Nonlinear correction method, device and equipment for excitation characteristics of mutual inductor

Abstract

The invention discloses a nonlinear correction method, device and equipment for excitation characteristics of a mutual inductor, and particularly relates to the technical field of mutual inductors. The method comprises the following steps: acquiring original data corresponding to a mutual inductor, wherein the original data comprises an original current signal, historical magnetic flux and distributed temperature data; performing noise reduction processing on the original current signal to obtain a purified current signal, and determining a magnetic conductivity correction coefficient and a temperature gradient according to the distributed temperature data; inputting the purification current signal and the magnetic conductivity correction coefficient into a hysteresis model to obtain a real-time magnetic flux, and inputting the purification current signal, the distributed temperature data, the temperature gradient and the historical magnetic flux into a magnetic flux correction network to obtain a magnetic flux correction value; and compensating the real-time magnetic flux according to the magnetic flux correction value to obtain a corrected magnetic flux result of the mutual inductor. According to the invention, multi-physical field fusion input can be realized, transient response can be improved, harmonic distortion compensation can be carried out, and high precision of data can be maintained.

Description

TECHNICAL FIELD

[0001] The embodiment of the present application relates to the technical field of mutual inductor, and particularly relates to a nonlinear correction method, device and equipment for excitation characteristics of a mutual inductor. BACKGROUND

[0002] With the rapid development of smart grid and industrial internet of things, high-precision current measurement has become a key foundation for power system state monitoring and fault diagnosis. The traditional current transformer has obvious nonlinear problems in excitation characteristics under complex working conditions, resulting in a decrease in measurement accuracy. For example, the rapid fluctuation of current caused by new energy grid connection (such as cloud shading transient of photovoltaic power station) can make the core working point frequently cross the nonlinear region, but the existing correction method based on the fixed parameter hysteresis model is difficult to adapt to such dynamic changes, and there is a large measured ratio error. In addition, when the power distribution equipment of the mutual inductor operates in a wide temperature range of-40℃ to 150℃, the silicon steel sheet permeability presents significant nonlinear temperature drift, and the traditional temperature compensation adopts single-point correction or linear coefficient. However, under the condition of non-uniform heating, the compensation residual will be further expanded. At the same time, the popularization of power electronic equipment leads to excessive current THD, which can cause additional harmonic excitation of the core and cannot meet the requirements of protection devices on rapidity.

[0003] Therefore, there is an urgent need for a new nonlinear correction method for excitation characteristics of a mutual inductor to solve the technical problems in the prior art. SUMMARY

[0004] The purpose of the present application is to at least solve one of the above technical defects.

[0005] In one aspect, the embodiment of the present application provides a nonlinear correction method for excitation characteristics of a mutual inductor, which comprises:

[0006] Obtaining original data corresponding to the mutual inductor, the original data including an original current signal, historical magnetic flux and distributed temperature data;

[0007] Performing noise reduction processing on the original current signal to obtain a purified current signal, and determining a permeability correction coefficient and a temperature gradient according to the distributed temperature data;

[0008] Inputting the purified current signal and the permeability correction coefficient into a hysteresis model to obtain real-time magnetic flux, and inputting the purified current signal, the distributed temperature data, the temperature gradient and the historical magnetic flux into a magnetic flux correction network to obtain a magnetic flux correction amount;

[0009] Compensating the real-time magnetic flux according to the magnetic flux correction amount to obtain a corrected magnetic flux result of the mutual inductor.

[0010] Optionally, the noise reduction processing on the original current signal to obtain the purified current signal comprises:

[0011] determine a real-time fundamental frequency according to the original current signal, and determine a wavelet decomposition layer number according to the real-time fundamental frequency;

[0012] wavelet-decompose the original current signal according to the wavelet decomposition layer number, to obtain a current signal of each layer, and perform double-threshold denoising processing on the current signal of each layer to obtain a purified current signal;

[0013] The double threshold value includes a first threshold value and a second threshold value, and the value of the second threshold value is twice the value of the first threshold value.

[0014] Optionally, the magnetic permeability correction coefficient is determined according to the distributed temperature data, including:

[0015] The distributed temperature data is subjected to finite element difference processing to obtain difference-processed distributed temperature data;

[0016] A preset standard temperature threshold value is obtained, and a temperature difference between the preset standard temperature threshold value and the difference-processed distributed temperature data is determined;

[0017] The temperature difference is input into a temperature drift compensation model to obtain the magnetic permeability correction coefficient.

[0018] Optionally, the purified current signal and the magnetic permeability correction coefficient are input into a hysteresis model to obtain a real-time magnetic flux, including:

[0019] A real-time magnetic field strength is determined according to the purified current signal, and a pre-stored model parameter is obtained;

[0020] The real-time magnetic field strength is iteratively solved according to the pre-stored model parameter until an iteration condition is met, to obtain a non-hysteresis magnetization strength;

[0021] The non-hysteresis magnetization strength is solved according to a fourth-order Runge-Kutta solving equation including a temperature dynamic term, to obtain a magnetization strength;

[0022] The real-time magnetic flux is obtained based on the magnetic permeability correction coefficient, the non-hysteresis magnetization strength and the magnetization strength.

