A mutual inductor error self-adaptive correction method and system

Abstract

The application discloses a kind of mutual inductor error self-adapting correction method and system, method includes: constructing group coordination feature and calculating real-time deviation to detect system anomaly;When finding abnormality, further calculate each mutual inductor individual contribution degree to realize accurate fault location;For the abnormal mutual inductor positioned, extract its residual error feature and fuse with macroscopic deviation, identify its abnormal type as fixed deviation or random error mutation by classification algorithm;According to the type of abnormality, dynamically determine the optimal prediction step using machine learning model, and accordingly self-adaptively select long short-term memory network or its hybrid model combined with attention mechanism for error prediction;Finally, the error value obtained by prediction is used to compensate the real-time measurement data online.The whole-process closed-loop management from anomaly detection, intelligent diagnosis, adaptive prediction to active compensation is realized, which significantly improves the accuracy of mutual inductor online monitoring, fault identification capability and long-term operation measurement precision.

Description

Technical Field

[0001] This invention belongs to the field of current transformer analysis technology, and particularly relates to a method and system for adaptive correction of current transformer errors. Background Technology

[0002] Voltage and current transformers are core basic measuring devices in power systems for energy metering, relay protection, and operation monitoring. Their long-term measurement accuracy directly affects the economy, safety, and reliability of the power grid. In actual operation, the measurement error of transformers is affected by a combination of factors, including ambient temperature, mechanical vibration, electromagnetic interference, and the aging of internal key components (such as optical devices and electronic components). This causes their output to deviate from the true value, and the error characteristics may exhibit slow-changing systematic deviations or unpredictable sudden jumps.

[0003] Currently, the management and response to transformer errors mainly rely on the following methods, but all of them have obvious limitations:

[0004] First, the mainstream method still relies on periodic offline calibration. This method requires taking the current transformer under test out of operation and comparing it with a higher-level standard current transformer in the laboratory or in the field. This approach is not only time-consuming and costly, but it also fails to reflect the true error state of the equipment under actual operating conditions, especially under dynamic changes, resulting in monitoring blind spots and making it difficult to provide early warning of error degradation trends.

[0005] Secondly, while some online monitoring technologies have been applied, they are mostly focused on simple threshold judgments of the output values ​​of individual instrument transformers. These methods fail to fully utilize the collective coordination information provided by multiple redundant instrument transformers at the same measurement point, making it difficult to effectively distinguish between normal load fluctuations on the primary side of the power grid and anomalies caused by faults in the instrument transformers themselves. This results in high false alarm and false negative rates, and insufficient accuracy and reliability in diagnosis. Summary of the Invention

[0006] This invention provides a method and system for adaptive correction of transformer errors, which solves the technical problem of difficulty in effectively distinguishing between normal load fluctuations on the primary side of the power grid and anomalies caused by transformer faults, resulting in high false alarm and missed alarm rates.

[0007] In a first aspect, the present invention provides a method for adaptive correction of transformer error, comprising:

[0008] Acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster contains at least one current transformer, and the real-time measurement data set contains a real-time measurement data set corresponding to the at least one current transformer.

[0009] Based on the pre-constructed group characteristic parameters, calculate the real-time deviation corresponding to the real-time measurement data set, and determine whether the real-time deviation is greater than a preset deviation threshold.

[0010] If the deviation exceeds a preset threshold, at least one abnormal current transformer is determined based on the individual contribution of each current transformer in the current transformer cluster.

[0011] Extract the residual features of the at least one abnormal current transformer, and fuse each residual feature with the real-time deviation to obtain at least one target deviation feature. Then, determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes fixed deviation type and random error mutation type.

[0012] Determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic;

[0013] Based on the type of anomaly, an error prediction model corresponding to the current transformer with the anomaly is selected using a preset model selection rule, and an error prediction is performed based on the optimal prediction step size and the error prediction model to obtain a prediction error value.

