key changing trends, thus providing more reliable data for subsequent metrological status judgment.

[0035] ;

[0036] in, I This contains the raw data of the phase currents on the secondary side of the CT. , The frequencies corresponding to the modal signals of the residual and noise terms are... For residual term signals, This is the noise signal.

[0037] S2. Perform wavelet decomposition on the residual term signal to obtain the approximate coefficient set; the approximate coefficient set is a low-frequency characteristic component representing the measurement error.

[0038] The residual terms of the three-phase voltage signal on the secondary side of the CT are processed using wavelet decomposition to obtain an approximate coefficient set. The wavelet decomposition formula is: ; ; In the formula, For time x The signal of the changing residual term, i.e. ; The wavelet basis is scaled and translated; These are scaling parameters used to control the wavelet scale; b ψ is the translation parameter used to control the wavelet position, and ψ is a fixed basic wavelet function.

[0039] S3. The approximate coefficient set is subjected to nonlinear dimensionality reduction processing using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data; S4. Estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining the preset reliability. The threshold value is obtained through the following steps: Multi-source data from current transformers were collected as training data. The training data is subjected to outlier detection, and outliers are removed to obtain accurate data. Perform wavelet decomposition on the accurate data to obtain a training set of approximation coefficients; The approximation coefficient training set is subjected to nonlinear dimensionality reduction using the t-distributed random neighbor embedding algorithm to obtain a low-dimensional feature data training set that can distinguish different error states; For a low-dimensional feature data training set under normal operating conditions, a density estimation method combined with a pre-set reliability is used to determine the threshold for exceeding the measurement error limit.

[0040] Specifically, the steps for using the t-distributed random neighbor embedding algorithm to perform nonlinear dimensionality reduction on the approximation coefficient training set to obtain a low-dimensional feature data training set that can distinguish different error states are as follows: Calculate the probability distribution between any two points in the original high-dimensional space. ;

[0041] in, For x i When centered, x j The probability of it being a neighboring point; x i For the first in the original high-dimensional space i One sample, x j For the first in the original high-dimensional space j One sample, N The number of samples; In low-dimensional space, with a degree of freedom of 1 t Calculate the probability density of the sample under the given distribution. : ; in, Represents the first in low-dimensional space i One sample, Represents the first in low-dimensional space j One sample, Represents the first in low-dimensional space k One sample, Represents the first in low-dimensional space l One sample.

[0042] Based on KL divergence, the similarity between the spatial distribution probability P of the high-dimensional space and the spatial distribution probability Q of the low-dimensional space before and after mapping is quantified to obtain the quantization result C. ; In the formula, Let KL divergence be the probability distribution between any two points in the high-dimensional space before and after the mapping. The quantization result C from the previous step is optimized using gradient descent.

[0043] .

[0044] Preferably, the kernel density estimation method is used to estimate the probability density distribution of low-dimensional feature data under normal operating conditions, and the measurement error exceeding the limit is determined by combining the pre-set reliability.

[0045] Specifically, a Gaussian kernel function is used to smooth each sample in the low-dimensional space, and the weight is determined by the distance between it and the remaining sample points. The probability density distribution of the overall data is constructed based on the kernel function and the determined weights for each sample. The confidence level is then used to further determine the distribution characteristics of the data, determine the threshold, and then establish anomaly identification criteria.

[0046] ; ; Where n is the number of samples The number of For the estimated point, h For smoothing parameters K For Gaussian kernel function, p Let be the probability density function. F Let be the probability distribution function. For confidence level, This is the final threshold value.

[0047] The VMD signal is used to filter the three-phase currents of phases A, B, and C to extract the residual term signal. Wavelet decomposition is used to process the residual term signals corresponding to different phases and different measurement error levels to obtain multiple sets of high-dimensional approximation coefficients. The t-SNE algorithm is used to perform nonlinear dimensionality reduction on the above multi-dimensional approximation coefficient sets, mapping them to a one-dimensional space, thereby obtaining low-dimensional feature data representing different error levels of each phase, preserving the nonlinear structure of the data, such as... Figure 5 As shown.