[0023] Optionally, the magnetic flux correction network includes an input layer, a separable convolution layer, a bidirectional gate recurrent unit, a residual connection module and an output layer, the purified current signal, the distributed temperature data, the temperature gradient and the historical magnetic flux are input into the magnetic flux correction network to obtain a magnetic flux correction amount, including:

[0024] The first-order difference processing is performed on the historical magnetic flux to obtain a magnetic flux change amount, and the purification current signal, the distributed temperature data, the temperature gradient and the historical magnetic flux change amount are normalized based on an input layer to obtain a normalized feature vector, the normalized feature vector including the normalized purification current signal, the normalized distributed temperature data, the normalized temperature gradient and the normalized historical magnetic flux change amount;

[0025] The normalized feature vector is input into a separable convolution layer for feature convolution processing to obtain a convolution feature vector;

[0026] The convolution feature vector is input into a bidirectional gated recurrent unit for capturing time sequence dependency processing to obtain a time sequence feature vector;

[0027] The time sequence feature vector is connected based on a residual connection module to obtain a connected feature vector, and the connected feature vector is input into an output layer to obtain a magnetic flux correction amount after global average pooling processing of the output layer.

[0028] Optionally, the real-time magnetic flux is compensated based on the magnetic flux correction amount to obtain a corrected magnetic flux result of the transformer, including:

[0029] The magnetic flux correction amount and the real-time magnetic flux are superimposed to obtain the corrected magnetic flux result of the transformer.

[0030] Optionally, the method further includes:

[0031] At least one real-time monitoring index is obtained, and a relationship between each real-time monitoring index and a corresponding threshold value is determined, the real-time monitoring index including a magnetic flux residual absolute value, a temperature gradient and a current harmonic distortion rate;

[0032] If any one of the real-time monitoring indexes is greater than the corresponding threshold value, the network parameters of the magnetic flux correction network are adjusted.

[0033] Optionally, the output layer includes a global pooling layer and a fully connected layer, and the network parameters of the magnetic flux correction network are adjusted, including:

[0034] The separable convolution layer and the residual connection module in the magnetic flux correction network are completely prohibited from being used;

[0035] The specified parameters in the bidirectional gated recurrent unit and the global pooling layer are adjusted according to a preset adjustment mode to obtain an adjusted bidirectional gated recurrent unit and global pooling layer;

[0036] All network parameters in the fully connected layer are adjusted to obtain an adjusted fully connected layer.

[0037] In another aspect, the embodiment of the present application provides a nonlinear correction device for excitation characteristics of a mutual inductor, comprising:

[0038] A data acquisition module is configured to acquire original data corresponding to the mutual inductor, wherein the original data comprises an original current signal, historical magnetic flux and distributed temperature data.

[0039] A data determination module is configured to perform noise reduction processing on the original current signal to obtain a purified current signal, and determine a magnetic permeability correction coefficient and a temperature gradient based on the distributed temperature data.

[0040] A correction amount determination module is configured to input the purified current signal and the magnetic permeability correction coefficient into a hysteresis model to obtain real-time magnetic flux, and input the purified current signal, the distributed temperature data, the temperature gradient and the historical magnetic flux into a magnetic flux correction network to obtain a magnetic flux correction amount.

[0041] A magnetic flux correction module is configured to compensate the real-time magnetic flux based on the magnetic flux correction amount to obtain a corrected magnetic flux result of the mutual inductor.

[0042] In another aspect, the embodiment of the present application provides an electronic device, comprising a processor and a memory:

[0043] The memory is configured to store machine-readable instructions, and the instructions, when executed by the processor, cause the processor to perform any one of the methods for nonlinear correction of excitation characteristics of a mutual inductor.

[0044] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects:

[0045] In the present application, the original current signal, the historical magnetic flux and the distributed temperature data of the mutual inductor can be acquired, and then the magnetic flux compensation amount is determined based on the hysteresis model and the magnetic flux correction network to correct the real-time magnetic flux of the mutual inductor. It can be seen that the magnetic flux correction network in the present application can simultaneously process the coupling effects of current amplitude, temperature field and historical magnetic flux trend, realize multi-physical field fusion input, and greatly reduce the step response time compared with the prior art, improve the transient response, and at the same time can compensate for harmonic distortion and maintain high precision of data.