[0014] The real-time measurement data is compensated based on a certain prediction error value to obtain the final corrected target real-time measurement data, wherein the real-time measurement data is the real-time measurement data of the abnormal current transformer.

[0015] In a second aspect, the present invention provides an adaptive error correction system for a current transformer, comprising:

[0016] The acquisition module is configured to acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster includes at least one current transformer, and the real-time measurement data set includes a real-time measurement data set corresponding to the at least one current transformer.

[0017] The judgment module is configured to calculate the real-time deviation corresponding to the real-time measurement data set based on the pre-constructed group characteristic parameters, and to determine whether the real-time deviation is greater than a preset deviation threshold.

[0018] The first determining module is configured to determine at least one abnormal current transformer based on the individual contribution of each current transformer in the current transformer cluster if the deviation exceeds a preset threshold.

[0019] The fusion module is configured to extract the residual features of the at least one abnormal current transformer, fuse each residual feature with the real-time deviation to obtain at least one target deviation feature, and determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes a fixed deviation type and a random error mutation type.

[0020] The second determination module is configured to determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic.

[0021] The prediction module is configured to select an error prediction model corresponding to the abnormal current transformer according to a preset model selection rule based on the abnormality type, and perform error prediction based on an optimal prediction step size and the error prediction model to obtain a prediction error value.

[0022] The compensation module is configured to compensate for a certain real-time measurement data based on a certain prediction error value to obtain a final corrected target real-time measurement data, wherein the certain real-time measurement data is the real-time measurement data of a certain abnormal current transformer.

[0023] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the current transformer error adaptive correction method according to any embodiment of the present invention.

[0024] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the mutual inductor error adaptive correction method according to any embodiment of the present invention.

[0025] The adaptive error correction method and system for current transformers disclosed in this application firstly utilizes the collective collaborative features constructed from historical data of multiple current transformers and real-time deviation calculations to achieve online high-sensitivity anomaly detection without the need for standard current transformers or power outages, and can accurately locate specific faulty equipment, greatly improving operation and maintenance efficiency. Secondly, by fusing the microscopic residual features of the abnormal equipment with the macroscopic deviation features of the system, and combining them with an intelligent classification algorithm, it achieves accurate identification of two anomaly modes: "fixed deviation" and "random error mutation," while also associating key component status data to provide a basis for fault root cause analysis. Thirdly, a machine learning model is introduced to dynamically optimize the future prediction time based on the identified anomaly type, enabling the prediction strategy to adaptively match different error change rhythms, solving the problem of poor adaptability of fixed prediction modes. For sudden random errors, a prediction model with an attention fusion mechanism is adopted, and real-time physical status data of key components are added for joint prediction, which not only improves the prediction accuracy under complex disturbances but also enhances the physical interpretability of the model prediction results. Most importantly, this invention directly uses the predicted error value to compensate for real-time measurement data, forming a closed loop from "detecting anomalies" to "active correction," directly translating the algorithm's value into real-time improvements in measurement accuracy. Finally, through a built-in model performance verification and update mechanism, the system ensures long-term tracking of equipment status changes, maintaining excellent robustness and self-maintenance capabilities, and providing continuous, reliable, and high-precision measurement assurance for the power system. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart of an adaptive error correction method for a mutual inductor provided in an embodiment of the present invention;

[0028] Figure 2 This is a structural block diagram of an adaptive error correction system for a mutual inductor provided in an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0031] Please see Figure 1 The diagram shows a flowchart of an adaptive error correction method for mutual inductors according to this application.

[0032] like Figure 1 As shown, the adaptive error correction method for mutual inductors specifically includes the following steps:

[0033] Step S101: Obtain the real-time measurement data set of the current transformer cluster, wherein the current transformer cluster includes at least one current transformer, and the real-time measurement data set includes the real-time measurement data set corresponding to the at least one current transformer.

