Method and System for Evaluating Error State of Current Transformer Based on Cascade of Multiple CNNs

Abstract

The present invention belongs to the technical field of power monitoring, and discloses a method and system for evaluating the error state of a current transformer based on cascaded multi-level CNNs, including collecting the current time-series data of the historical operation of the current transformer, processing the time-series data to obtain a time-domain image and a frequency-domain image; using a cascaded multi-level CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively to obtain time-domain features and frequency-domain features, fusing the time-domain features and the frequency-domain features to obtain hybrid features; inputting the time-domain features and the frequency-domain features processed by the first-level CNN and the hybrid features processed by the multi-level CNN into an improved LSTM model to obtain the error state of the current transformer. The present invention fuses the time-domain information and the frequency-domain information of the secondary output data of the current transformer to realize the evaluation of the error state with high accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of power monitoring, and particularly to a method and system for evaluating the error state of a current transformer based on cascaded multi-level CNNs. Background Art

[0002] The statements in this section merely provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In a power system, as a key device in the electric energy metering system, the metering performance accuracy of a transformer directly affects the fairness and reliability of electric energy trading. A current transformer (CT), a precision device based on the principle of electromagnetic induction, can convert a high-magnitude primary current into a secondary current that is easy to measure. Its main structure includes a closed iron core and coils wound thereon. However, in the actual operating environment, the metering error of a CT may gradually deviate from the standard range over time due to limitations in the acquisition mechanism and the influence of environmental conditions. This not only requires us to have accurate and rapid fault diagnosis capabilities but also to be able to always grasp the real-time state of the CT metering error, thereby guiding maintenance personnel to reasonably plan maintenance strategies. If the abnormality of the transformer state is not detected in time, it will surely pose a potential threat to the overall operation of the power grid.

[0004] Therefore, ensuring the real-time accuracy of current transformers, reducing electric energy metering losses, and ensuring the stable operation of measurement and control and protection devices have become technical challenges that need to be solved urgently. How to achieve real-time online evaluation of the CT error state undoubtedly constitutes a key problem that the power industry urgently needs to overcome. The existing regular inspection of CT errors has high costs, low efficiency, and poor timeliness, and cannot detect faulty devices in time, seriously affecting the real-time accuracy of electric energy trade settlement. It is urgent to break through the online monitoring technology bottleneck and achieve real-time online inspection of the CT error state. Summary of the Invention

[0005] To solve the above problems, the present invention proposes a method and system for evaluating the error state of a current transformer based on cascaded multi-level CNNs, which fuse the time-domain information and frequency-domain information of the secondary output data of the current transformer to achieve the evaluation of the error state with high accuracy.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a method for evaluating the error state of a current transformer based on cascaded multi-level CNNs, including the following steps:

[0008] Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image;

[0009] The multi - level cascaded CNN algorithm is used to extract features from the time - domain image and the frequency - domain image respectively, obtaining time - domain features and frequency - domain features, and the time - domain features and frequency - domain features are fused to obtain hybrid features;

[0010] The time - domain features and frequency - domain features after the first - level CNN processing and the hybrid features after the multi - level CNN processing are input into the improved LSTM model together to obtain the error state of the current transformer.

[0011] As an alternative implementation, the current time - series data of the current transformer includes zero - sequence unbalance time - series data and negative - sequence unbalance time - series data.

[0012] As an alternative implementation, the time - series data is processed to obtain a time - domain image and a frequency - domain image. Specifically:

[0013] The Gram - Angle Field is used to convert the time - domain signal into a time - domain image;

[0014] The fast Fourier transform is used to convert the time - domain signal into a frequency - domain signal, and further the Gram - Angle is used to convert the frequency - domain signal into a frequency - domain image.

[0015] As an alternative implementation, the time - domain features and frequency - domain features are fused in a cross - fusion manner.

[0016] As an alternative implementation, the time - domain features and frequency - domain features are up - sampled to obtain the sampled features;

[0017] Average pooling and max - pooling operations are respectively performed on the time - domain features and frequency - domain features, a feature concatenation operation is performed on the obtained features, and then after a multi - layer perceptron and an activation function, the time - domain feature weight and the frequency - domain feature weight are generated. The sampled features are weighted using the generated feature weights to obtain the first time - domain weighted feature and the first frequency - domain weighted feature, and the first time - domain weighted feature and the first frequency - domain weighted feature are fused to obtain the hybrid features.