[0046] In the embodiment of the present application, the noise reduction processing is performed by a double-threshold method, which can ensure linear transition of the signal between the double thresholds, avoid step distortion of the hard threshold, and preserve the true amplitude of the signal, effectively avoiding the influence of Gaussian white noise and pulse interference. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0048] Figure 1 A flowchart of a nonlinear correction method for the excitation characteristic of a transformer provided by the embodiments of the present application;

[0049] Figure 2 A wavelet decomposition diagram provided by the embodiments of the present application;

[0050] Figure 3 A specific connection circuit diagram of a current transformer provided by the embodiments of the present application;

[0051] Figure 4 A structure diagram of a nonlinear correction device for the excitation characteristic of a transformer provided by the embodiments of the present application;

[0052] Figure 5 A structure diagram of an electronic device provided by the embodiments of the present application. DETAILED DESCRIPTION

[0053] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the drawings, in which the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present application, and cannot be interpreted as a limitation of the present application.

[0054] Those skilled in the art can understand that, unless specifically stated, the singular forms "a", "an" and "the" used herein also include the plural forms. It should be further understood that the phrase "comprising" used in the specification of the present application means that the features, integers, steps, operations, elements and / or components exist, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to the other element, or there can be intermediate elements. In addition, "connected" or "coupled" used herein can include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any single unit and all combinations of the associated listed items.

[0055] In order to make the purposes, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0056] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0057] Specifically, such as Figure 1 As shown, the method may include:

[0058] Step S101: Obtain the raw data corresponding to the current transformer. The raw data includes the raw current signal, historical magnetic flux, and distributed temperature data.

[0059] Optionally, the current transformer can be configured with a multi-source sensor unit, which can acquire the raw data of the current transformer. This raw data specifically includes the raw current signal, historical magnetic flux, and distributed temperature data of the current transformer. The raw current signal can be the instantaneous current signal I(t) acquired by the current sampling module; the distributed temperature data is the spatial temperature field data T(x,y) output by the distributed temperature sensing array embedded on the surface of the iron core; and the historical magnetic flux is the magnetic flux value Φ(t-Δt) of the previous cycle stored in the historical state memory. The specific duration of each cycle can be set according to actual needs, such as 3 days or a week, etc., and this embodiment does not limit this.

[0060] Step S102: The original current signal is denoised to obtain a purified current signal, and the permeability correction coefficient and temperature gradient are determined based on the distributed temperature data.

[0061] Optionally, since the instantaneous current signal acquired by the current sampling module may be subject to interference from other signals, noise reduction processing can be performed on the original current signal after acquisition to obtain a purified current signal. This purified current signal has eliminated interference from other signals, ensuring a purer and more accurate signal. Optionally, while processing the original current signal, the permeability correction coefficient and temperature gradient can also be determined based on distributed temperature data. This can improve resource utilization and increase data processing speed.

[0062] In an optional embodiment of this application, noise reduction processing is performed on the original current signal to obtain a purified current signal, including:

[0063] The real-time fundamental frequency is determined based on the original current signal, and the number of wavelet decomposition layers is determined based on the real-time fundamental frequency.

[0064] The original current signal is decomposed into wavelet decomposition based on the number of wavelet decomposition levels to obtain the current signal of each level. Then, the current signal of each level is subjected to double threshold noise reduction processing to obtain the purified current limiting signal.

[0065] The dual thresholds include a first threshold and a second threshold, with the second threshold being twice the value of the first threshold.

[0066] Optionally, in existing technologies, the current signal is decomposed into a fixed number of layers (e.g., 5 layers). In this case, at a high fundamental frequency, high-frequency details are excessively compressed, leading to high-frequency signal distortion. Conversely, at a low fundamental frequency, the effective signal bandwidth is not fully covered, resulting in residual low-frequency noise. Based on this, in the embodiments of this application, the real-time fundamental frequency can be determined based on the original current signal using the following formula:

[0067]

[0068] Where f0 is the real-time fundamental frequency, f is the candidate frequency, I(t) is the original current signal, and f s Where N is the sampling frequency and N is the analysis window length.

[0069] Furthermore, the number of wavelet decomposition levels can be determined based on the determined real-time fundamental frequency using the following formula:

[0070]

[0071] Where L is the number of decomposition layers, the coefficient 2.5 is the optimized value determined experimentally to balance bandwidth resolution and computational load, f0 is the real-time fundamental frequency, and f s The sampling frequency.

[0072] Furthermore, such as Figure 2 As shown, the original current signal can be decomposed into a defined number of decomposition levels L using wavelet decomposition, yielding the final current signal, specifically including high-frequency details DL and low-frequency approximation AL. Then, noise reduction processing is performed on the current signal of each level to obtain a purified current-limiting signal. However, existing technologies typically use a single threshold (hard threshold or soft threshold). Research has found that using a hard threshold truncates the signal at λ, which can easily lead to signal oscillation, while using a soft threshold results in overall compression, weakening the effective signal amplitude.

[0073] Based on this, this application employs a dual-threshold method to denoise each layer of the decomposed signal, obtaining a purified current-limiting signal. Specifically, this can be determined using the following formula:

[0074]

[0075] Where, d j For the j-th layer current signal, The purified current signal is obtained after noise reduction, where λ1 is the first threshold, λ2 is the second threshold, and λ2 = 2λ1. MAD is the median absolute deviation of the detail coefficients.