[0034] Step S102: Calculate the real-time deviation corresponding to the real-time measurement data set based on the pre-constructed group characteristic parameters, and determine whether the real-time deviation is greater than a preset deviation threshold.

[0035] In this step, the group feature parameters include the group centroid vector and the group association covariance matrix;

[0036] The group centroid vector and the group correlation covariance matrix are calculated based on the historical normal measurement data of the current transformer cluster, wherein:

[0037] Calculate the i-th element in the population centroid vector. The expression is:

[0038] ,

[0039] In the formula, Let be the i-th element in the group center value vector, representing the mean value of the measurements taken by the i-th fiber optic current transformer. This represents the total number of sampling points for historical normal data. The historical normal measurement value of the i-th current transformer at the t-th sampling time;

[0040] Calculate the element in the i-th row and j-th column of the group association covariance matrix. The expression is:

[0041] ,

[0042] In the formula, Let be the element in the i-th row and j-th column of the group correlation covariance matrix, representing the covariance between the measured values ​​of the i-th current transformer and the j-th current transformer. Let j be the j-th element in the group center value vector, representing the mean value of the measurements taken by the j-th fiber optic current transformer. This represents the historical normal measurement value of the j-th current transformer at the t-th sampling time.

[0043] Furthermore, the expression for calculating the real-time deviation is:

[0044] ,

[0045] In the formula, Let be the deviation at the t-th sampling time. For the measurement data at the t-th sampling time, The group centrality value, This is the transpose symbol.

[0046] Step S103: If the deviation is greater than the preset deviation threshold, at least one abnormal current transformer is determined based on the individual contribution of each current transformer in the current transformer cluster.

[0047] In this step, it is determined whether the individual contribution of a certain current transformer is greater than a preset contribution threshold. The expression for calculating the individual contribution is as follows:

[0048] ,

[0049] In the formula, Let be the individual contribution of the i-th current transformer to the deviation at the t-th sampling time. For the measurement data at the t-th sampling time, The group centrality value, Let be the deviation between the real-time measured value of the i-th current transformer and the group center value vector. Historical covariance matrix The element in the i-th row and z-th column of the inverse matrix, Let be the deviation between the real-time measured value of the z-th current transformer and the group center value vector;

[0050] The expression for the preset contribution threshold is:

[0051] ,

[0052] In the formula, To preset the contribution threshold, To adjust the coefficient, This represents the total number of current transformers in the current transformer cluster.

[0053] If the contribution is not greater than the preset contribution threshold, then the current transformer is defined as an abnormal current transformer; otherwise, the current transformer is defined as a normal current transformer.

[0054] Step S104: Extract the residual features of the at least one abnormal current transformer, and fuse each residual feature with the real-time deviation to obtain at least one target deviation feature. Determine the abnormal type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormal type includes fixed deviation type and random error mutation type.

[0055] In this step, based on principal component analysis, the real-time measurement data of the abnormal current transformer is projected onto the residual space determined by historical normal data to obtain the residual vector.

[0056] At least one statistical feature is extracted from the residual vector as a residual feature. The statistical features include standard deviation, trend slope, and fluctuation period.

[0057] Each residual feature is fused with the real-time deviation to obtain at least one target deviation feature. The K-nearest neighbor classification algorithm is used to compare the target deviation feature corresponding to each abnormal current transformer with a preset training sample set to classify its abnormality type as fixed deviation type or random error mutation type.

[0058] Step S105: Determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic.

[0059] In this step, the specific anomaly type and the specific target deviation feature are input into the XGBoost regression model;

[0060] The XGBoost regression model outputs an optimal prediction step size that adapts to the current error change trend. When the anomaly type is a fixed deviation type, the output optimal prediction step size is constrained to a first value range. When the anomaly type is a random error mutation type, the output optimal prediction step size is constrained to a second value range that is smaller than the first value range.

