Metering degradation trend prediction method and device of capacitor voltage transformer

Abstract

The invention discloses a measurement degradation trend prediction method and device for a capacitor voltage transformer. The method comprises the following steps: carrying out a gradual change degradation test on the capacitor voltage transformer in a mode that a preset ratio difference is gradually increased, and obtaining secondary side voltage time sequence sample data; calculating a performance state index sample of the capacitor voltage transformer according to the secondary side voltage time sequence sample data; decomposing the performance state index sample, obtaining performance state degradation trend sample data and performance state fluctuation sample data, training a double-Gaussian model, obtaining a degradation trend prediction model, training a KAN algorithm, and obtaining a fluctuation prediction model; and respectively carrying out degradation trend and fluctuation prediction on the obtained historical sequence data of the to-be-tested capacitor voltage transformer, obtaining a degradation trend prediction sequence and a fluctuation prediction sequence of the to-be-tested capacitor voltage transformer, and determining a degradation trend prediction result of the to-be-tested capacitor voltage transformer.

Description

Technical Field

[0001] The present invention relates to the technical field of online prediction of the metering performance state of a capacitor voltage transformer, and more specifically, to a method and device for predicting the metering degradation trend of a capacitor voltage transformer. Background Art

[0002] Capacitive voltage transformer (CVT) converts high voltage into low voltage through capacitor voltage division, and then transforms it through the intermediate transformer to provide it for metering, relay protection, automatic control, signal indication, etc. CVT has the advantages of low cost, high reliability, small size and simple installation. Therefore, CVT has been widely used in high-voltage power grids.

[0003] Compared with the traditional electromagnetic voltage transformer, the CVT structure is more complex. Capacitor voltage dividers and compensating reactors are added to the CVT structure, so it is not as stable as the traditional electromagnetic voltage transformer and is prone to faults such as measurement error drift. Among them, the measurement error of CVT is the most common fault of CVT after long-term operation. It is stipulated that the measurement ratio difference of CVT shall not exceed ±0.2% when it is in normal working state, and the CVT needs to be replaced when it exceeds ±0.2%. The measurement error of CVT is not conducive to the fairness of electricity trading. The most commonly used detection technology in the power grid is to use standard CVT equipment for regular offline experiments. However, since the experiment is generally carried out once every few years and has the disadvantages of large time consumption and high cost, regular offline experiments are not conducive to timely detection of CVT ratio drift.

[0004] Currently, scholars have proposed various types of error identification methods for CVT errors, including methods based on model methods, data-driven methods, and statistical indicators. The model method identifies the measurement error of CVT by establishing an accurate model of CVT. Regarding the data-driven method, methods such as LSTM, Markov model, and gate recursive unit have been proposed for CVT measurement error identification. Regarding the statistical indicator method, scholars have proposed a method of using the PCA method to calculate the Q value of the secondary voltage as an indicator for evaluating the error state of CVT [15-18]. The above are all error identification methods. There is currently a lack of research on the prediction of future changes in CVT measurement errors, and there is a lack of characterization methods for the measurement error performance of CVT. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and device for predicting the metrological degradation trend of a capacitive voltage transformer.

[0006] According to one aspect of the present invention, a method for predicting a metrological degradation trend of a capacitor voltage transformer is provided, comprising:

[0007] A gradual degradation test of the capacitor voltage transformer is performed by gradually increasing the preset ratio difference to obtain secondary side voltage time series sample data;

[0008] The performance status index samples of capacitor voltage transformer are calculated based on the secondary side voltage time series sample data by using wavelet transform method.

[0009] The performance status indicator samples are decomposed using the variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data;

[0010] The double Gaussian model is trained using the performance state degradation trend sample data to obtain the degradation trend prediction model, and the KAN algorithm is trained using the performance state fluctuation sample data to obtain the fluctuation prediction model;

[0011] Using a degradation trend prediction model and a fluctuation prediction model, respectively, the acquired historical sequence data of the capacitor voltage transformer to be tested are used to perform degradation trend and fluctuation prediction, so as to obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested;

[0012] According to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested, a degradation trend prediction result of the capacitor voltage transformer to be tested is determined.

[0013] Optionally, the expression of the preset ratio difference is:

[0014]

[0015] Where V s is the secondary voltage amplitude, V f is the primary voltage amplitude, and k is the rated transformation ratio of the capacitive voltage transformer.

[0016] Optionally, a performance status index sample of the capacitor voltage transformer is calculated based on the secondary side voltage time series sample data using a wavelet transform method, including:

[0017] The secondary side voltage time series sample data is decomposed by wavelet transform to obtain the approximate coefficient sequence;

[0018] According to the maximum approximation coefficient in the approximation coefficient sequence, a performance status index sample of the capacitor voltage transformer is calculated.

[0019] Optionally, the calculation formula of the performance status indicator sample is:

[0020]

[0021] Where, MAC iIt represents the maximum approximation coefficient of the calculated test data, and MAC0 represents the maximum approximation coefficient of the normal state.

[0022] Optionally, the performance status indicator samples are decomposed using a variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data, including:

[0023] The VMD algorithm is used to construct the variational subject according to the performance status indicator samples;

[0024] The average difference is used to solve the variational subject, and the performance state degradation trend sample data and the performance state fluctuation sample data are obtained.

[0025] Optionally, a VMD algorithm is used to construct a variational subject according to the performance status indicator sample, including:

[0026] The one-sided spectrum of each mode is calculated by Hilbert transform;

[0027] Based on the unilateral spectrum of each mode, the spectrum of each mode is modulated to the response frequency range;

[0028] Based on the response frequency range of each mode and the performance status index samples, the variational subject of unconstrained variation is constructed.