[0076] In the embodiments of this application, noise reduction is performed using a dual-threshold method, which can ensure that the signal transitions linearly between the two thresholds, avoid step distortion caused by hard thresholds, and preserve the authenticity of the signal amplitude, effectively avoiding the effects of Gaussian white noise and impulse interference.

[0077] In an optional embodiment of this application, determining the permeability correction coefficient based on distributed temperature data includes:

[0078] The distributed temperature data is processed by finite element interpolation to obtain the interpolated distributed temperature data.

[0079] The temperature difference is input into the temperature drift compensation model to obtain the permeability correction coefficient.

[0080] Optionally, the distributed temperature data is obtained based on the reflected wavelength conversion of each fiber Bragg grating sensor. This results in distributed temperature data containing multiple temperature values. Outlier data points can be removed using Grubbs spectroscopy to obtain the removed distributed temperature data. The discrete temperature values ​​in the removed distributed temperature data are then mapped to a two-dimensional mesh of the iron core for finite element interpolation processing to obtain the differentiated distributed temperature data. A preset standard temperature threshold (e.g., 25℃, denoted as Tref=25℃) is then obtained, and the temperature difference between the preset standard temperature threshold and the differentiated distributed temperature data is determined. Finally, the temperature difference is input into the temperature drift compensation model to obtain the permeability correction coefficient. The temperature drift compensation model is determined using the following formula:

[0081] μ corr (x,y)=μ0[1+β1ΔT(x,y)+β2(ΔT(x,y)) 2 ]

[0082] Where ΔT(x,y) is the temperature difference, β and β2 are material coefficients, β1=0.0023, β2=1.7·10 -6 μ corr (x,y) is the permeability correction coefficient, and μ0 is the calibration coefficient.

[0083] Step S103: Input the purification current signal and the permeability correction coefficient into the hysteresis model to obtain the real-time magnetic flux, and input the purification current signal, distributed temperature data, temperature gradient and historical magnetic flux into the magnetic flux correction network to obtain the magnetic flux correction amount.

[0084] Optionally, after obtaining the purification current signal and the permeability correction coefficient, they can be input into the hysteresis model. At this time, the real-time magnetic flux can be obtained. Simultaneously, the obtained purification current signal, distributed temperature data, temperature gradient, and historical magnetic flux are input into the magnetic flux correction network, which will output the magnetic flux correction amount.

[0085] In an optional embodiment of this application, the purification current signal and the permeability correction coefficient are input into the hysteresis model to obtain the real-time magnetic flux, including:

[0086] The real-time magnetic field strength is determined based on the purification current signal, and the pre-stored model parameters are obtained.

[0087] The real-time magnetic field strength is iteratively solved based on the pre-stored model parameters until the iteration condition is met, and the hysteresis-free magnetization intensity is obtained.

[0088] The magnetization intensity is obtained by solving the hysteresis-free magnetization intensity using the fourth-order Runge-Kutta equation that includes a temperature dynamic term.

[0089] The real-time magnetic flux is obtained based on the permeability correction coefficient, hysteresis-free magnetization, and magnetization.

[0090] Optionally, the number of coil turns and the effective magnetic path length can be obtained, and then the real-time magnetic field strength can be calculated based on the purification current signal, the number of coil turns, and the effective magnetic path length using existing magnetic field strength calculation formulas. Further, the real-time magnetic field strength can be iteratively solved using model parameters until the iteration conditions are met to obtain the hysteresis-free magnetization. The model parameters are as follows:

[0091]

[0092] Where Man is the hysteresis-free magnetization, Ms is the saturation magnetization, H is the real-time magnetic field strength, α is the molecular field coefficient, a(T) is the temperature-dependent shape parameter, and M is the dynamic variable.

[0093] Furthermore, the model parameters can be iterated using the Newton-Raphson method until a preset iteration condition is met, resulting in hysteresis-free magnetization. The iteration condition can be set according to actual needs, and this embodiment does not impose any limitations on it.

[0094] Correspondingly, a fourth-order Runge-Kutta equation including a temperature dynamic term can be obtained. Then, based on this Runge-Kutta equation, a dynamic hysteresis differential equation is solved for the hysteresis-free magnetization to obtain the magnetization. Finally, based on the permeability correction coefficient, the hysteresis-free magnetization, and the magnetization, the magnetic flux density is calculated using the following formula:

[0095] B(t) = μ0μ corr (H(t)+M(T))

[0096] Where B(t) is the magnetic flux density, μ0 is the calibration coefficient, H(t) is the magnetization, M(T) is the hysteresis-free magnetization, and μ corr This is the permeability correction factor.

[0097] Furthermore, the obtained magnetic flux density can be converted into real-time magnetic flux using existing formulas.