[0018] As an alternative implementation, the scale factor of the improved LSTM model is an adaptive scale factor. Specifically:

[0019]

[0020] Among them, and are the maximum and minimum step sizes determined based on experience, 、 、 are the reference values of zero - sequence unbalance, negative - sequence unbalance, and sampled current respectively, 、 、 All are parameters for determining step size adjustment. , , are hyperparameters, is the variance.

[0021] In a second aspect, the present invention provides a current transformer error state evaluation system based on cascaded multi-level CNNs, including:

[0022] A data acquisition and preprocessing module, configured to: acquire the current time series data of the historical operation of the current transformer, process the time series data, and obtain a time domain image and a frequency domain image;

[0023] A multi-level cascaded module, configured to: use the multi-level cascaded CNN algorithm to extract features from the time domain image and the frequency domain image respectively, obtain time domain features and frequency domain features, and fuse the time domain features and the frequency domain features to obtain mixed features;

[0024] An output module, configured to: input the time domain features and frequency domain features after being processed by the first-level CNN and the mixed features after being processed by the multi-level CNN into an improved LSTM model together to obtain the error state of the current transformer.

[0025] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the following steps are completed:

[0026] Acquire the current time series data of the historical operation of the current transformer, process the time series data, and obtain a time domain image and a frequency domain image;

[0027] Use the multi-level cascaded CNN algorithm to extract features from the time domain image and the frequency domain image respectively, obtain time domain features and frequency domain features, and fuse the time domain features and the frequency domain features to obtain mixed features;

[0028] Input the time domain features and frequency domain features after being processed by the first-level CNN and the mixed features after being processed by the multi-level CNN into an improved LSTM model together to obtain the error state of the current transformer.

[0029] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the following steps are implemented:

[0030] Acquire the current time series data of the historical operation of the current transformer, process the time series data, and obtain a time domain image and a frequency domain image;

[0031] The multi - stage cascaded CNN algorithm is used to extract features from the time - domain image and the frequency - domain image respectively, obtaining time - domain features and frequency - domain features, and the time - domain features and frequency - domain features are fused to obtain hybrid features;

[0032] The time - domain features and frequency - domain features after the first - stage CNN processing and the hybrid features after the multi - stage CNN processing are input into an improved LSTM model together to obtain the error state of the current transformer.

[0033] In a fifth aspect, the present invention provides a computer program product, including a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0034] Collect the current time - series data of the historical operation of the current transformer, process the time - series data to obtain a time - domain image and a frequency - domain image;

[0035] The multi - stage cascaded CNN algorithm is used to extract features from the time - domain image and the frequency - domain image respectively, obtaining time - domain features and frequency - domain features, and the time - domain features and frequency - domain features are fused to obtain hybrid features;

[0036] The time - domain features and frequency - domain features after the first - stage CNN processing and the hybrid features after the multi - stage CNN processing are input into an improved LSTM model together to obtain the error state of the current transformer.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] The present invention proposes a method and system for evaluating the error state of a current transformer based on multi - stage CNN cascading, designs an effective hybrid architecture, realizes mutual complementation in the feature extraction process through multiple CNN branches, and realizes two - way information exchange between time - domain features and frequency - domain features through the fusion of intermediate features. This fusion strategy improves the complementarity of features and enhances the model evaluation accuracy.

[0039] The advantages of the additional aspects of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The specification drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention.

[0041] Figure 1 It is a framework diagram of the method for evaluating the error state of a current transformer based on multi - stage CNN cascading provided in Embodiment 1 of the present invention;

[0042] Figure 2 It is a framework diagram of the cross - fusion of time - frequency domain feature maps of the present invention;

[0043] Figure 3 It is a schematic structural diagram of the improved LSTM model of the present invention. Detailed implementation manners

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0045] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0046] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those clearly listed steps or units, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0047] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0048] Term explanation:

[0049] Gramian Angular Field (GAF): It is an effective method for converting one-dimensional time series data into a two-dimensional image representation. Its core principle is to calculate the cosine values of the angles between data points in the time series and map these values to the pixels of a two-dimensional image, thereby generating an image that can reflect the dynamic and periodic characteristics of the time series.

[0050] Embodiment 1

[0051] As Figure 1 shown, this embodiment provides a method for evaluating the error state of a current transformer based on cascaded multi-level CNNs, including the following steps:

[0052] S1. Collect the current time series data of the historical operation of the current transformer, process the time series data to obtain a time domain image and a frequency domain image;

[0053] S2. Use the cascaded multi-level CNN algorithm to extract features from the time domain image and the frequency domain image respectively, obtain time domain features and frequency domain features, and fuse the time domain features and the frequency domain features to obtain mixed features;

[0054] S3. Input the time-domain features and frequency-domain features processed by the first-level CNN and the mixed features processed by the multi-level CNN into the improved LSTM model to obtain the error state of the current transformer.