[0029] Alternatively, the expression of the variational body of the unconstrained variation is:

[0030]

[0031] Where k is the number of eigenmode components, {μ k}={μ1,...,μ K} represents K modal components; {ω k}={ω1,...,ω K} represents the frequency center of the modal component; represents the partial derivative with respect to t, K is the number of modal components, λ is the Lagrange multiplier, α is the quadratic penalty factor, δ(t) is the impulse function, t is the time index; j is the imaginary unit, f(t) is the sum of all modal components, L() is the unconstrained variational modal constraint model, ||*|| 2 is the L2 norm, ∑ K () is the summation symbol.

[0032] Optionally, the average difference is quoted to solve the variational subject to obtain performance state degradation trend sample data and performance state fluctuation sample data, including:

[0033] When the number of decomposition layers of the variational subject is 1, the current average difference is determined based on the current residual and the original performance status indicator sample;

[0034] The average difference is compared with the preset threshold. When the average difference is not less than the preset threshold, the performance status indicator sample is decomposed to the next layer and the average difference of the next layer is calculated. When the average difference is less than the preset threshold, the decomposition is stopped to obtain the performance status degradation trend sample data and the performance status fluctuation sample data.

[0035] According to another aspect of the present invention, there is provided a device for predicting a metrological degradation trend of a capacitor voltage transformer, comprising:

[0036] A test module, used for performing a gradual degradation test on the capacitor voltage transformer by gradually increasing a preset ratio difference, and obtaining secondary side voltage time series sample data;

[0037] A calculation module, used to calculate the performance status index sample of the capacitor voltage transformer according to the secondary side voltage time series sample data by using the wavelet transform method;

[0038] A decomposition module is used to decompose the performance status indicator samples by using the variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data;

[0039] A training module is used to train a double Gaussian model using performance state degradation trend sample data to obtain a degradation trend prediction model, and to train a KAN algorithm using performance state fluctuation sample data to obtain a fluctuation prediction model;

[0040] A prediction module, used to use a degradation trend prediction model and a fluctuation prediction model to respectively predict the degradation trend and fluctuation of the acquired historical sequence data of the capacitor voltage transformer to be tested, and obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested;

[0041] The determination module is used to determine the degradation trend prediction result of the capacitor voltage transformer to be tested according to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested.

[0042] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute the method described in any one of the above aspects of the present invention.

[0043] According to another aspect of the present invention, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of the above aspects of the present invention.

[0044] Thus, the present invention first performs a wavelet transform on the measured CVT secondary voltage. Approximate coefficients and detail coefficients are formed after multi-layer wavelet decomposition. The maximum approximate coefficient is selected from the approximate coefficients, and the maximum approximate coefficient is used to calculate the metrological performance state of the CVT to characterize the metrological error. Subsequently, a variational mode decomposition-average difference method is proposed to perform multi-layer decomposition on the SOP of the CVT. The decomposed residual term is used to characterize the degradation trend part of the SOP, and a double Gaussian model is used for modeling and prediction; the intrinsic mode function of the SOP is decomposed to characterize the degradation fluctuation part of the SOP, and the KAN network algorithm is used for prediction. Finally, all the prediction results are accumulated to obtain a predicted value to characterize the SOP change trend. The prediction of the disadvantage change of the capacitive voltage transformer is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:

[0046] Figure 1 It is a flow chart of a method for predicting a metering degradation trend of a capacitor voltage transformer provided by an exemplary embodiment of the present invention;

[0047] Figure 2 is a structural schematic diagram of a capacitive voltage transformer provided by an exemplary embodiment of the present invention;

[0048] Figure 3 is a schematic diagram of a KAN structure provided by an exemplary embodiment of the present invention;

[0049] Figure 4a to Figure 4c They are schematic diagrams of voltage test data of phase A, phase B and phase C in a degradation experiment provided by an exemplary embodiment of the present invention;

[0050] Figure 5 It is a schematic diagram of SOP degradation trend of each phase degradation experiment provided by an exemplary embodiment of the present invention;

[0051] Figure 6a to Figure 6c They are schematic diagrams of SOP decomposition results of phase A, phase B, and phase C provided by an exemplary embodiment of the present invention;

[0052] Figures 7a to 7c They are schematic diagrams of double Gaussian fitting results of SOP residual terms of degradation experiments of phase A, phase B, and phase C provided by an exemplary embodiment of the present invention;

[0053] Figure 8a to Figure 8e They are respectively schematic diagrams of prediction results of IMF1, IMF2, IMF3, IMF4 and IMF5 of the SOP decomposition of phase A provided by an exemplary embodiment of the present invention;

[0054] Figure 9a to Figure 9eThey are respectively schematic diagrams of prediction results of IMF1, IMF2, IMF3, IMF4 and IMF5 of the B-phase SOP decomposition provided by an exemplary embodiment of the present invention;

[0055] Figure 10a to Figure 10e They are respectively schematic diagrams of prediction results of IMF1, IMF2, IMF3, IMF4, and IMF5 of the C-phase SOP decomposition provided by an exemplary embodiment of the present invention;

[0056] Figure 11a to Figure 11c 1 is a schematic diagram of the overall prediction results of the SOP of the A phase, the B phase and the C phase of the CVT provided by an exemplary embodiment of the present invention;

[0057] Fig.12 It is a structural schematic diagram of a device for predicting a metering degradation trend of a capacitor voltage transformer provided by an exemplary embodiment of the present invention;

[0058] Fig.13 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0059] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0060] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0061] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0062] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.