[0098] In an optional embodiment of this application, the magnetic flux correction network includes an input layer, a separable convolutional layer, a bidirectional gated recurrent unit, a residual connection module, and an output layer. The purified current signal, distributed temperature data, temperature gradient, and historical magnetic flux are input to the magnetic flux correction network to obtain the magnetic flux correction value, including:

[0099] The historical magnetic flux is processed by first-order difference to obtain the magnetic flux change. Based on the input layer, the purification current signal, distributed temperature data, temperature gradient and historical magnetic flux change are normalized to obtain a normalized feature vector. The normalized feature vector includes the normalized purification current signal, the normalized distributed temperature data, the normalized temperature gradient and the normalized historical magnetic flux change.

[0100] The normalized feature vector is input into a separable convolutional layer for feature convolution processing to obtain the convolutional feature vector.

[0101] The convolutional feature vector is input into a bidirectional gated recurrent unit to capture temporal dependencies, thus obtaining a temporal feature vector.

[0102] The temporal feature vector is processed by residual connection module to obtain the connected feature vector. The connected feature vector is then input into the output layer so that the output layer can obtain the magnetic flux correction after global average pooling.

[0103] Optionally, the acquired historical magnetic flux can be processed using first-order difference to obtain the flux change, that is, the original historical magnetic flux is converted into a rate of change, thereby enhancing the perception of dynamic characteristics. Further, based on the input layer, the purification current signal, distributed temperature data, temperature gradient, and historical flux change are normalized to obtain the normalized purification current signal, normalized distributed temperature data, normalized temperature gradient, and normalized historical flux change. At this point, the normalized purification current signal represents the current excitation intensity, with a normalization range of [-1,1]; the normalized distributed temperature data represents the overall thermal state influence, with a normalization range of [0,1]; and the normalized historical flux change represents the flux change trend, with a normalization range of [-1,1].

[0104] Furthermore, the obtained normalized feature vector is input into a separable convolutional layer for feature convolution processing to obtain a convolutional feature vector. The kernel size of this separable convolutional layer is 3, the stride is 1, and the padding is 'same', and its computational cost is only 1 / 8 of that of a standard convolution. Then, the convolutional feature vector is input into a bidirectional gated recurrent unit for capturing temporal dependencies to obtain a temporal feature vector.

[0105] Correspondingly, residual connection processing can be performed on temporal features based on residual connection modules to obtain connected feature vectors. The connected feature vectors are then input into the output layer, where the output layer can perform global average pooling on the input feature vectors to obtain the magnetic flux correction.

[0106] In this application, the magnetic flux correction network can simultaneously handle the coupling effect of current amplitude (I), temperature field (T), and historical magnetic flux trend (ΔΦ), realizing multi-physics field fusion input. Compared with the prior art, it can significantly reduce the step response time, improve the transient response, and perform harmonic distortion compensation to maintain high data accuracy.

[0107] Step S104: Compensate the real-time magnetic flux according to the magnetic flux correction amount to obtain the corrected magnetic flux result of the current transformer.

[0108] In an optional embodiment of this application, the real-time magnetic flux is compensated according to the magnetic flux correction amount to obtain the corrected magnetic flux result of the current transformer, including:

[0109] The magnetic flux correction value is superimposed with the real-time magnetic flux to obtain the corrected magnetic flux result of the current transformer.

[0110] Optionally, the magnetic flux correction can be superimposed with the real-time magnetic flux, and the sum can be used as the corrected magnetic flux result of the mutual inductor.

[0111] In an optional embodiment of this application, the method further includes:

[0112] At least one real-time monitoring indicator is acquired, and the relationship between each real-time monitoring indicator and its corresponding threshold is determined. The real-time monitoring indicators include the absolute value of magnetic flux residual, temperature gradient, and current harmonic distortion rate.

[0113] If any real-time monitoring indicator exceeds the corresponding threshold, the network parameters of the magnetic flux correction network will be adjusted.

[0114] Optionally, in this embodiment, at least one real-time indicator of the current transformer can be monitored. This indicator can be the absolute value of the residual, the temperature gradient, or the current harmonic distortion rate. Then, the relationship between each monitored real-time indicator and the corresponding threshold is calculated. If any real-time monitoring indicator is found to be greater than the corresponding threshold, it indicates that there may be some errors in the current magnetic flux correction network. At this time, the network parameters of the magnetic flux correction network can be adjusted.

[0115] In an optional embodiment of this application, the output layer includes a global pooling layer and a fully connected layer, adjusting the network parameters of the magnetic flux correction network, including:

[0116] Completely disable the use of separable convolutional layers and residual connection modules in the magnetic flux correction network;

[0117] The specified parameters in the bidirectional gated loop unit and the global pooling layer are adjusted according to the preset adjustment method to obtain the adjusted bidirectional gated loop unit and the global pooling layer.

[0118] Adjust all network parameters in the fully connected layer to obtain the adjusted fully connected layer.