[0055] Among them, S1. Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain the time-domain image and frequency-domain image. Specifically:

[0056] First, collect the secondary current data of the historical operation of the current transformer. The secondary current data of the current transformer includes the currents of three phases A, B, and C , , .

[0057] Collect the zero-sequence unbalance time-series data of the current transformer: , the negative-sequence unbalance time-series data: , where T represents the historical time period, , represent the zero-sequence unbalance and negative-sequence unbalance at the t-th moment.

[0058] Zero-sequence unbalance and negative-sequence unbalance calculation:

[0059] (1)

[0060] Zero-sequence unbalance: (2)

[0061] Negative-sequence unbalance: (3)

[0062] Among them , , are the three-phase currents, , , is the corresponding zero-sequence component of phase a. , respectively represent the zero-sequence unbalance and negative-sequence unbalance without the time dimension. And , has a time dimension T, , respectively represent the zero-sequence unbalance and negative-sequence unbalance at the t-th moment, , respectively represent the zero-sequence unbalance and negative-sequence unbalance within the historical time period T.

[0063] Use the Gramian Angular Field (GAF) to process the time-domain signal , Converted into a time-domain image 、 , specifically:

[0064] (4)

[0065] (5)

[0066] Using the Fast Fourier Transform (FFT) to convert the time-domain signal 、 into a frequency-domain signal, and further using the Gram angle to convert the frequency-domain signal 、 into a frequency-domain image 、 , specifically:

[0067] (6)

[0068] (7)

[0069] (8)

[0070] (9)

[0071] S2. Adopt the multi-level cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively, obtain the time-domain features and the frequency-domain features, and fuse the time-domain features and the frequency-domain features to get the hybrid features, specifically:

[0072] Denote the time-domain image TG and the frequency-domain image FG as:

[0073] (10)

[0074] (11)

[0075] Feature extraction of the time-domain image:

[0076] (12)

[0077] (13)

[0078] (14)

[0079] Feature extraction of the frequency-domain image:

[0080] (15)

[0081] (16)

[0082] Among them, i represents the number of cascades, represents the i-th level of time-domain hybrid features, represents the (i - 1)-th level of time-domain image features, represents the (i - 1)-th level of time-frequency hybrid features, represents the 0-th level of time-domain image features, represents the 0-th level of time-frequency hybrid features, represents the i-th level of frequency-domain image features, represents the (i - 1)-th level of frequency-domain image features, represents the 0-th level of frequency-domain image features.

[0083] The time-frequency domain features are fused by means of cross-fusion, and the specific process is as Figure 2 shown.

[0084] (1) First, upsample the time-domain image features and frequency-domain image features, and then splice the sampled data for feature concatenation to achieve the interaction of features between two different domains, integrate feature information of different scales and different levels, which helps the model to understand the input data more comprehensively, thereby improving its performance and accuracy.

[0085] (17)

[0086] (18)

[0087] (19)

[0088] Among them, Conv is the convolution operation, upsample is the upsampling operation, 、 are the sampled time-domain image features and frequency-domain image features, || is the feature concatenation symbol; is the contrast function, represents the weight of the i-th level of fusion.

[0089] (2) Take the time-domain image features and frequency-domain image features as inputs, after average pooling (AP) and max pooling (MP), perform feature concatenation, and then generate time-domain feature weights and frequency-domain feature weights through a multi-layer perceptron (MLP) and an activation function (Sigmoid); then use the generated features to weight the sampled features to obtain the first time-domain weighted features and the first frequency-domain weighted features;

[0090] (20)

[0091] (21)

[0092] (22)

[0093] (23)

[0094] Among them, represents the activation function, is the average pooling function, is the max pooling function, and MLP is the hidden function of the multi-layer perceptron. , are the weights of the time-domain image features and the frequency-domain image features, respectively.

[0095] (3)Fuse the first time-domain weighted feature and the first frequency-domain weighted feature to obtain the fused feature.

[0096] (24)

[0097] Among them, is the i-th level fused feature, is the cross product operation, is the direct sum operation.

[0098] Feature weighted fusion can combine the advantages of multiple features by assigning different weights to different features, improving the expression ability of the model. This fusion method enables the model to better capture the key information in the data, thereby improving the accuracy of classification.

