Methods, devices and electronic equipment for predicting the error status of capacitive voltage transformers

By extracting features and training regression fitting models from historical error data of capacitive voltage transformers, the problem of power outages required for error state inspection of capacitive voltage transformers has been solved, enabling real-time monitoring without power outages and improving the reliability and stability of the power system.

CN118035717BActive Publication Date: 2026-11-17MARKETING SERVICE CENT OF STATE GRID JILIN ELECTRIC POWER CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202311532506.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-16
Publication Date
2026-11-17
Estimated Expiration
2043-11-16

AI Technical Summary

Technical Problem

In the existing technology, the error status inspection of capacitive voltage transformers (CVTs) requires periodic power outages, which leads to high operational difficulty and cost.

Method used

By acquiring historical error data of capacitive voltage transformers, feature extraction and regression fitting models are used to extract and fit features, construct a training set, predict the error state of capacitive voltage transformers, and achieve real-time monitoring without power outages.

Benefits of technology

It enables real-time, convenient, and low-cost monitoring of the error status of capacitive voltage transformers under uninterrupted power conditions, thereby improving the reliability and stability of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118035717B_ABST
    Figure CN118035717B_ABST
Patent Text Reader

Abstract

This invention provides a method, apparatus, and electronic device for predicting the error state of a capacitive voltage transformer. The method includes: acquiring a first error dataset and a second error dataset of the capacitive voltage transformer; inputting the first error dataset into a feature extraction model for feature extraction to obtain a feature dataset, and constructing a training set based on the feature dataset and the second error dataset; inputting the training set into an initial regression fitting model for training to determine a fully trained regression fitting model; acquiring the test data of the capacitive voltage transformer, inputting the test data into the feature extraction model to obtain a feature dataset, and inputting the feature dataset into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer. This invention enables real-time, convenient, low-cost, and effective monitoring of the error state of a capacitive voltage transformer without power interruption.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power metering and monitoring technology, specifically to a method for predicting the error state of a capacitive voltage transformer. Background Technology

[0002] Capacitive voltage transformers (CVTs) have good insulation and economy, so they are widely used in power systems of 110kV and above. However, their structure is complex, their metering error is easily affected by a variety of factors, and their stability is low.

[0003] In existing technologies, the method of periodic power outage testing is often used to test the error status of CVTs. However, this method has the problems of high difficulty and high cost in power outage operations. Summary of the Invention

[0004] In view of this, it is necessary to provide a method for predicting the error status of a capacitive voltage transformer to solve the technical problems of the existing CVT error inspection method with periodic power outage inspection, which has the problems of high difficulty in power outage operation and high cost.

[0005] To address the aforementioned problems, this invention provides a method for predicting the error state of a capacitive voltage transformer, comprising:

[0006] Historical error data of a capacitive voltage transformer is obtained. The historical error data is error data collected at equal intervals. The historical error data is divided into a first error dataset including a first preset number of error data and a second error dataset including a second preset number of error data according to the time sequence.

[0007] The first error dataset is input into the feature extraction model for feature extraction to obtain a feature dataset, and a training set is constructed based on the feature dataset and the second error dataset;

[0008] The training set is input into the initial regression fitting model for training, and a fully trained regression fitting model is determined.

[0009] The test data of the capacitive voltage transformer is obtained, the test data is input into the feature extraction model to obtain the feature dataset, and the feature dataset is input into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer.

[0010] In some possible implementations, the feature extraction model includes a second preset number of feature extraction modules, each of which includes a signal decomposition unit, a feature extraction unit, a size restoration unit, and a residual unit.

[0011] The first error dataset is input into the feature extraction model for feature extraction to obtain a feature dataset, including:

[0012] Based on the signal decomposition unit, the first error dataset is decomposed to obtain several intrinsic mode components and margins;

[0013] Based on the feature extraction unit, feature extraction is performed on the plurality of intrinsic modal components and residuals to determine feature information;

[0014] Based on the size restoration unit, the feature information is accumulated column by column to determine the feature data;

[0015] Based on the residual unit, the first error dataset and the feature data are calculated and then input into the next feature extraction module for feature extraction. The above process is repeated until the last feature extraction module outputs feature data. A feature dataset is constructed based on the preset number of feature data output by the preset number of feature extraction modules.

[0016] In some possible implementations, the signal decomposition unit is used to perform signal decomposition on the first error dataset using a set of empirical model decomposition methods.