[0061] Step S106: Based on the type of anomaly, select an error prediction model corresponding to the current transformer with a preset model selection rule, and perform error prediction based on the optimal prediction step size and the error prediction model to obtain a prediction error value.

[0062] In this step, if an anomaly type is a fixed bias type, a long short-term memory network model is selected as the error prediction model; if an anomaly type is a random error mutation type, a hybrid model combining attention mechanism and long short-term memory network is selected as the error prediction model.

[0063] A real-time measurement data of a certain abnormal current transformer is projected onto the residual space determined by the historical normal measurement data of the same abnormal current transformer based on principal component analysis, resulting in a residual vector sequence arranged in chronological order, which serves as the historical error sequence.

[0064] Using a certain optimal prediction step size as the time length of the prediction target, the historical error sequence of a certain abnormal current transformer is input into the error prediction model;

[0065] The error prediction model outputs an error prediction sequence for the future time period, and the predicted values ​​corresponding to the current and future compensation times in the error prediction sequence are used as the prediction error values.

[0066] When a hybrid model combining attention mechanism and long short-term memory network is selected, error prediction based on the error prediction model also includes: inputting the time series data of the key component status of a certain abnormal current transformer collected synchronously into the hybrid model; the hybrid model dynamically weights and fuses the historical error sequence and the time series data of the key component status through the attention mechanism, and then processes it through the long short-term memory network to output the error prediction sequence for the future time period.

[0067] Key component status timing data includes LED drive current and cooler temperature.

[0068] Step S107: Compensate a real-time measurement data according to a certain prediction error value to obtain a final corrected target real-time measurement data, wherein the real-time measurement data is the real-time measurement data of the certain abnormal current transformer.

[0069] In summary, the method of this application firstly utilizes the group collaborative features constructed from historical data of multiple instrument transformers and real-time deviation calculations to achieve online high-sensitivity anomaly detection without the need for standard instrument transformers or power outages, and can accurately locate specific faulty equipment, greatly improving operation and maintenance efficiency. Secondly, by fusing the microscopic residual features of the abnormal equipment with the macroscopic deviation features of the system, and combining them with intelligent classification algorithms, it achieves accurate identification of two anomaly modes: "fixed deviation" and "random error mutation." Simultaneously, it correlates key component status data, providing a basis for fault root cause analysis. Thirdly, it introduces a machine learning model to dynamically optimize the future prediction time based on the identified anomaly type, enabling the prediction strategy to adaptively match different error change rhythms, solving the problem of poor adaptability of fixed prediction modes. For sudden random errors, it adopts a prediction model that integrates an attention mechanism and incorporates real-time physical state data of key components for joint prediction, which not only improves prediction accuracy under complex disturbances but also enhances the physical interpretability of the model prediction results. Most importantly, this invention directly uses the predicted error value to compensate for real-time measurement data, forming a closed loop from "detecting anomalies" to "active correction," directly translating the algorithm's value into real-time improvements in measurement accuracy. Finally, through a built-in model performance verification and update mechanism, the system ensures long-term tracking of equipment status changes, maintaining excellent robustness and self-maintenance capabilities, and providing continuous, reliable, and high-precision measurement assurance for the power system.

[0070] In one specific embodiment, the mutual inductor error adaptive correction method specifically includes the following steps:

[0071] S1: Data Acquisition, Preprocessing, and Construction of Normal Population Patterns

[0072] Data Acquisition: Real-time measurement datasets from n current transformers are collected synchronously. , Let be the measurement data of the nth current transformer at the t-th sampling time. The transpose symbol represents the state characteristics of key components of the transformer (such as LED current, cooler temperature, etc.).

[0073] Group Normal Pattern Construction: Based on historical normal datasets, statistical methods are used to calculate the group centroid vector and the group correlation covariance matrix. The group correlation covariance matrix reflects the interrelationship and coordination of n current transformers under normal operating conditions.