[0099] S3. Input the time-domain features and frequency-domain features processed by the first-level CNN and the mixed features processed by the multi-level CNN into the improved LSTM model (as Figure 3 shown) to obtain the error state of the current transformer. The specific calculation process is as follows:

[0100] The process of evaluating the LSTM adopted in the present invention is as follows: The sequence value of the feature sequence X at time t is input into the network, and the update calculations of the forget gate, input gate, output gate, and memory unit are respectively performed. The specific process is as follows:

[0101] Forget gate: (25)

[0102] Input gate: (26)

[0103] Temporary memory: (27)

[0104] Memory update: (28)

[0105] Output gate: (29)

[0106] Hidden state: (30)

[0107] Among them, 、 、 are the information input, hidden state output, and hidden state input at time t respectively, 、 、 correspond to different gating structures respectively, is supplementary information, is memory information, is the weight matrix, b is the bias, 、 are activation functions that unify the data into the ranges of [0, 1] and [-1, 1] respectively.

[0108] In order to strengthen the information interaction between the features at the current moment and the error state at the previous moment and improve the prediction accuracy, the present invention is processed by the following formula:

[0109] (31)

[0110] (32)

[0111] Among them, is the activation function, is the scale factor, 、 are the weights of the input and output convolutional layers respectively, 、 are the biases of the input and output layers.

[0112] Since the input features and output states change with the time dimension and show a certain trend of change, the in the present invention is an adaptive scale, specifically:

[0113] (33)

[0114] Among them, and are the maximum and minimum step sizes determined according to experience, 、 、 are the reference values for describing zero-sequence imbalance, negative-sequence imbalance, and sampling current during the modeling period respectively, and the parameters 、 、 that determine the step size adjustment are only determined by the variance between two consecutive sampling points, 、 、 is a hyperparameter, p = phases A, B, and C, and exp is the exponential function with base e. is the variance.

[0115] The present invention evaluates the error state of current transformers, timely discovers out-of-tolerance current transformers, and thus ensures the stability of power grid operation.

[0116] Define that the out-of-tolerance operation state of the current transformer is a positive example, and the normal state of the current transformer is a negative example. Respectively count the number of correct and incorrect judgments for positive and negative examples, and use the accuracy ACC and precision P as the measurement indicators of its effectiveness. Among them, the accuracy describes the probability that the error estimation state of this method is consistent with the actual state, and the precision describes the probability of actual out-of-tolerance among all current transformers judged to be out-of-tolerance. The calculation formulas are shown in Eqs. (34) and (35):

[0117] (34)

[0118]

[0119] In the formula, TP: the sample is positive and the prediction result is positive; FP: the sample is negative and the prediction result is positive; TN: the sample is negative and the prediction result is negative; FN: the sample is positive and the prediction result is negative.

[0120] Ten current transformers were selected and evaluated using the method of the present invention. The accuracy ACC reached more than 98.817%, and the precision C reached more than 99.223%.

[0121] Embodiment 2

[0122] This embodiment provides a current transformer error state evaluation system based on cascaded multi-level CNNs, including:

[0123] A data acquisition and preprocessing module, configured to: acquire the current time-series data of the historical operation of the current transformer, process the time-series data, and obtain a time-domain image and a frequency-domain image;

[0124] A multi-level cascading module, configured to: respectively extract features from the time-domain image and the frequency-domain image using the multi-level cascaded CNN algorithm, obtain time-domain features and frequency-domain features, and fuse the time-domain features and the frequency-domain features to obtain hybrid features;

[0125] An output module, configured to: input the time-domain features and frequency-domain features after being processed by the first-level CNN and the hybrid features after being processed by the multi-level CNN into an improved LSTM model together to obtain the error state of the current transformer.

[0126] It should be noted here that the above modules correspond to the steps described in Embodiment 1. The examples and application scenarios implemented by the above modules and the corresponding steps are the same, but are not limited to the content disclosed in the above Embodiment 1. It should be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0127] In more embodiments, there is also provided:

[0128] An electronic device, including a memory and a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the following steps are completed:

[0129] Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image;

[0130] Use a multi-stage cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively, obtain time-domain features and frequency-domain features, and fuse the time-domain features and the frequency-domain features to obtain mixed features;

[0131] Input the time-domain features and frequency-domain features after being processed by the first-stage CNN and the mixed features after being processed by the multi-stage CNN into an improved LSTM model together to obtain the error state of the current transformer.

[0132] It should be understood that in this embodiment, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0133] The memory can include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory can also include a non-volatile random memory. For example, the memory can also store information about the device type.