[0017] In some possible implementations, the feature extraction unit is used to extract features from the plurality of intrinsic modal components and residuals using an Inception layer.

[0018] In some possible implementations, the regression fitting model is any one of the Lasso regression model, the ridge regression model, and the elastic network regression model.

[0019] In some possible implementations, the regression fitting model is a regularized regression model, and the penalty term in the loss function of the regression fitting model is set with a Pearson correlation coefficient.

[0020] In some possible implementations, the loss function of the regression fitting model is:

[0021]

[0022]

[0023] z j =Pearson(X) j ,Y)

[0024] In the formula, β is the weighting coefficient, and J λ,α (β) is the loss function, X jLet Y be the j-th feature data, Y be the second error dataset, T be the number of data points in the second error dataset, and L be the number of feature extraction modules in the feature extraction model. For the sum of squared residuals, The term is the penalty term, where λ and α are the penalty coefficients, and |β| is the penalty term. j | is the L1 norm, For L2 norm, z j For X j The Pearson correlation coefficient between Y and Y.

[0025] On the other hand, the present invention also provides a capacitive voltage transformer error state prediction device, comprising:

[0026] The data acquisition unit is used to acquire historical error data of the capacitive voltage transformer. The historical error data is error data collected at equal intervals. The historical error data is divided into a first error dataset including a first preset number of error data and a second error dataset including a second preset number of error data according to the time sequence. It is also used to acquire the test data of the capacitive voltage transformer.

[0027] The feature extraction unit is used to input the first error dataset into the feature extraction model to extract features and obtain a feature dataset, and to construct a training set based on the feature dataset and the second error dataset. It is also used to input the test data into the feature extraction model to obtain a feature dataset set of the test data.

[0028] The regression fitting unit is used to input the training set into the initial regression fitting model for training, determine the fully trained regression fitting model, input the feature dataset of the test data into the fully trained regression fitting model, and obtain the error state prediction result of the capacitive voltage transformer.

[0029] On the other hand, the present invention also provides an electronic device, including a memory and a processor;

[0030] The memory is used to store programs;

[0031] The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps of the above-described method for predicting the error state of a capacitive voltage transformer.

[0032] On the other hand, the present invention also provides a computer-readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the above-described method for predicting the error state of a capacitive voltage transformer.

[0033] The beneficial effects of the above embodiments are as follows: The method for predicting the error state of a capacitive voltage transformer provided by the present invention first acquires historical error data of the capacitive voltage transformer. This historical error data consists of error data collected at equal intervals. The historical error data is then divided chronologically into a first error dataset including a first preset number of error data points and a second error dataset including a second preset number of error data points. The first error dataset is then input into a feature extraction model for feature extraction to obtain a feature dataset. A training set is constructed based on the feature dataset and the second error dataset. Finally, the training set is input into an initial regression fitting model for training, determining the completeness of the training. The regression fitting model is used to obtain the test data of the capacitive voltage transformer. The test data is then input into the feature extraction model to obtain the feature dataset of the test data. The feature dataset of the test data is then input into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer. By predicting the error state of the capacitive voltage transformer through the feature extraction model and the regression fitting model, the error state of the capacitive voltage transformer can be monitored in real time, conveniently, cost-effectively, and effectively without power interruption. This can provide targeted guidance for the operation and maintenance of the capacitive voltage transformer, promptly resolve error deviation problems, and improve the reliability and stability of the power system. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 A schematic flowchart of an embodiment of the capacitive voltage transformer error state prediction method provided by the present invention;

[0036] Figure 2 For the present invention Figure 1 A schematic diagram of an embodiment of step S102;

[0037] Figure 3 A schematic diagram of an embodiment of the capacitive voltage transformer error state prediction device provided by the present invention.

[0038] Figure 4 A schematic diagram of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0039] 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 a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0040] It should be understood that the illustrative drawings are not drawn to scale. The flowcharts used in this invention illustrate operations implemented according to some embodiments of the invention. It should be understood that the operations in the flowcharts may be implemented out of order, and steps without logical contextual relationships may be reversed or performed simultaneously. Furthermore, those skilled in the art, guided by the content of this invention, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.

[0041] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. "And / or" describes the relationship between related objects, indicating that three relationships may exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone.

[0042] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0043] This invention provides a method for predicting the error state of a capacitive voltage transformer, which will be described in detail below.