[0063] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0064] In addition, the term "and / or" in the present invention is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects before and after are in an "or" relationship.

[0065] It should also be understood that the description of the various embodiments of the present invention focuses on the differences between the various embodiments, and the same or similar aspects thereof can be referenced to each other, and for the sake of brevity, they will not be described one by one.

[0066] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0067] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0068] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0069] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0070] Embodiments of the present invention can be applied to electronic devices such as terminal devices, computer systems, servers, etc., which can operate with many other general or special computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, servers, etc. include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network personal computers, small computer systems, large computer systems, and distributed cloud computing technology environments including any of the above systems, etc.

[0071] Electronic devices such as terminal devices, computer systems, servers, etc. can be described in the general context of computer system executable instructions (such as program modules) executed by computer systems. Generally, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in a distributed cloud computing environment, where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can be located on local or remote computing system storage media including storage devices.

[0072] Exemplary Methods

[0073] Figure 1FIG. 1 is a flow chart of a method for predicting a measurement degradation trend of a capacitor voltage transformer provided by an exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as Figure 1 As shown, the method 100 for predicting the metering degradation trend of a capacitor voltage transformer comprises the following steps:

[0074] Step 101, performing a gradual degradation test on a capacitor voltage transformer by gradually increasing a preset ratio difference, and obtaining secondary side voltage time series sample data;

[0075] Step 102, using a wavelet transform method, based on the secondary side voltage time series sample data, calculate the performance status index sample of the capacitor voltage transformer;

[0076] Step 103, decomposing the performance status indicator samples by using the variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data;

[0077] Step 104, using the performance state degradation trend sample data to train the double Gaussian model to obtain a degradation trend prediction model, and using the performance state fluctuation sample data to train the KAN algorithm to obtain a fluctuation prediction model;

[0078] Step 105, using a degradation trend prediction model and a fluctuation prediction model to respectively perform degradation trend and fluctuation prediction on the acquired historical sequence data of the capacitor voltage transformer to be tested, to obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested;

[0079] Step 106: Determine a degradation trend prediction result of the capacitor voltage transformer to be tested according to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested.

[0080] Specifically, this patent designs a metrological performance index for CVT to characterize the CVT metrological error, and proposes a metrological performance prediction method based on the double Gaussian model-KAN. First, a wavelet transform is performed on the measured CVT secondary voltage. After multi-layer wavelet decomposition, approximate coefficients and detail coefficients are formed. The maximum approximate coefficient is selected from the approximate coefficients, and the maximum approximate coefficient is used to calculate the metrological performance state (State of Performance, SOP) of the CVT to characterize the metrological error. Subsequently, the variational modal decomposition-mean difference (Variational Modal Decomposition Mean Difference, VMD-MD) method is proposed to perform multi-layer decomposition on the SOP of CVT. The decomposed residual term is used to characterize the degradation trend part of the SOP, and the double Gaussian model is used for modeling and prediction; the intrinsic mode function (Intrinsic Mode Functions, IMF) of the SOP is decomposed to characterize the degradation fluctuation part of the SOP, and the Kolmogorov-Arnold network (KAN) algorithm is used for prediction. Finally, all the prediction results are accumulated to obtain the predicted value to characterize the SOP change trend. Finally, the proposed prediction method was tested by using the SWCVT-3 CVT online test system of the China Electric Power Research Institute to collect three-phase voltage data with gradually increasing ratio differences. In the test, the measurement SOP of CVT calculated by wavelet transform method can effectively characterize the measurement error of CVT, and the proposed dual Gaussian model-KAN hybrid prediction can accurately predict the SOP degradation process of CVT. The specific technical scheme is as follows:

[0081] 1. CVT and ratio difference:

[0082] like Figure 2 As shown in Figure 1, CVT consists of a capacitive voltage divider (CVD), a compensating inductor (CI), and an intermediate voltage transformer (IVT). The CVD is composed of hundreds of polypropylene film capacitors connected in series, which is equivalent to a high-voltage capacitor C1 and a low-voltage capacitor C2. The high-voltage signal is decomposed into a medium-voltage signal by the CVD, and then measured by the IVT.

[0083] The measurement error of CVT includes ratio difference and phase shift. Ratio difference describes the difference between the actual rate of change and the rated rate of change. Phase difference describes the phase difference between the primary voltage and the secondary voltage phase. Ratio difference is the main error fault mode of CVT.

[0084] The ratio error of CVT is defined as e in equation (1), where Vs is the secondary voltage amplitude, V f is the primary voltage amplitude, k is the rated transformation ratio.

[0085]

[0086] In order to meet the requirements of redundant configuration, many CVTs are connected to the transmission lines, busbars and power transformers in the substation in the same phase. The impedance between these CVTs can be ignored because there is no other equipment between them except the conductor. Therefore, the CVTs connected to the same phase actually measure the same primary voltage, which can be considered as a single-phase relationship. Suppose n CVTs are connected to the same single phase, and the secondary voltage of the i-th CVT is shown in formula (2):

[0087]

[0088] Among them, V i f and V i s is the secondary voltage and primary voltage amplitude of the i-th CVT, e i is the ratio difference of the i-th CVT. Since all CVTs are connected to the same primary conductor at the same voltage level, all V i f are equal to the primary voltage V f .