[0074] Residual feature extraction: Principal component analysis (PCA) is used to establish a normal operating mode model on historical normal datasets. Residual matrix. Obtained by projection onto the residual space, where... To standardize data vectors, Principal element load matrix, This is the identity matrix. The residual matrix concentrates the subtle information in the original data that does not match the normal pattern of the population, i.e., error information.

[0075] S2: Online detection of measurement errors and precise location of abnormal equipment (group collaborative detection)

[0076] Real-time error detection (calibration-free): Utilizing a real-time measurement dataset, calculate its real-time deviation from the population center and compare it with the error control limit obtained based on historical data statistics. Perform real-time comparisons. , This represents the statistical mean of historical deviations. This represents the standard deviation of historical deviation. If the real-time deviation > If the error is detected, the error state is determined to be abnormal, triggering the fault diagnosis and prediction process. This detection process normalizes the group correlation effect and can effectively distinguish between normal fluctuations in primary current and equipment errors.

[0077] Anomaly location: If the state is abnormal, calculate the individual contribution of the i-th current transformer to quantify the contribution of the device to the total real-time deviation.

[0078] S3: Enhanced residual feature extraction and anomaly type classification (fault diagnosis)

[0079] Enhanced Feature Set Construction: Constructing a multidimensional feature vector set includes:

[0080] PCA residual characteristics: Extracting the standard deviation of the residuals from the residual matrix. Trend slope These reflect the microscopic fluctuations of individual devices.

[0081] ,

[0082] ,

[0083] in, For residual samples, The mean of the residuals, For time series data, the corresponding time. The target phase residual sequence. The residual fluctuation period needs to be calculated by first performing a Fourier transform on the residual samples and then calculating the period corresponding to the peak frequency.

[0084] ,

[0085] ,

[0086] in, The frequency components are obtained after performing a Fourier transform on the residual samples. The sampling time interval, For the largest The corresponding frequency, It is the load matrix of the principal element space.

[0087] Group deviation characteristics: such as real-time deviation relative to The proportion reflects the degree of macroscopic anomaly in the group.

[0088] KNN classification and diagnosis: The K-Nearest Neighbors (KNN) classification algorithm is used to train and classify the enhanced feature set to accurately identify the anomaly type as a fixed bias (such as drift caused by gradual temperature changes) or a sudden random error (such as instantaneous electromagnetic interference or sudden component failure). Before classification, the features need to be standardized to ensure that the samples to be classified and the training set are standardized using the same standard to avoid data leakage.

[0089] Faulty component location: Based on the anomaly type and synchronously collected key component status parameters, the faulty component is located. Fixed deviation: If the LED current in the key component status parameters shows a long-term stable downward trend, the fault is located as aging and decay of the light source (LED). Random error mutation: If the cooler temperature or power supply ripple in the key component status parameters fluctuates drastically, the fault is located as a cooler fault or a momentary power supply interference.

[0090] S4: Adaptive Prediction Step Size Optimization (Policy Decision) Based on XGBoost

[0091] Based on the classification results of step S3, the Extreme Gradient Boosting (XGBoost) regression model is adopted to minimize the prediction error (such as MAE or RMSE) and output the optimal prediction step size that adapts to the current error change trend.

[0092] XGBoost inputs include the enhanced feature set and KNN classification results.

[0093] In fixed-bias scenarios, XGBoost outputs a longer prediction step size to balance computational efficiency.

[0094] Scenarios with sudden random errors: XGBoost outputs a shorter prediction step size (e.g., Lopt) [0.5, 3] minutes), to ensure the ability to capture mutation details.

[0095] S5: Dynamic Model Selection and Error Prediction (High-Precision Prediction)

[0096] Based on the optimal prediction step size and anomaly type, dynamically select and construct the prediction model:

[0097] Fixed bias type: Select a multilayer long short-term memory network (LSTM) model, input historical error sequences, and use its excellent ability to capture long-term time dependence for prediction.