[0044] Figure 1 A schematic flowchart of an embodiment of the capacitive voltage transformer error state prediction method provided by the present invention is shown below. Figure 1 As shown, the error state prediction method for capacitive voltage transformers includes:

[0045] S101. Obtain historical error data of the capacitive voltage transformer. The historical error data is error data collected at equal intervals. Divide the historical error data into a first error dataset including the first preset number of error data and a second error dataset including the last preset number of error data according to the time sequence.

[0046] S102. Input the first error dataset into the feature extraction model to extract features, obtain the feature dataset, and construct a training set based on the feature dataset and the second error dataset;

[0047] S103. Input the training set into the initial regression fitting model for training, and determine the fully trained regression fitting model.

[0048] S104. Obtain the test data of the capacitive voltage transformer, input the test data into the feature extraction model to obtain the feature dataset of the test data, input the feature dataset of the test data into the fully trained regression fitting model, and determine the error state prediction result of the capacitive voltage transformer.

[0049] Compared with existing technologies, this invention first acquires historical error data of the capacitive voltage transformer. This historical error data consists of error data collected at equal intervals. The historical error data is then divided chronologically into a first error dataset (including a first preset number of error data points) and a second error dataset (including a second preset number of error data points). The first error dataset is then input into a feature extraction model for feature extraction, resulting in a feature dataset. A training set is constructed based on the feature dataset and the second error dataset. This training set is then input into an initial regression fitting model for training, determining a fully trained regression fitting model. Finally, the test data of the capacitive voltage transformer is acquired and input into the feature extraction model to obtain a feature dataset. This feature dataset is then input into the fully trained regression fitting model to obtain the predicted error state of the capacitive voltage transformer. By predicting the error state of the capacitive voltage transformer through the feature extraction model and the regression fitting model, the error state of the capacitive voltage transformer can be monitored in real-time, conveniently, cost-effectively, and efficiently without power interruption. This allows for targeted guidance on the operation and maintenance of the capacitive voltage transformer, timely resolution of error deviation problems, and improvement of the reliability and stability of the power system.

[0050] To better extract features from the first error dataset and thus better predict the error state of the capacitive voltage transformer, in some embodiments, the feature extraction model includes a second preset number of feature extraction modules. Each feature extraction module includes a signal decomposition unit, a feature extraction unit, a size restoration unit, and a residual unit. (Refer to...) Figure 2 Step S102 specifically includes:

[0051] S201. Based on the signal decomposition unit, perform signal decomposition on the first error dataset to obtain several intrinsic mode components and margins;

[0052] It should be noted that, in order to obtain better signal decomposition results, in this embodiment, the Ensemble Empirical Mode Decomposition (EEMD) method is used to decompose the first error dataset. In other embodiments, the Empirical Mode Decomposition (EMD) method can also be used.

[0053] Specifically, EMD decomposes data based on its own time-scale characteristics, breaking down complex data into a finite number of Intrinsic Mode Function (IMF) components and a residual. Each IMF component contains local feature information of the original sequence at different time scales. It does not require any pre-defined basis functions and is a time-frequency domain signal processing method. EMD has significant advantages in processing non-stationary and nonlinear data, is suitable for analyzing nonlinear and non-stationary sequences, and has a high signal-to-noise ratio.

[0054] The essence of the EEMD method is a multiple empirical mode decomposition with superimposed Gaussian white noise. It utilizes the statistical characteristic that Gaussian white noise has a uniform frequency distribution. By adding different white noises of the same amplitude each time, the extreme point characteristics of the signal are changed. Then, the corresponding IMFs obtained from multiple EMDs are averaged to cancel the added white noise, thereby effectively suppressing the generation of mode mixing. EEMD can convert one-dimensional data into two-dimensional data so that it can be processed by two-dimensional feature extraction methods. At the same time, the data processed by this method retains the internal features of the original data. Therefore, this embodiment uses the EEMD method.