[0089] 2. CVT SOP definition:

[0090] Currently, there is no definition of SOP index for CVT. This patent proposes a SOP index definition method based on wavelet transform.

[0091] Wavelet transform is an effective technology that has been widely used in the field of signal analysis. Wavelet transform can decompose the signal into approximate signal and detail signal, thereby mining the basic characteristics of the signal. Assuming f(x) is the measurement signal, the wavelet transform formula is:

[0092]

[0093] Where ψ(x) is the mother wavelet, and a, b and c(a,b) are the scaling parameter, translation parameter and wavelet coefficients of the signal respectively.

[0094] Assume {V k} k∈Z For orthogonal multi-resolution analysis, {W k} k∈Z is the associated wavelet space. f(x) in V k The projection on is as follows:

[0095]

[0096] in and They are 2k+1 Resolution V k+1 and W k+1 The projection of f(x) on ; and Respectively represent 2 k+1 The scale factor and wavelet coefficient of resolution f(x); φ k+1 and ψ k+1 Respectively represent 2 k+1 The scale function and wavelet function of the resolution. Therefore, c k+1 and d k+1 Represents 2 k+1 The approximate coefficients and detail coefficients of f(x) at resolution {V k} k∈Z This breaks down as follows:

[0097]

[0098] Through decomposition, a series of approximate coefficients and detail coefficients can be formed. After multi-layer decomposition of the original signal, an approximate coefficient sequence can be obtained. The maximum approximate coefficient is selected from the sequence, and the performance index SOP of CVT is calculated using the following formula:

[0099]

[0100] Where MACi represents the calculated maximum approximation coefficient of the test data, and MAC0 represents the maximum approximation coefficient of the normal state.

[0101] 3. SOP degradation prediction method:

[0102] For the measured three-phase signal data of the secondary side of CVT, the maximum approximation coefficient is first decomposed by wavelet transform to calculate the SOP of CVT. Then, the VMD-MD decomposition method is used to perform multi-layer decomposition of the SOP of CVT. The decomposed SOP degradation trend part is modeled and predicted using the double Gaussian model, and the decomposed SOP fluctuation part is predicted using the KAN algorithm.

[0103] 3.1 VMD-MD signal decomposition method:

[0104] Due to the influence of the primary voltage of the power grid and the error of the measuring equipment itself, the SOP signal may experience local jitter. In order to more accurately predict the degradation trend of CVT, this patent will use VMD-MD to decompose the SOP signal.

[0105] The VMD algorithm is a novel signal processing method that adaptively decomposes the SOP signal into the center frequency ω k Different mode sets μ k(k=1,...,K). It transforms the decomposition into two detailed processes: constructing the variational body and solving the variational body. The process of constructing the variational problem is as follows.

[0106] Step 1: Calculate the one-sided spectrum of each mode through Hilbert transform. The one-sided spectrum is expressed as:

[0107]

[0108] Where δ(t) is the pulse function, t is the time index, and j is the imaginary unit.

[0109] Step 2: Modulate the spectrum of each mode to the response frequency range:

[0110]

[0111] Step 3: Construct constrained variational problem:

[0112]

[0113] where {μ k}={μ1,...,μ K} represents K modal components; {ω k}={ω1,...,ω K} represents the frequency center of the modal component; denotes the partial derivative with respect to t, and K is the number of modal components.

[0114] In order to overcome the variational problem, the Lagrangian multiplier λ and the quadratic penalty factor α are introduced to transform the constrained variational problem into an unconstrained variational problem. The augmented Lagrangian is described as:

[0115]

[0116] The VMD algorithm that introduces the mean difference (MD) is the VMD-MD algorithm.

[0117] Determine the decomposition level h by MD. When h=1, the current residual and the original signal are calculated. When k>1, the MD of the decomposed residual term and the residual term of the previous decomposition is used. When MD is lower than the threshold, it is considered that the residual has not changed much, and the decomposition is terminated.

[0118]

[0119] Where R h i is the i-th parameter of the residual of the h-th decomposition of the SOP of the CVT. n is the length of the decomposed residual, that is, the number of elements.

[0120] 3.2. Double Gaussian model:

[0121] In real-life scenarios, the collected signals are more complex, and a single Gaussian model may not be able to explain the collected signals, which can be regarded as a combination of multiple groups of Gaussian distributions. Therefore, the use of classic signal fitting methods such as polynomial fitting and Gaussian fitting cannot achieve ideal results. Therefore, this patent proposes a dual Gaussian model method to decompose complex data into two independent components, and quantify the characteristics of each independent component by adjusting the parameters of the model, so as to more accurately fit the changing trend of the signal, significantly improve the accuracy of the fitting, and obtain a more accurate model, providing a more reliable scientific basis for subsequent signal change trend predictions. The working principle of the dual Gaussian model can be summarized as the following formula.

[0122]

[0123] Where Y is the target variable, x is the input feature, a1 and a2 are the amplitudes of two independent signals, b1 and b2 are the means of two independent signals, and c1 and c2 are the variances of two independent signals.

[0124] 3.3 KAN

[0125] KAN is a novel deep learning model. Different from the traditional MLP model, it uses the Kolmogorov-Arnold theorem as its basic working principle. It represents any multivariable continuous function by a two-layer nested combination of multiple sets of single-variable functions. The KAN model adopts edge learning in structure. What is connected between the layers of KAN is not a specific weight value, but a one-dimensional function represented by B-Spline, which replaces the linear transformation and nonlinear transformation learning method in MLP and solves the drawbacks of the MLP model in practical applications, such as large number of model parameters and poor interpretability. The specific working principle of Kolmogorov-Arnold is as follows:

[0126]

[0127] where x p represents the pth variable in the multivariate list, q is the number of indices of the external function, p is the dimension of the input vector, i.e. the number of variables, Φ is the combination of single variable functions, Function of a single variable.