[0119] Optionally, in this embodiment, when adjusting the network parameters of the magnetic flux correction network, the separable convolutional layer and residual connection module in the magnetic flux correction network can be frozen, i.e., completely disabled. At this time, all parameters of the layer (convolutional kernel weights, biases) remain fixed. Furthermore, the specified parameters in the bidirectional gated recurrent unit and the global pooling layer in the output layer are adjusted according to a preset adjustment method, such as adjusting the gate weights, hidden state weights, and bias terms of the bidirectional gated recurrent unit. This can enhance the response speed to temperature change signals. At the same time, all network parameters in the fully connected layer can be adjusted to obtain the adjusted fully connected layer, thereby quickly compensating for the systematic deviation caused by the temperature gradient.

[0120] Optionally, the current transformer in this application embodiment can be a current transformer. The excitation test unit of the current transformer is developed after in-depth theoretical research based on the traditional current transformer volt-ampere characteristic ratio polarity comprehensive tester based on voltage regulator, voltage booster and current booster. The unit adopts high-performance DSP and FPGA and advanced manufacturing process to ensure stable and reliable product performance, complete functions, high degree of automation and high testing efficiency.

[0121] Optionally, the current transformer excitation test unit in this embodiment uses the principle of low-frequency voltage for boost testing, which can be performed at a frequency lower than the rated frequency, avoiding the winding and secondary terminals from being subjected to unacceptable voltages. The principle of CT excitation characteristic measurement includes opening the primary side and applying voltage from the secondary side. During the test, the relationship curve between the applied voltage V and the input current I can be measured. This curve approximates the relationship curve between the CT's excitation potential E and the excitation current I. Furthermore, an appropriate sinusoidal excitation voltage of the rated frequency is applied to the full-turn terminals of the transformer's secondary winding, with all other terminals open, and the excitation current is measured. For current transformers with selectable secondary winding tap ratios, the excitation characteristics of ratios other than the maximum ratio can be calculated. The specific connection lines of the current transformer are as follows... Figure 3 As shown.

[0122] In this application, the raw current signal, historical magnetic flux, and distributed temperature data of the current transformer can be acquired. Then, based on the hysteresis model and the magnetic flux correction network, the magnetic flux compensation amount is determined to correct the real-time magnetic flux of the current transformer. It is evident that the magnetic flux correction network in this application can simultaneously handle the coupling effect of current amplitude, temperature field, and historical magnetic flux trend, achieving multi-physics field fusion input. Compared to existing technologies, it can significantly reduce the step response time, improve transient response, and simultaneously perform harmonic distortion compensation, maintaining high data accuracy.

[0123] This application provides a nonlinear correction device for the excitation characteristics of a current transformer, such as... Figure 4 As shown, the device may include: a data acquisition module 401, a data determination module 402, a correction amount determination module 403, and a magnetic flux correction module 404, wherein,

[0124] The data acquisition module is used to acquire the raw data corresponding to the current transformer, including raw current signal, historical magnetic flux and distributed temperature data.

[0125] The data determination module is used to perform noise reduction processing on the original current signal to obtain the purified current signal, and to determine the permeability correction coefficient and temperature gradient based on the distributed temperature data.

[0126] The correction amount determination module is used to input the purification current signal and the permeability correction coefficient into the hysteresis model to obtain the real-time magnetic flux, and to input the purification current signal, distributed temperature data, temperature gradient and historical magnetic flux into the magnetic flux correction network to obtain the magnetic flux correction amount.

[0127] The magnetic flux correction module is used to compensate the real-time magnetic flux based on the magnetic flux correction amount, so as to obtain the corrected magnetic flux result of the current transformer.

[0128] Optionally, when the data determination module performs noise reduction processing on the original current signal to obtain the purified current signal, it is specifically used for:

[0129] The real-time fundamental frequency is determined based on the original current signal, and the number of wavelet decomposition layers is determined based on the real-time fundamental frequency.

[0130] The original current signal is decomposed into wavelet decomposition based on the number of wavelet decomposition levels to obtain the current signal of each level. Then, the current signal of each level is subjected to double threshold noise reduction processing to obtain the purified current limiting signal.

[0131] The dual thresholds include a first threshold and a second threshold, with the second threshold being twice the value of the first threshold.

[0132] Optionally, when determining the permeability correction coefficient based on distributed temperature data, the data determination module is specifically used for:

[0133] The distributed temperature data is processed by finite element interpolation to obtain the interpolated distributed temperature data.

[0134] Obtain a preset standard temperature threshold and determine the temperature difference between the preset standard temperature threshold and the distributed temperature data after the difference;

[0135] The temperature difference is input into the temperature drift compensation model to obtain the permeability correction coefficient.

[0136] Optionally, the correction determination module, when inputting the purification current signal and permeability correction coefficient into the hysteresis model to obtain the real-time magnetic flux, is specifically used for:

[0137] The real-time magnetic field strength is determined based on the purification current signal, and the pre-stored model parameters are obtained.

[0138] The real-time magnetic field strength is iteratively solved based on the pre-stored model parameters until the iteration condition is met, and the hysteresis-free magnetization intensity is obtained.

[0139] The magnetization intensity is obtained by solving the hysteresis-free magnetization intensity using the fourth-order Runge-Kutta equation that includes a temperature dynamic term.