[0048] Experimental results demonstrate that the proposed dual decomposition scheme based on VMD and wavelet decomposition can retain information in the metrological errors of each CT phase to the greatest extent while eliminating redundant information. VMD can eliminate high-frequency noise signals by means of bandwidth control, frequency adaptation, and variational optimization, and based on the characteristics of degraded signals. Multiresolution analysis in wavelet decomposition compresses the information of the decomposed residuals, further eliminating redundant information. In addition, t-SNE maps the high-dimensional approximation coefficient set to a one-dimensional space, retains the data nonlinearity, and clearly presents the low-dimensional characteristic data of each CT phase, i.e., the error characteristics.

[0049] Based on the extraction of degradation features during CT deterioration, a kernel density estimation method is applied to establish a threshold for identifying excessive CT errors. To ensure the reliability of the final threshold, it needs to be determined based on data under normal operating conditions. This invention uses normal data with measurement errors within 0.2% to determine the excessive CT error index. With a confidence level set to 99%, the calculated low-dimensional feature data corresponding to the three phases of the secondary side of the CT scan are 1.6818467698208412, -0.7405597525400477, and -2.0088016164450466, respectively.

[0050] S5. Compare the low-dimensional feature data with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

[0051] The final CT error exceeding the limit identification result is as follows: Figure 6 As shown in the figure. The experimental results clearly show that the CT error over-limit identification criterion proposed in this invention can effectively monitor working conditions where the ratio difference is outside 0.2%, meeting the 0.2 level monitoring standard.

[0052] Example 2 like Figure 7 As shown, based on the same inventive concept as the above embodiments, the present invention also provides an online identification device for current transformer metering errors, comprising: The filtering module is used to decompose the three-phase current signal on the secondary side of the current transformer using VMD, and then filter out the residual signal after decomposition. The decomposition module is used to perform wavelet decomposition on the residual term signal to obtain an approximate coefficient set. The processing module is used to perform nonlinear dimensionality reduction processing on the approximation coefficient set using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data; The estimation module is used to estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining the preset confidence level. The identification module is used to compare the low-dimensional feature data with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

[0053] Example 3 like Figure 8 As shown, the present invention also provides an electronic device 100 for implementing an online identification method for current transformer metering errors; The electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104.

[0054] The memory 101 can be used to store the computer program 103. The processor 102 implements the steps of the online identification method for current transformer metering error in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101.

[0055] The memory 101 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0056] At least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 102 may be a microprocessor or any conventional processor. Processor 102 is the control center of electronic device 100, connecting various parts of electronic device 100 via various interfaces and lines.

[0057] The memory 101 in the electronic device 100 stores multiple instructions to implement an online identification method for current transformer metering errors, and the processor 102 can execute multiple instructions to achieve the following: VMD is used to decompose the three-phase current signal on the secondary side of the current transformer, and the residual signal is then screened out after decomposition. Perform wavelet decomposition on the residual term signal to obtain the approximate coefficient set; The approximation coefficient set is nonlinearly reduced using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data. Estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining it with a pre-set reliability. The low-dimensional feature data is compared with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

[0058] Example 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0059] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0061] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0062] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0063] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0064] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for online identification of metering errors in current transformers, characterized in that, include: VMD is used to decompose the three-phase current signal on the secondary side of the current transformer, and the residual signal is then screened out after decomposition. Perform wavelet decomposition on the residual term signal to obtain the approximate coefficient set; The approximation coefficient set is nonlinearly reduced using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data. Estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining it with a pre-set reliability. The low-dimensional feature data is compared with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

2. The online identification method for current transformer metering errors as described in claim 1, characterized in that, The threshold value is obtained through the following steps: Multi-source data from current transformers were collected as training data. The training data is subjected to outlier detection, and outliers are removed to obtain accurate data. Perform wavelet decomposition on the accurate data to obtain a training set of approximation coefficients; The approximation coefficient training set is subjected to nonlinear dimensionality reduction using the t-distributed random neighbor embedding algorithm to obtain a low-dimensional feature data training set that can distinguish different error states; For a low-dimensional feature data training set under normal operating conditions, a density estimation method combined with a pre-set reliability is used to determine the threshold for exceeding the measurement error limit.