[0128] The activation function of the KAN model can be dynamically adjusted based on the B-spline method, which can automatically adapt to the dynamic changes of the input data, so that KAN can dynamically learn based on the data, thereby improving the robustness of the model and improving the efficiency and interpretability of the model. The KAN network architecture is as follows: Figure 3 shown.

[0129] In one embodiment of the present application, an experiment is conducted using the method for predicting the measurement degradation trend of a capacitive voltage transformer proposed in the present application. The specific process is as follows:

[0130] 1. Experimental process:

[0131] 1) Perform wavelet transform on the collected three-phase voltage data on the secondary side of the CVT, extract the maximum approximation coefficient, and convert it into the SOP of the CVT through formula (6), so as to quantify the metering performance status of the CVT.

[0132] 2) Divide the data into training data and test data.

[0133] 3) Use VMD-MD algorithm to perform multi-layer decomposition on the CVT SOP obtained in the above step, wherein the number of decomposition layers is determined by the proposed average difference method.

[0134] 4) For the SOP change trend data decomposed by VMD-MD, the double Gaussian model method is used to model and predict future trends. For other fluctuation data of decomposed SOP, the KAN method is used to predict future trends.

[0135] 5) Accumulate the predicted results to obtain the predicted SOP metrological performance prediction data and compare it with the test data.

[0136] 2. CVT error over-limit fault identification

[0137] Based on the CVT online control system SWCVT-3, the A-phase error over-limit test of CVT is carried out. The system includes HEW99 full-function transformer tester and six voltage transformers. The output capacity of each phase of this system is 10kVA, the three phases are balanced, and the angle between the two phases is 120 degrees. The equipment can set the ratio difference of the secondary side of the built-in CVT and obtain the secondary side voltage data of the corresponding ratio difference.

[0138] In the experiment, the ratio difference of one phase is set to gradually increase, while the ratio difference of the other two phases fluctuates within the normal range, and an experiment is carried out on phases A, B, and C. The primary side rated voltage and secondary side rated voltage of the CVT in the system are 110kV and 57.7V respectively, and the three-phase signal frequency is 50Hz.

[0139] The gradual degradation experiment was carried out by gradually increasing the ratio difference. During the degradation experiment, the initial ratio difference of phase A was -0.028%, and the final ratio difference was 0.63%. The initial ratio difference and final ratio difference of phase B were 0.127% and 0.789%, respectively. The initial ratio difference and final ratio difference of phase C were -0.028% and 0.627%, respectively. Due to the adjustment limitation of the online control system, a maximum of 36 samples with different ratio differences can be collected in this interval. Each sample collects 1 data point per second for a total of 2 minutes, so each sample point includes 120 data points. A total of 4320 data points were collected in this experiment.

[0140] The process of gradually increasing the ratio difference corresponding to the first ten moments of the three phases A, B, and C collected in the experiment is shown in Table 1. The three-phase voltage data collected in the three groups of degradation experiments are shown in Table 1. Figure 4a to Figure 4c As shown in the figure, it can be clearly observed that the voltage amplitude of the phase where the CVT deteriorates shows an increasing trend as the error of the CVT increases.

[0141] Table 1 Gradual degradation process of each phase

[0142] Collection time Experiment 1 (Phase A) Experiment 2 (Phase B) Experiment 3 (Phase C) 1 -0.028％ 0.127％ -0.028％ 2 -0.016％ 0.140％ -0.017％ 3 -0.004％ 0.161％ -0.003％ 4 0.024％ 0.181％ 0.015％ 5 0.049％ 0.208％ 0.05％ 6 0.071％ 0.238％ 0.072％ 7 0.082％ 0.249％ 0.077％ 8 0.093％ 0.270％ 0.09％ 9 0.114％ 0.292％ 0.111％ 10 0.132％ 0.317％ 0.132％

[0143] The 36 measured data are evenly divided into training data and test data, that is, 18 training data and 18 test data. Then, using formula (6), the SOP degradation trends of CVT A, B, and C phases are calculated as follows: Figure 5 As shown. Figure 5 In the figure, as the ratio difference increases, the SOP of CVT shows a downward trend. Therefore, this patent proposes to use wavelet transform to first calculate the maximum approximation coefficient, and then calculate the SOP calculation method to reflect the process of CVT's metering performance degradation.

[0144] VMD-MD was used to perform decomposition, and the threshold was set to 1×10 -4 The decomposition calculation is carried out on it, and the threshold range is exceeded when it is decomposed to the seventh layer. Therefore, it is calculated that the SOP of the three phases A, B and C need to be decomposed to the sixth layer.

[0145] Table 2 Average difference of VMD-MD decomposition residuals corresponding to different h values

[0146]

[0147] The VMD-MD algorithm is used to decompose the SOP of the three phases of CVT into six layers, and the following are obtained: Figure 6a to Figure 6c The decomposition result.