[0055] S202. Based on the feature extraction unit, feature extraction is performed on several intrinsic modal components and residuals to determine feature information;

[0056] It should be noted that, to achieve better feature extraction results, in this embodiment, the feature extraction unit uses the Inception (also known as GoogLeNet) layer to extract features from several intrinsic modal components and residuals. Specifically, the core idea of ​​the Inception layer is to combine different convolutional layers in parallel, and then concatenate the resulting matrices processed by different convolutional layers along the depth dimension to form a deeper matrix. It can be repeatedly stacked to form a larger network, efficiently expanding the depth and width of the network, improving the accuracy of deep learning networks while preventing overfitting. It can also perform dimensionality reduction on larger matrices and aggregate information at different sizes, facilitating feature extraction from different scales. Therefore, this embodiment uses the Inception module for feature extraction. In other embodiments, CNN and other network structures can also be used.

[0057] S203. Based on the size restoration unit, the feature information is accumulated column by column to determine the feature data;

[0058] It should be noted that the original data can be obtained by summing the IMF components and the remainder after EEMD decomposition. This is one of the properties of EEMD. In this embodiment, this property is also used to sum the feature information column by column to restore the data size.

[0059] S204. Based on the residual unit, the first error dataset and feature data are calculated and then input into the next feature extraction module for feature extraction.

[0060] S205. Repeat the above process until the last feature extraction module outputs feature data. Construct a feature dataset based on the preset number of feature data output by the preset number of feature extraction modules.

[0061] It should be noted that adding residual modules to the model's network structure makes the model easier to optimize, increases network depth while maintaining the original characteristics of the data, alleviates the gradient vanishing problem, and improves accuracy.

[0062] To achieve better prediction results, in some embodiments, the loss function of the regression fitting model in step S103 is:

[0063]

[0064]

[0065]

[0066] In the formula, β is the weighting coefficient, and J λ,α (β) is the loss function, [X] 1,...,X L [X] represents the feature dataset. j Let Y be the j-th feature data, Y be the second error dataset, T be the number of data points in the second error dataset (i.e., the first preset number), and L be the number of feature extraction modules in the feature extraction model (i.e., the second preset number). For the sum of squared residuals, As a penalty term, |β j | is the L1 norm, Let z be the L2 norm, λ and α be penalty coefficients, where λ controls the overall sparsity of the weighted coefficient solution, and α determines the relative weights of the L1 and L2 norms. j For X j The Pearson correlation coefficient between X and Y, conv(X) j (,Y) is X j covariance of Y For X j Standard deviation, σ Y The standard deviation of Y is given.

[0067] It should be noted that the regression fitting model above is based on the elastic network regression model, with the addition of the Pearson correlation coefficient to the penalty term.

[0068] Elastic network regression is a linear regression model that combines Lasso regression and ridge regression. Its goal is to find the optimal regression coefficients. Based on the least squares method, its loss function includes a mean squared error term and a regularization term. The mean squared error term measures the difference between predicted and actual values, while the regularization term consists of Lasso regression and ridge regression terms. By adjusting the weights of these two regularization terms, elastic network regression can simultaneously consider the magnitude and number of feature values ​​in the model. Compared to other regression models, elastic network regression performs better in this invention. Therefore, in this embodiment, elastic network regression is used. In other embodiments, linear regression models such as Lasso regression and ridge regression can also be used.

[0069] The Pearson correlation coefficient is used to measure the linear correlation between two variables and can indicate the relationship between a single feature and the CVT error. In the regression fitting model of this invention, the value of the Pearson correlation coefficient has a certain influence on the weight coefficient of the linear regression model. Specifically, the larger the Pearson correlation coefficient between a certain influencing factor and the CVT error state, the more significant the influence of the influencing factor on the error value. In this case, the relevant information of the influencing factor should be retained and a smaller penalty should be given so that different penalty weights can be set in the elastic network regression according to the Pearson correlation coefficient of each influencing factor, so as to better fit the error value and calculate the optimal weight coefficient.

[0070] A well-trained regression model can be expressed as:

[0071]

[0072] In the formula, Y′ is the predicted value of the error data, and X j Let L be the number of feature extraction modules, and β be the j-th feature data. j These are the optimal weighting coefficients.

[0073] By inputting the data to be tested into the feature extraction model and the fully trained regression fitting model described above, the predicted error data value for the next time period can be obtained. Then, the error state of the capacitive voltage transformer can be judged based on the predicted error data value for the next time period.

[0074] To better implement the capacitive voltage transformer error state prediction method in this embodiment of the invention, based on the existing capacitive voltage transformer error state prediction method, correspondingly, as follows: Figure 3 As shown, this embodiment of the invention also provides a capacitive voltage transformer error state prediction device 300, comprising:

[0075] The data acquisition unit 301 is used to acquire historical error data of the capacitive voltage transformer. The historical error data is error data collected at equal intervals. The historical error data is divided into a first error dataset including the first preset number of error data and a second error dataset including the last preset number of error data according to the time sequence. It is also used to acquire the test data of the capacitive voltage transformer.