3. The online identification method for current transformer metering errors as described in claim 2, characterized in that, The multi-source data includes real-time error monitoring data from the online monitoring module, instrument transformer attribute data, archive data, calibration data, and operating condition data.

4. The online identification method for current transformer metering errors as described in claim 2, characterized in that, The steps of performing outlier detection and deletion on the training data to obtain accurate data include: The training data is outlier removal using a quartile box plot outlier detection algorithm and a DBSCAN clustering algorithm. Based on the constant impedance characteristic of the secondary circuit of the current transformer and combined with the core BH curve, the full-range error value is fitted using the error estimation formula and the secondary circuit impedance calculation formula to expand the training data.

5. The online identification method for current transformer metering errors as described in claim 4, characterized in that... The error estimation formula is as follows: ; Where ε is the error value. This refers to the number of turns in the secondary winding. The length of the magnetic circuit of the iron core. For secondary loop magnetic flux, This refers to the secondary circuit current. The formula for calculating the impedance of the secondary circuit is as follows: ; in, S For the core area, f For operating frequency, k For the core lamination factor, B Z represents the magnetic flux density, and Z represents the impedance of the secondary circuit.

6. The online identification method for metering errors of current transformers as described in claim 1, characterized in that, The step of using the t-distributed random neighbor embedding algorithm to perform nonlinear dimensionality reduction processing on the approximation coefficient set to obtain low-dimensional feature data includes: Calculate the probability distribution between any two points in the original high-dimensional space. : ； in, For x i When centered, x j The probability of it being a neighboring point; x i For the first in the original high-dimensional space i One sample, x j For the first in the original high-dimensional space j One sample, N The number of samples; In a low-dimensional space, under the condition of a t-distribution with 1 degree of freedom, calculate the probability density of the sample. : ； in, Represents the first in low-dimensional space i One sample, Represents the first in low-dimensional space j One sample, Represents the first in low-dimensional space k One sample, Represents the first in low-dimensional space l One sample.

7. The online identification method for current transformer metering errors as described in claim 6, characterized in that, Also includes: Based on KL divergence, the similarity between the spatial distribution probability P of the high-dimensional space and the spatial distribution probability Q of the low-dimensional space before and after mapping is quantified to obtain the quantization result C. ； In the formula, Let KL divergence be the probability distribution between any two points in the high-dimensional space before and after the mapping. The quantization results were optimized using gradient descent. 。 8. An online identification device for metering errors of current transformers, characterized in that, include: The filtering module is used to decompose the three-phase current signal on the secondary side of the current transformer using VMD, and then filter out the residual signal after decomposition. The decomposition module is used to perform wavelet decomposition on the residual term signal to obtain an approximate coefficient set. The processing module is used to perform nonlinear dimensionality reduction processing on the approximation coefficient set using the t-distributed random neighbor embedding algorithm to obtain low-dimensional feature data; The estimation module is used to estimate the probability density distribution of low-dimensional feature data and determine the threshold for exceeding the measurement error limit by combining the preset confidence level. The identification module is used to compare the low-dimensional feature data with the over-limit threshold. If the low-dimensional feature data is less than the over-limit threshold, it indicates that the current transformer is in normal operation. Otherwise, it is considered that the current transformer may be in abnormal operation.

9. An electronic device, characterized in that, It includes a processor and a memory, the processor being used to execute a computer program stored in the memory to implement the online identification method for current transformer metering errors as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, which, when executed by a processor, implements the online identification method for current transformer metering errors as described in any one of claims 1 to 7.

Images