[0148] This patent uses double Gaussian fitting combined with the Levenberg–Marquardt (LM) algorithm to optimize the parameters in its model, thereby improving its fitting effect on the residual term. For each phase degradation object, the parameters of the double Gaussian fitting model are determined as shown in Table 3. The RMSE, MAE and R between the final prediction results and the true value are shown in Table 3. 2 As shown in Table 4, the final prediction results are Figure 7a to Figure 7c From the experimental results in Table 4 and Figure 8, it can be clearly seen that RMSE and MAE are small, and R 2 Greater than 0.99, both close to 1, indicating that the double Gaussian fitting model is used in the task of predicting the SOP residual term. The double Gaussian model has a more detailed signal decomposition capability, that is, decomposing the signal into two Gaussian distribution signals. It can accurately capture the main degradation trend of the CVT over time, which is of great significance for the overall prediction of the subsequent CVT degradation performance indicator SOP.

[0149] Table 3 Double Gaussian model parameter determination

[0150]

[0151] Table 4 Prediction results of double Gaussian model for SOP decomposition trend term

[0152] RMSE MAE <![CDATA[R 2 ]]> Phase A 0.0040769 0.002798 0.99521 Phase B 0.0055827 0.0038658 0.99048 Phase C 0.0033573 0.002981 0.99711

[0153] KAN is used to predict the IMF of SOP, that is, the trend of the fluctuation signal, and compared with LSTM, GRU and SVR. The final prediction results of each signal are as follows: Figure 8a to Figure 8e , Figure 9a to Figure 9e and Figure 10a to Figure 10e The performance indicators of each model are shown in Table 5, Table 6 and Table 7. The IMF signal of SOP changes rapidly and irregularly, which increases the difficulty of prediction. However, KAN adds a "side learning" working mode to the connection part of each layer of neurons, and realizes automatic selection of activation functions through B-spline, which enhances the prediction ability of the KAN model in complex situations. By observing the experimental results Figure 8a to Figure 8e , Figure 9a to Figure 9e and Figure 10a to Figure 10eAs can be clearly seen from the performance indicators in Tables 5, 6 and 7, in the prediction task of the SOP decomposition signal, the overall prediction accuracy of KAN is relatively ideal. For example, in the A-phase IMF prediction process, the IMF1, IMF3 and IMF5 signals are relatively complex. In addition, the degradation experiment in this patent has only 18 training samples and 18 test samples, which is a small number. Through the final comparison results of each model, it can be clearly found that for the A-phase SOP decomposition signal IMF3, the RMSE corresponding to the final prediction result of KAN is 0.0039267 and the MAE is 0.0032691. Compared with SVR, LSTM and GRU, RMSE decreased by 0.0012371, 0.0009206, and 0.0006357, respectively, and MAE decreased by 0.0009502, 0.0012599 and 0.0009506, respectively, which further verifies the superiority and applicability of KAN in this patent. Experiments have shown that although the superiority of LSTM and GRU in time prediction sequences has been verified through a large number of studies. However, most of the SOP decomposition signals have a large frequency of change, complex changes and a small sample size, resulting in the model not being fully trained, thus limiting its practicality in the SOP prediction task of this patent. In addition, the superiority of SVR in small sample scenarios has also been proven, but its data static model fails to capture the correlation characteristics of data in the time series space, and its prediction performance in time series prediction is poor. For scenarios with small data volumes and more complex data changes, compared to the above-mentioned comparison models, KAN has improved the robustness and flexibility of the prediction model due to its learnable activation function, and can obtain more accurate prediction results in more complex decomposition signals.

[0154] Table 5 Comparison results of prediction of IMF signal of phase A SOP decomposition

[0155]

[0156] Table 6 Comparison results of prediction of IMF signal of phase B SOP decomposition

[0157]

[0158]

[0159] Table 7 Comparison results of SOP decomposition IMF signal prediction for phase C

[0160]

[0161] The prediction results of the residual term based on the double Gaussian model and the prediction results of the other decomposition signals based on multiple prediction models are superimposed to obtain the overall SOP prediction results of the CVT degradation degree, such as Figures 11a to 11cThe prediction performance comparison indicators of the relevant models are shown in Table 8. Among the three groups of experiments in this patent, the prediction effect of the double Gaussian model combined with KAN is the best. The following will be explained in detail in combination with the results of the A phase degradation experiment. The RMSE, MAE and R corresponding to KAN in this part 2 They are 0.00863, 0.0066781, and 0.98108, respectively. Compared with the KAN combined with the double Gaussian fitting model, the RMSE corresponding to the final experimental results of the SVR, LSTM, and GRU combined with the double Gaussian fitting model are reduced by 0.002238, 0.00347, and 0.003149, respectively, and the MAE is reduced by 0.0017093, 0.0034869, and 0.0021796, respectively. 2 The increases were 0.01108, 0.01828, and 0.01633, respectively. Combined with the overall experimental results of SOP, it further proves that the KAN model determines the activation function by "learning on the side", which can effectively capture the key features in the data, be robust to noise and outliers when processing complex data sets, and perform well in small sample data tasks.

[0162] Table 8 Comparison results of the overall prediction results of the three-phase SOP

[0163]

[0164]

[0165] Therefore, the present invention first performs a wavelet transform on the measured CVT secondary side voltage. Approximate coefficients and detail coefficients are formed after multi-layer wavelet decomposition. The maximum approximate coefficient is selected from the approximate coefficients, and the maximum approximate coefficient is used to calculate the metrological performance state of the CVT to characterize the metrological error. Subsequently, a variational mode decomposition-average difference method is proposed to perform multi-layer decomposition on the SOP of the CVT. The decomposed residual term is used to characterize the degradation trend part of the SOP, and a double Gaussian model is used for modeling and prediction; the intrinsic mode function of the SOP is decomposed to characterize the degradation fluctuation part of the SOP, and the KAN network algorithm is used for prediction. Finally, all the prediction results are accumulated to obtain a predicted value to characterize the SOP change trend.