[0098] Random error mutation type: Select the Attention-LSTM hybrid model and implement deep feature fusion. The model dynamically weights and fuses the state features of key components of the transformer (such as LED current, cooler temperature, etc.) with temporal features through an attention mechanism and then makes predictions, which greatly improves the prediction accuracy and physical correlation of the model in nonlinear, high-noise mutation scenarios.

[0099] Please see Figure 2 The diagram shows a structural block diagram of an adaptive error correction system for a current transformer according to this application.

[0100] like Figure 2 As shown, the current transformer error adaptive correction system 200 includes an acquisition module 210, a judgment module 220, a first determination module 230, a fusion module 240, a second determination module 250, a prediction module 260, and a compensation module 270.

[0101] The acquisition module 210 is configured to acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster contains at least one current transformer, and the real-time measurement data set contains a real-time measurement data set corresponding to the at least one current transformer; the judgment module 220 is configured to calculate the real-time deviation corresponding to the real-time measurement data set based on pre-constructed group characteristic parameters, and determine whether the real-time deviation is greater than a preset deviation threshold; the first determination module 230 is configured to determine at least one abnormal current transformer based on the individual contribution of each current transformer in the current transformer cluster if the deviation is greater than the preset deviation threshold; the fusion module 240 is configured to extract the residual features of the at least one abnormal current transformer, fuse each residual feature with the real-time deviation to obtain at least one target deviation feature, and then, based on... The at least one target deviation feature determines the anomaly type of the at least one abnormal current transformer, the anomaly type including a fixed deviation type and a random error mutation type; the second determining module 250 is configured to determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation feature; the prediction module 260 is configured to select an error prediction model corresponding to the certain abnormal current transformer according to the certain anomaly type using a preset model selection rule, and perform error prediction based on the certain optimal prediction step size and the error prediction model to obtain a certain prediction error value; the compensation module 270 is configured to compensate a certain real-time measurement data based on the certain prediction error value to obtain a final corrected target real-time measurement data, wherein the certain real-time measurement data is the real-time measurement data of the certain abnormal current transformer.

[0102] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0103] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the mutual inductor error adaptive correction method in any of the above method embodiments.

[0104] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0105] Acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster contains at least one current transformer, and the real-time measurement data set contains a real-time measurement data set corresponding to the at least one current transformer.

[0106] Based on the pre-constructed group characteristic parameters, calculate the real-time deviation corresponding to the real-time measurement data set, and determine whether the real-time deviation is greater than a preset deviation threshold.

[0107] If the deviation exceeds a preset threshold, at least one abnormal current transformer is determined based on the individual contribution of each current transformer in the current transformer cluster.

[0108] Extract the residual features of the at least one abnormal current transformer, and fuse each residual feature with the real-time deviation to obtain at least one target deviation feature. Then, determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes fixed deviation type and random error mutation type.

[0109] Determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic;

[0110] Based on the type of anomaly, an error prediction model corresponding to the current transformer with the anomaly is selected using a preset model selection rule, and an error prediction is performed based on the optimal prediction step size and the error prediction model to obtain a prediction error value.

[0111] The real-time measurement data is compensated based on a certain prediction error value to obtain the final corrected target real-time measurement data, wherein the real-time measurement data is the real-time measurement data of the abnormal current transformer.

[0112] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the current transformer error adaptive correction system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, which can be connected to the current transformer error adaptive correction system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0113] Figure 3This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the adaptive correction method for transformer errors described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the adaptive correction system for transformer errors. The output device 340 may include a display screen or other display device.

[0114] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0115] In one implementation, the above-described electronic device is applied in a current transformer error adaptive correction system for a client, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0116] Acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster contains at least one current transformer, and the real-time measurement data set contains a real-time measurement data set corresponding to the at least one current transformer.