[0140] The real-time magnetic flux is obtained based on the permeability correction coefficient, hysteresis-free magnetization, and magnetization.

[0141] Optionally, the magnetic flux correction network includes an input layer, a separable convolutional layer, a bidirectional gated recurrent unit, a residual connection module, and an output layer. The correction amount determination module, when inputting the purified current signal, distributed temperature data, temperature gradient, and historical magnetic flux into the magnetic flux correction network to obtain the magnetic flux correction amount, is specifically used for:

[0142] The historical magnetic flux is processed by first-order difference to obtain the magnetic flux change. Based on the input layer, the purification current signal, distributed temperature data, temperature gradient and historical magnetic flux change are normalized to obtain a normalized feature vector. The normalized feature vector includes the normalized purification current signal, the normalized distributed temperature data, the normalized temperature gradient and the normalized historical magnetic flux change.

[0143] The normalized feature vector is input into a separable convolutional layer for feature convolution processing to obtain the convolutional feature vector.

[0144] The convolutional feature vector is input into a bidirectional gated recurrent unit to capture temporal dependencies, thus obtaining a temporal feature vector.

[0145] The temporal feature vector is processed by residual connection module to obtain the connected feature vector. The connected feature vector is then input into the output layer so that the output layer can obtain the magnetic flux correction after global average pooling.

[0146] Optionally, when the magnetic flux correction module compensates for the real-time magnetic flux based on the magnetic flux correction amount to obtain the corrected magnetic flux result of the current transformer, it is specifically used for:

[0147] The magnetic flux correction value is superimposed with the real-time magnetic flux to obtain the corrected magnetic flux result of the current transformer.

[0148] Optionally, the device also includes a model adjustment module, which is used for:

[0149] At least one real-time monitoring indicator is acquired, and the relationship between each real-time monitoring indicator and its corresponding threshold is determined. The real-time monitoring indicators include the absolute value of magnetic flux residual, temperature gradient, and current harmonic distortion rate.

[0150] If any real-time monitoring indicator exceeds the corresponding threshold, the network parameters of the magnetic flux correction network will be adjusted.

[0151] Optionally, the output layer includes a global pooling layer and a fully connected layer. When adjusting the network parameters of the magnetic flux correction network, the model tuning module is specifically used for:

[0152] Completely disable the use of separable convolutional layers and residual connection modules in the magnetic flux correction network;

[0153] The specified parameters in the bidirectional gated loop unit and the global pooling layer are adjusted according to the preset adjustment method to obtain the adjusted bidirectional gated loop unit and the global pooling layer.

[0154] Adjust all network parameters in the fully connected layer to obtain the adjusted fully connected layer.

[0155] The nonlinear correction device for the excitation characteristics of a current transformer in this embodiment can execute the nonlinear correction method for the excitation characteristics of a current transformer shown in the embodiment of this application. The implementation principle is similar, and will not be described again here.

[0156] This application provides an electronic device, which includes a processor and a memory configured to store machine-readable instructions that, when executed by the processor, cause the processor to perform a nonlinear correction method for the excitation characteristics of a current transformer.

[0157] This application provides an electronic device, such as... Figure 5 As shown, Figure 5 The illustrated electronic device includes a processor 2001 and a memory 2003. The processor 2001 and the memory 2003 are connected, for example, via a bus 2002. Optionally, the electronic device 2000 may further include a transceiver 2004. It should be noted that in practical applications, the transceiver 2004 is not limited to one type, and the structure of this electronic device 2000 does not constitute a limitation on the embodiments of this application.

[0158] Processor 2001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 2001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0159] Bus 2002 may include a pathway for transmitting information between the aforementioned components. Bus 2002 may be a PCI bus or an EISA bus, etc. Bus 2002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0160] The memory 2003 may be ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0161] The memory 2003 stores the application code that executes the scheme of this application, and its execution is controlled by the processor 2001. The processor 2001 executes the application code stored in the memory 2003 to implement... Figure 4 The embodiment shown illustrates the operation of a nonlinear correction device for the excitation characteristics of a current transformer.

[0162] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0163] The above description is only a partial embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for nonlinear correction of the excitation characteristics of a current transformer, characterized in that, include: Acquire the raw data corresponding to the current transformer, including the raw current signal, historical magnetic flux and distributed temperature data; The original current signal is denoised to obtain a purified current signal, and the permeability correction coefficient and temperature gradient are determined based on the distributed temperature data. The purification current signal and the permeability correction coefficient are input into the hysteresis model to obtain the real-time magnetic flux, and the purification current signal, distributed temperature data, temperature gradient and historical magnetic flux are input into the magnetic flux correction network to obtain the magnetic flux correction amount. The real-time magnetic flux is compensated based on the magnetic flux correction amount to obtain the corrected magnetic flux result of the mutual inductor.