[0166] Exemplary Devices

[0167] Fig.12 FIG. 1 is a schematic diagram of a structure of a device for predicting a metering degradation trend of a capacitor voltage transformer provided by an exemplary embodiment of the present invention. Fig.12 As shown, the apparatus 1200 includes:

[0168] The test module 1210 is used to perform a gradual degradation test on the capacitor voltage transformer by gradually increasing a preset ratio difference, and obtain secondary side voltage time series sample data;

[0169] A calculation module 1220, configured to calculate a performance status index sample of the capacitor voltage transformer according to the secondary side voltage time series sample data by using a wavelet transform method;

[0170] A decomposition module 1230 is used to decompose the performance status indicator sample by using a variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data;

[0171] The training module 1240 is used to train a double Gaussian model using the performance state degradation trend sample data to obtain a degradation trend prediction model, and to train a KAN algorithm using the performance state fluctuation sample data to obtain a fluctuation prediction model;

[0172] A prediction module 1250 is used to use the degradation trend prediction model and the fluctuation prediction model to respectively perform degradation trend and fluctuation prediction on the acquired historical sequence data of the capacitor voltage transformer to be tested, so as to obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested;

[0173] The determination module 1260 is used to determine the degradation trend prediction result of the capacitor voltage transformer to be tested according to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested.

[0174] Optionally, the expression of the preset ratio difference is:

[0175]

[0176] Where V s is the secondary voltage amplitude, V f is the primary voltage amplitude, and k is the rated transformation ratio of the capacitive voltage transformer.

[0177] Optionally, the calculation module 1220 includes:

[0178] A decomposition submodule is used to decompose the secondary side voltage time series sample data using wavelet transform to obtain an approximate coefficient sequence;

[0179] The calculation submodule is used to calculate the performance status index sample of the capacitor voltage transformer according to the maximum approximate coefficient in the approximate coefficient sequence.

[0180] Optionally, the calculation formula of the performance status indicator sample is:

[0181]

[0182] Where, MAC iIt represents the maximum approximation coefficient of the calculated test data, and MAC0 represents the maximum approximation coefficient of the normal state.

[0183] Optionally, the decomposition module 1230 includes:

[0184] A construction submodule, used to construct a variational subject according to the performance status indicator samples using the VMD algorithm;

[0185] The solving submodule is used to solve the variational subject by referencing the average difference, and obtain the performance state degradation trend sample data and the performance state fluctuation sample data.

[0186] Optionally, construct a submodule, including:

[0187] A first calculation unit is used to calculate the one-sided spectrum of each mode by Hilbert transform;

[0188] A modulation unit, used for modulating the spectrum of each mode to a response frequency range based on the unilateral spectrum of each mode;

[0189] The construction unit is used to construct a variational subject of unconstrained variation based on the response frequency range of each mode and the performance status indicator sample.

[0190] Alternatively, the expression of the variational body of the unconstrained variation is:

[0191]

[0192] Where k is the number of eigenmode components, {μ k}={μ1,...,μ K} represents K modal components; {ω k}={ω1,...,ω K} represents the frequency center of the modal component; represents the partial derivative with respect to t, K is the number of modal components, λ is the Lagrange multiplier, α is the quadratic penalty factor, δ(t) is the impulse function, t is the time index; j is the imaginary unit, f(t) is the sum of all modal components, L() is the unconstrained variational modal constraint model, ||*|| 2 is the L2 norm, ∑ K () is the summation symbol.

[0193] Optionally, solve submodules, including:

[0194] The second calculation unit is used to, when the number of decomposition layers of the variational subject is 1, perform calculations based on the current residual and the original performance status indicator sample to determine the current average difference;

[0195] The solving unit is used to compare the average difference with a preset threshold. When the average difference is not less than the preset threshold, the performance status indicator sample is decomposed into the next layer and the average difference of the next layer is calculated. When the average difference is less than the preset threshold, the decomposition is stopped to obtain the performance status degradation trend sample data and the performance status fluctuation sample data.

[0196] Exemplary Electronic Devices

[0197] Fig.13 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. Fig.13 As shown, the electronic device 130 includes one or more processors 131 and a memory 132 .

[0198] The processor 131 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0199] The memory 132 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 131 may run the program instructions to implement the methods of the software programs of the various embodiments of the present invention described above and / or other desired functions. In one example, the electronic device may also include: an input device 133 and an output device 134, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0200] In addition, the input device 133 may also include, for example, a keyboard, a mouse, etc.

[0201] The output device 134 can output various information to the outside, and can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto.

[0202] Of course, to simplify, Fig.13 Only some of the components related to the present invention in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, the electronic device may further include any other appropriate components according to specific application conditions.

[0203] Exemplary computer program products and computer-readable storage media

[0204] In addition to the above-mentioned methods and devices, an embodiment of the present invention may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present invention described in the above-mentioned "Exemplary Method" section of this specification.

[0205] The computer program product may be written in any combination of one or more programming languages ​​to write program code for performing the operations of the embodiments of the present invention, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0206] In addition, an embodiment of the present invention may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present invention described in the above “Exemplary Method” section of this specification.

[0207] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, system or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0208] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details disclosed above are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0209] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0210] The block diagrams of the devices, systems, equipment, and systems involved in the present invention are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, systems, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The word "such as" used here refers to the phrase "such as but not limited to", and can be used interchangeably with it.