[0076] The feature extraction unit 302 is used to input the first error dataset into the feature extraction model to extract features and obtain a feature dataset, and to construct a training set based on the feature dataset and the second error dataset. It is also used to input the test data into the feature extraction model to obtain the feature dataset of the test data.

[0077] The state prediction unit 303 is used to input the training set into the initial regression fitting model for training, determine the fully trained regression fitting model, and also to input the feature dataset of the test data into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer.

[0078] The capacitive voltage transformer error state prediction device 300 provided in the above embodiments can realize the technical solutions described in the above embodiments of the capacitive voltage transformer error state prediction method. The specific implementation principles of each unit can be found in the corresponding content in the above embodiments of the capacitive voltage transformer error state prediction method, and will not be repeated here.

[0079] like Figure 4As shown, the present invention also provides an electronic device 400. The electronic device 400 includes a processor 401, a memory 402, and a display 403. Figure 4 Only some components of the electronic device 400 are shown, but it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

[0080] In some embodiments, memory 402 may be an internal storage unit of electronic device 400, such as a hard disk or memory of electronic device 400. In other embodiments, memory 402 may also be an external storage device of electronic device 400, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on electronic device 400.

[0081] Furthermore, the memory 402 may include both internal storage units of the electronic device 400 and external storage devices. The memory 402 is used to store application software and various types of data installed on the electronic device 400.

[0082] In some embodiments, processor 401 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 402 or process data, such as the capacitive voltage transformer error state prediction method of the present invention.

[0083] In some embodiments, display 403 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 403 is used to display information from electronic device 400 and to display a visual user interface. Components 401-403 of electronic device 400 communicate with each other via a system bus.

[0084] In some embodiments of the present invention, when the processor 401 executes the capacitive voltage transformer error state prediction program in the memory 402, the following steps can be implemented:

[0085] The historical error data of the capacitive voltage transformer is obtained. The historical error data is the error data collected at equal intervals. The historical error data is divided into a first error dataset including the first preset number of error data and a second error dataset including the last preset number of error data according to the time sequence.

[0086] The first error dataset is input into the feature extraction model to extract features, resulting in a feature dataset. A training set is then constructed based on the feature dataset and the second error dataset.

[0087] The training set is input into the initial regression fitting model for training, and the fully trained regression fitting model is determined.

[0088] The test data of the capacitive voltage transformer is obtained, and the test data is input into the feature extraction model to obtain the feature dataset of the test data. The feature dataset of the test data is then input into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer.

[0089] It should be understood that when the processor 401 executes the error state prediction program for the capacitive voltage transformer in the memory 402, in addition to the functions mentioned above, it can also perform other functions, as detailed in the description of the corresponding method embodiments above.

[0090] Furthermore, this embodiment of the invention does not specifically limit the type of electronic device 400 mentioned. Electronic device 400 can be a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, laptop computer, or other portable electronic device. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic device can also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the invention, electronic device 400 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0091] Accordingly, this application also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by a processor, they can implement the steps or functions of the capacitive voltage transformer error state prediction method provided in the above-described method embodiments.