2. The method according to claim 1, characterized in that, The noise reduction process for the original current signal to obtain the purified current signal includes: The real-time fundamental frequency is determined based on the original current signal, and the number of wavelet decomposition layers is determined based on the real-time fundamental frequency. The original current signal is decomposed into wavelet decomposition based on the wavelet decomposition level to obtain the current signal of each level, and the current signal of each level is subjected to double threshold noise reduction processing to obtain the purification current limiting signal. The dual thresholds include a first threshold and a second threshold, wherein the value of the second threshold is twice the value of the first threshold.

3. The method according to claim 1, characterized in that, The step of determining the permeability correction coefficient based on the distributed temperature data includes: The distributed temperature data is subjected to finite element interpolation to obtain the interpolated distributed temperature data. Obtain a preset standard temperature threshold and determine the temperature difference between the preset standard temperature threshold and the distributed temperature data after the difference. The temperature difference is input into the temperature drift compensation model to obtain the permeability correction coefficient.

4. The method according to claim 3, characterized in that, The purification current signal and the permeability correction coefficient are input into the hysteresis model to obtain the real-time magnetic flux, including: The real-time magnetic field strength is determined based on the purification current signal, and the pre-stored model parameters are obtained. The real-time magnetic field strength is iteratively solved according to the pre-stored model parameters until the iteration condition is met, and the hysteresis-free magnetization intensity is obtained. The hysteresis-free magnetization is solved by the fourth-order Runge-Kutta equation, which includes a temperature dynamic term, to obtain the magnetization intensity. The real-time magnetic flux is obtained based on the permeability correction coefficient, the hysteresis-free magnetization, and the magnetization.

5. The method according to claim 1, characterized in that, The magnetic flux correction network includes an input layer, a separable convolutional layer, a bidirectional gated recurrent unit, a residual connection module, and an output layer. The process of inputting the purified current signal, distributed temperature data, the temperature gradient, and historical magnetic flux into the magnetic flux correction network to obtain the magnetic flux correction value includes: The historical magnetic flux is subjected to first-order difference processing to obtain the magnetic flux change. Based on the input layer, the purification current signal, the distributed temperature data, the temperature gradient, and the historical magnetic flux change are normalized to obtain a normalized feature vector. The normalized feature vector includes the normalized purification current signal, the normalized distributed temperature data, the normalized temperature gradient, and the normalized historical magnetic flux change. The normalized feature vector is input into the separable convolutional layer for feature convolution processing to obtain the convolutional feature vector. The convolutional feature vector is input into the bidirectional gated recurrent unit to capture temporal dependency processing, thereby obtaining a temporal feature vector; The residual connection module performs residual connection processing on the temporal feature vector to obtain a connected feature vector, and inputs the connected feature vector into the output layer so that the output layer obtains the magnetic flux correction amount after global average pooling processing.

6. The method according to claim 1, characterized in that, The step of compensating the real-time magnetic flux based on the magnetic flux correction to obtain the corrected magnetic flux result of the current transformer includes: The magnetic flux correction value is superimposed with the real-time magnetic flux to obtain the corrected magnetic flux result of the current transformer.

7. The method according to claim 5, characterized in that, The method further includes: At least one real-time monitoring indicator is acquired, and the relationship between each real-time monitoring indicator and its corresponding threshold is determined. The real-time monitoring indicators include the absolute value of magnetic flux residual, temperature gradient, and current harmonic distortion rate. If any of the real-time monitoring indicators is greater than the corresponding threshold, the network parameters of the magnetic flux correction network are adjusted.

8. The method according to claim 7, characterized in that, The output layer includes a global pooling layer and a fully connected layer. Adjusting the network parameters of the magnetic flux correction network includes: Completely disable the separable convolutional layer and the residual connection module in the magnetic flux correction network; The specified parameters in the bidirectional gated loop unit and the global pooling layer are adjusted according to a preset adjustment method to obtain the adjusted bidirectional gated loop unit and global pooling layer; All network parameters in the fully connected layer are adjusted to obtain the adjusted fully connected layer.

9. A nonlinear correction device for the excitation characteristics of a current transformer, characterized in that, include: The data acquisition module is used to acquire the raw data corresponding to the current transformer, including raw current signal, historical magnetic flux and distributed temperature data. The data determination module is used to perform noise reduction processing on the original current signal to obtain a purified current signal, and to determine the permeability correction coefficient and temperature gradient based on the distributed temperature data. The correction amount determination module is used to input the purification current signal and the permeability correction coefficient into the hysteresis model to obtain the real-time magnetic flux, and to input the purification current signal, distributed temperature data, the temperature gradient and the historical magnetic flux into the magnetic flux correction network to obtain the magnetic flux correction amount. The magnetic flux correction module is used to compensate the real-time magnetic flux according to the magnetic flux correction amount, so as to obtain the corrected magnetic flux result of the current transformer.

10. An electronic device, characterized in that, Including the processor and memory: The memory is configured to store a computer program that, when executed by the processor, causes the processor to perform the method according to any one of claims 1-8.

Images