[0211] The method and system of the present invention may be implemented in many ways. For example, the method and system of the present invention may be implemented by software, hardware, firmware or any combination of software, hardware, firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present invention are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present invention may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present invention. Thus, the present invention also covers a recording medium storing a program for executing the method according to the present invention.

[0212] It should also be noted that in the system, device and method of the present invention, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. The above description of the disclosed aspects is provided to enable any technician in the field to make or use the present invention. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined here can be applied to other aspects without departing from the scope of the present invention. Therefore, the present invention is not intended to be limited to the aspects shown here, but in accordance with the widest range consistent with the principles and novel features disclosed here.

[0213] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present invention to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for predicting the metrological degradation trend of a capacitive voltage transformer, characterized in that: include: A gradual degradation test is performed on the capacitor voltage transformer by gradually increasing the preset ratio difference to obtain secondary side voltage time series sample data; Calculating the performance status index sample of the capacitive voltage transformer according to the secondary side voltage time series sample data by using a wavelet transform method; Decomposing the performance status indicator samples by using variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data; Using the performance state degradation trend sample data to train a double Gaussian model to obtain a degradation trend prediction model, and using the performance state fluctuation sample data to train a KAN algorithm to obtain a fluctuation prediction model; Using the degradation trend prediction model and the fluctuation prediction model to respectively perform degradation trend and fluctuation prediction on the acquired historical sequence data of the capacitor voltage transformer to be tested, to obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested; A degradation trend prediction result of the capacitor voltage transformer to be tested is determined according to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested.

2. The method according to claim 1, characterized in that The expression of the preset ratio difference is: Where V s is the secondary voltage amplitude, V f is the primary voltage amplitude, and k is the rated transformation ratio of the capacitive voltage transformer.

3. The method according to claim 1, characterized in that Calculating the performance status index sample of the capacitor voltage transformer according to the secondary side voltage time series sample data by using the wavelet transform method, including: Decomposing the secondary side voltage time series sample data by wavelet transform to obtain an approximate coefficient sequence; The performance status indicator sample of the capacitive voltage transformer is calculated according to the maximum approximation coefficient in the approximation coefficient sequence.

4. The method according to claim 3, characterized in that: The calculation formula of the performance status indicator sample is: In the formula, MAC i It represents the maximum approximation coefficient of the calculated test data, and MAC0 represents the maximum approximation coefficient of the normal state.

5. The method according to claim 1, characterized in that The performance status indicator samples are decomposed by using the variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data, including: Using VMD algorithm to construct a variational subject according to the performance status indicator sample; The variational subject is solved by quoting the average difference to obtain the performance state degradation trend sample data and the performance state fluctuation sample data.

6. The method according to claim 5, characterized in that The VMD algorithm is used to construct a variational subject according to the performance status indicator sample, including: The one-sided spectrum of each mode is calculated by Hilbert transform; Based on the unilateral spectrum of each mode, modulate the spectrum of each mode to a response frequency range; Based on the response frequency range of each mode and the performance status indicator sample, a variational subject of unconstrained variation is constructed.

7. The method according to claim 6, characterized in that The expression of the variational body of the unconstrained variation is: Where k is the number of eigenmode components, {μ k }={μ1,...,μ K } represents K modal components; {ω k }={ω1,...,ω K } represents the frequency center of the modal component; represents the partial derivative with respect to t, K is the number of modal components, λ is the Lagrange multiplier, α is the quadratic penalty factor, δ(t) is the impulse function, t is the time index; j is the imaginary unit, f(t) is the sum of all modal components, L() is the unconstrained variational modal constraint model, ||*|| 2 is the L2 norm, ∑ K () is the summation symbol.

8. The method according to claim 5, characterized in that The variational subject is solved by quoting the average difference to obtain the performance state degradation trend sample data and the performance state fluctuation sample data, including: When the number of decomposition layers of the variational subject is 1, the current average difference is determined based on the current residual and the original performance status indicator sample; The average difference is compared with a preset threshold. When the average difference is not less than the preset threshold, the performance status indicator sample is decomposed into the next layer and the average difference of the next layer is calculated. When the average difference is less than the preset threshold, the decomposition is stopped to obtain the performance status degradation trend sample data and the performance status fluctuation sample data.

9. A device for predicting the measurement degradation trend of a capacitive voltage transformer, characterized in that: include: A test module, used for performing a gradual degradation test on the capacitor voltage transformer by gradually increasing a preset ratio difference, and obtaining secondary side voltage time series sample data; A calculation module, used for calculating the performance status index sample of the capacitive voltage transformer according to the secondary side voltage time series sample data by using a wavelet transform method; A decomposition module, used to decompose the performance status indicator sample by using a variational mode decomposition-mean difference method to obtain performance status degradation trend sample data and performance status fluctuation sample data; A training module, used to train a double Gaussian model using the performance state degradation trend sample data to obtain a degradation trend prediction model, and to train a KAN algorithm using the performance state fluctuation sample data to obtain a fluctuation prediction model; A prediction module, used to use the degradation trend prediction model and the fluctuation prediction model to respectively perform degradation trend and fluctuation prediction on the acquired historical sequence data of the capacitor voltage transformer to be tested, and obtain a degradation trend prediction sequence and a fluctuation prediction sequence of the capacitor voltage transformer to be tested; A determination module is used to determine a degradation trend prediction result of the capacitor voltage transformer to be tested according to the degradation trend prediction sequence and the fluctuation prediction sequence of the capacitor voltage transformer to be tested.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1-8.

Images