[0092] The above provides a detailed description of the error state prediction method for a capacitive voltage transformer provided by the present invention. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method of predicting error state of a capacitive voltage transformer, characterized by, include: Historical error data of a capacitive voltage transformer is obtained. The historical error data is error data collected at equal intervals. The historical error data is divided into a first error dataset including a first preset number of error data and a second error dataset including a second preset number of error data according to the time sequence. The first error dataset is input into the feature extraction model for feature extraction to obtain a feature dataset, and a training set is constructed based on the feature dataset and the second error dataset; the feature extraction model includes a second preset number of feature extraction modules, and the feature extraction module includes a signal decomposition unit, a feature extraction unit, a size restoration unit and a residual unit; The first error dataset is input into a feature extraction model for feature extraction to obtain a feature dataset, which includes: decomposing the first error dataset into a signal based on the signal decomposition unit to obtain several intrinsic modal components and residuals; extracting features from the several intrinsic modal components and residuals based on the feature extraction unit to determine feature information; accumulating the feature information column by column based on the size restoration unit to determine feature data; and inputting the first error dataset and the feature data into the next feature extraction module for feature extraction after calculation based on the residual unit. The above process is repeated until the last feature extraction module outputs feature data, and a feature dataset is constructed based on the preset number of feature data output by the preset number of feature extraction modules. The training set is input into the initial regression fitting model for training to determine a fully trained regression fitting model; the regression fitting model is a regularized regression model, and the penalty term in the loss function of the regression fitting model is set with the Pearson correlation coefficient; the loss function of the regression fitting model is: In the formula, These are the weighting coefficients. For loss function, For the first Each feature data, For the second error dataset, The number of data points in the second error dataset. This refers to the number of feature extraction modules in the feature extraction model. For the sum of squared residuals, As a penalty item, and The penalty coefficient is... It is an L1 norm. It is the L2 norm. for and The Pearson correlation coefficient between them; The test data of the capacitive voltage transformer is obtained, and the test data is input into the feature extraction model to obtain the feature dataset of the test data. The feature dataset of the test data is then input into the fully trained regression fitting model to determine the error state prediction result of the capacitive voltage transformer.

2. The method for predicting the error state of a capacitive voltage transformer according to claim 1, characterized in that, The signal decomposition unit is used to decompose the first error dataset using a set of empirical model decomposition methods.

3. The method for predicting the error state of a capacitive voltage transformer according to claim 1, characterized in that, The feature extraction unit is used to extract features from the several intrinsic modal components and residuals using the Inception layer.

4. The method for predicting the error state of a capacitive voltage transformer according to claim 1, characterized in that, The regression fitting model can be any one of the Lasso regression model, ridge regression model, and elastic network regression model.

5. A device for predicting the error state of a capacitive voltage transformer, characterized in that, include: The data acquisition unit is used to acquire historical error data of the capacitive voltage transformer. The historical error data is error data collected at equal intervals. The historical error data is divided into a first error dataset including a first preset number of error data and a second error dataset including a second preset number of error data according to the time sequence. It is also used to acquire the test data of the capacitive voltage transformer. The feature extraction unit is used to input the first error dataset into the feature extraction model for feature extraction to obtain a feature dataset, and to construct a training set based on the feature dataset and the second error dataset. It is also used to input the test data into the feature extraction model to obtain a feature dataset of the test data. The feature extraction model includes a second preset number of feature extraction modules, and the feature extraction modules include a signal decomposition unit, a feature extraction unit, a size restoration unit, and a residual unit. The first error dataset is input into a feature extraction model for feature extraction to obtain a feature dataset, which includes: decomposing the first error dataset into a signal based on the signal decomposition unit to obtain several intrinsic modal components and residuals; extracting features from the several intrinsic modal components and residuals based on the feature extraction unit to determine feature information; accumulating the feature information column by column based on the size restoration unit to determine feature data; and inputting the first error dataset and the feature data into the next feature extraction module for feature extraction after calculation based on the residual unit. The above process is repeated until the last feature extraction module outputs feature data, and a feature dataset is constructed based on the preset number of feature data output by the preset number of feature extraction modules. The regression fitting unit is used to train an initial regression fitting model by inputting the training set into the model, thereby determining a fully trained regression fitting model. It is also used to input the feature dataset of the test data into the fully trained regression fitting model to obtain the error state prediction result of the capacitive voltage transformer. The regression fitting model is a regularized regression model, and the penalty term in the loss function of the regression fitting model includes a Pearson correlation coefficient. The loss function of the regression fitting model is: In the formula, These are the weighting coefficients. For loss function, For the first Each feature data, For the second error dataset, The number of data points in the second error dataset. This refers to the number of feature extraction modules in the feature extraction model. For the sum of squared residuals, As a penalty item, and The penalty coefficient is... It is an L1 norm. It is the L2 norm. for and The Pearson correlation coefficient between them.

6. An electronic device, characterized in that, Including memory and processor; The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps of the capacitive voltage transformer error state prediction method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The readable storage medium stores a program or instructions, which, when executed by a processor, implement claim 1 above. The steps of the method for predicting the error state of a capacitive voltage transformer as described in any one of 4.

Citation Information

Patent Citations

  • Public steel structure building micro-strain prediction method and system

    CN113343470A

  • Method and system suitable for measuring errors of capacitive transformer

    CN113567904A