[0117] Based on the pre-constructed group characteristic parameters, calculate the real-time deviation corresponding to the real-time measurement data set, and determine whether the real-time deviation is greater than a preset deviation threshold.

[0118] If the deviation exceeds a preset threshold, at least one abnormal current transformer is determined based on the individual contribution of each current transformer in the current transformer cluster.

[0119] Extract the residual features of the at least one abnormal current transformer, and fuse each residual feature with the real-time deviation to obtain at least one target deviation feature. Then, determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes fixed deviation type and random error mutation type.

[0120] Determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic;

[0121] Based on the type of anomaly, an error prediction model corresponding to the current transformer with the anomaly is selected using a preset model selection rule, and an error prediction is performed based on the optimal prediction step size and the error prediction model to obtain a prediction error value.

[0122] The real-time measurement data is compensated based on a certain prediction error value to obtain the final corrected target real-time measurement data, wherein the real-time measurement data is the real-time measurement data of the abnormal current transformer.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0124] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for adaptive correction of transformer error, characterized in that, include: Acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster contains at least one current transformer, and the real-time measurement data set contains a real-time measurement data set corresponding to the at least one current transformer. Based on the pre-constructed group characteristic parameters, calculate the real-time deviation corresponding to the real-time measurement data set, and determine whether the real-time deviation is greater than a preset deviation threshold. If the deviation exceeds a preset threshold, at least one abnormal current transformer is determined based on the individual contribution of each current transformer in the current transformer cluster. This includes: determining whether the individual contribution of a certain current transformer exceeds a preset contribution threshold, wherein the expression for calculating the individual contribution is: ， In the formula, Let be the individual contribution of the i-th current transformer to the deviation at the t-th sampling time. For the measurement data at the t-th sampling time, The group centrality value, Let be the deviation between the real-time measured value of the i-th current transformer and the group center value vector. Historical covariance matrix The element in the i-th row and z-th column of the inverse matrix, Let be the deviation between the real-time measured value of the z-th current transformer and the group center value vector; The expression for the preset contribution threshold is: ， In the formula, To preset the contribution threshold, To adjust the coefficient, This represents the total number of current transformers in the current transformer cluster. If the contribution is not greater than the preset contribution threshold, then the current transformer is defined as an abnormal current transformer; otherwise, the current transformer is defined as a normal current transformer. Extract the residual features of the at least one abnormal current transformer, and fuse each residual feature with the real-time deviation to obtain at least one target deviation feature. Then, determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes fixed deviation type and random error mutation type. Determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic; Based on the type of anomaly, an error prediction model corresponding to the current transformer with the anomaly is selected using a preset model selection rule, and an error prediction is performed based on the optimal prediction step size and the error prediction model to obtain a prediction error value. The real-time measurement data is compensated based on a certain prediction error value to obtain the final corrected target real-time measurement data, wherein the real-time measurement data is the real-time measurement data of the abnormal current transformer.

2. The adaptive error correction method for a current transformer according to claim 1, characterized in that, The group feature parameters include the group centroid vector and the group association covariance matrix; The group centroid vector and the group correlation covariance matrix are calculated based on the historical normal measurement data of the current transformer cluster, wherein: Calculate the i-th element in the population centroid vector. The expression is: ， In the formula, Let be the i-th element in the group center value vector, representing the mean value of the measurements taken by the i-th fiber optic current transformer. This represents the total number of sampling points for historical normal data. The historical normal measurement value of the i-th current transformer at the t-th sampling time; Calculate the element in the i-th row and j-th column of the group association covariance matrix. The expression is: ， In the formula, Let be the element in the i-th row and j-th column of the group correlation covariance matrix, representing the covariance between the measured values ​​of the i-th current transformer and the j-th current transformer. Let j be the j-th element in the group center value vector, representing the mean value of the measurements taken by the j-th fiber optic current transformer. This represents the historical normal measurement value of the j-th current transformer at the t-th sampling time.

3. The adaptive error correction method for a current transformer according to claim 1, characterized in that, The expression for calculating the real-time deviation is: ， In the formula, Let be the deviation at the t-th sampling time. For the measurement data at the t-th sampling time, The group centrality value, This is the transpose symbol.

4. The adaptive error correction method for a current transformer according to claim 1, characterized in that, The process of determining the optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic includes: The specific anomaly type and the specific target deviation feature are input into the XGBoost regression model; The XGBoost regression model outputs an optimal prediction step size that adapts to the current error change trend. When the anomaly type is a fixed deviation type, the output optimal prediction step size is constrained to a first value range. When the anomaly type is a random error mutation type, the output optimal prediction step size is constrained to a second value range that is smaller than the first value range.

5. The adaptive error correction method for a current transformer according to claim 1, characterized in that, The step of performing error prediction based on the optimal prediction step size and the error prediction model to obtain a prediction error value includes: A real-time measurement data of a certain abnormal current transformer is projected onto the residual space determined by the historical normal measurement data of the certain abnormal current transformer based on the principal component analysis method, and a residual vector sequence arranged in chronological order is obtained as the historical error sequence. Using the optimal prediction step size as the time length for the prediction target, the historical error sequence of the abnormal current transformer is input into the error prediction model; The error prediction model outputs an error prediction sequence for the future time period, and the predicted values ​​in the error prediction sequence corresponding to the current and future compensation times are used as the prediction error values.

6. A current transformer error adaptive correction system, characterized in that, include: The acquisition module is configured to acquire a real-time measurement data set of a current transformer cluster, wherein the current transformer cluster includes at least one current transformer, and the real-time measurement data set includes a real-time measurement data set corresponding to the at least one current transformer. The judgment module is configured to calculate the real-time deviation corresponding to the real-time measurement data set based on the pre-constructed group characteristic parameters, and to determine whether the real-time deviation is greater than a preset deviation threshold. The first determining module is configured to, if the deviation exceeds a preset threshold, determine at least one abnormal current transformer based on the individual contribution of each current transformer in the current transformer cluster, including: determining whether the individual contribution of a certain current transformer exceeds a preset contribution threshold, wherein the expression for calculating the individual contribution is: ， In the formula, Let be the individual contribution of the i-th current transformer to the deviation at the t-th sampling time. For the measurement data at the t-th sampling time, The group centrality value, Let be the deviation between the real-time measured value of the i-th current transformer and the group center value vector. Historical covariance matrix The element in the i-th row and z-th column of the inverse matrix, Let be the deviation between the real-time measured value of the z-th current transformer and the group center value vector; The expression for the preset contribution threshold is: ， In the formula, To preset the contribution threshold, To adjust the coefficient, This represents the total number of current transformers in the current transformer cluster. If the contribution is not greater than the preset contribution threshold, then the current transformer is defined as an abnormal current transformer; otherwise, the current transformer is defined as a normal current transformer. The fusion module is configured to extract the residual features of the at least one abnormal current transformer, fuse each residual feature with the real-time deviation to obtain at least one target deviation feature, and determine the abnormality type of the at least one abnormal current transformer based on the at least one target deviation feature. The abnormality type includes a fixed deviation type and a random error mutation type. The second determination module is configured to determine an optimal prediction step size corresponding to a certain abnormal current transformer based on a certain anomaly type and a certain target deviation characteristic. The prediction module is configured to select an error prediction model corresponding to the abnormal current transformer according to a preset model selection rule based on the abnormality type, and perform error prediction based on an optimal prediction step size and the error prediction model to obtain a prediction error value. The compensation module is configured to compensate for a certain real-time measurement data based on a certain prediction error value to obtain a final corrected target real-time measurement data, wherein the certain real-time measurement data is the real-time measurement data of a certain abnormal current transformer.

7. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method described in any one of claims 1 to 5.

Images