[0134] A computer-readable storage medium for storing computer instructions. When the computer instructions are executed by the processor, the following steps are implemented:

[0135] Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image;

[0136] Use a multi-stage cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively, obtain time-domain features and frequency-domain features, and fuse the time-domain features and the frequency-domain features to obtain mixed features;

[0137] The time-domain features and frequency-domain features after being processed by the first-level CNN and the mixed features after being processed by multiple-level CNN are input into an improved LSTM model together to obtain the error state of the current transformer.

[0138] The method in Embodiment 1 can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory, or electrically erasable programmable memory, register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method. To avoid repetition, it will not be described in detail here.

[0139] A computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0140] Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image;

[0141] Adopt a multi-level cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively, obtain time-domain features and frequency-domain features, and fuse the time-domain features and the frequency-domain features to obtain mixed features;

[0142] The time-domain features and frequency-domain features after being processed by the first-level CNN and the mixed features after being processed by multiple-level CNN are input into an improved LSTM model together to obtain the error state of the current transformer.

[0143] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. This computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to execute the process / method as described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided as needed. The machine-executable instructions for program modules can be executed locally or within a distributed device. In a distributed device, program modules can be located in local and remote storage media.

[0144] The computer program code for implementing the method of the present invention can be written in one or more programming languages. This computer program code can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, so that when the program code is executed by the computer or other programmable data processing devices, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the computer, partially on the computer, as an independent software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0145] In the context of the present invention, the computer program code or relevant data can be carried by any appropriate carrier, so that the device, apparatus, or processor can execute the various processes and operations described above. Examples of carriers include signals, computer-readable media, and so on. Examples of signals can include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0146] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in conjunction with this embodiment can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0147] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, it is not a limitation to the protection scope of the present invention. Those skilled in the art should understand that, based on the technical solution of the present invention, various modifications or deformations that can be made by those skilled in the art without creative efforts are still within the protection scope of the present invention.

Claims

1. A method for evaluating the error state of a current transformer based on cascaded multi-level CNNs, characterized in that Including the following steps: Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image; Use the multi-stage cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively to obtain time-domain features and frequency-domain features, and use the cross-fusion method to fuse the time-domain features and the frequency-domain features to obtain mixed features; Input the time-domain features and frequency-domain features after being processed by the first-stage CNN and the mixed features after being processed by the multi-stage CNN into an improved LSTM model to obtain the error state of the current transformer. The scale factor of the improved LSTM model is an adaptive scale factor, specifically: ； Among them, and are the maximum and minimum step sizes determined based on experience, is the zero-sequence imbalance, is the negative-sequence imbalance, 、 、 are the reference values of zero-sequence imbalance, negative-sequence imbalance, and sampled current respectively, 、 、 are all parameters that determine step size adjustment, 、 、 are hyperparameters, is the variance.

2. The method for evaluating the error state of a current transformer based on cascaded multi-level CNNs according to claim 1, wherein The current time-series data of the historical operation of the current transformer includes zero-sequence unbalanced time-series data and negative-sequence unbalanced time-series data.

3. The method for evaluating the error state of a current transformer based on cascaded multi-level CNNs according to claim 1, wherein, Process the time-series data to obtain a time-domain image and a frequency-domain image, specifically: Use the Gram angular field to convert the time-domain signal into a time-domain image; Use the fast Fourier transform to convert the time-domain signal into a frequency-domain signal, and further use the Gram angle to convert the frequency-domain signal into a frequency-domain image.

4. The method for evaluating the error state of a current transformer based on cascaded multi-level CNNs according to claim 1, characterized in that, Upsample the time-domain features and the frequency-domain features to obtain the sampled features; Perform average pooling and max pooling operations on the time-domain features and the frequency-domain features respectively, perform a feature splicing operation on the obtained features, and then generate time-domain feature weights and frequency-domain feature weights after passing through a multi-layer perceptron and an activation function. Use the generated feature weights to weight the sampled features to obtain the first time-domain weighted feature and the first frequency-domain weighted feature, and fuse the first time-domain weighted feature and the first frequency-domain weighted feature to obtain mixed features.

5. A current transformer error state evaluation system based on cascaded multi-level CNNs, characterized in that, Including: A data acquisition and preprocessing module configured to: collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image; A multi-stage cascaded module configured to: use the multi-stage cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively, obtain time-domain features and frequency-domain features in a cross-fusion manner, and fuse the time-domain features and the frequency-domain features to obtain mixed features; An output module configured to: input the time-domain features and frequency-domain features after being processed by the first-stage CNN and the mixed features after being processed by the multi-stage CNN into an improved LSTM model to obtain the error state of the current transformer. The scale factor of the improved LSTM model is an adaptive scale factor, specifically: ； Among them, and are the maximum and minimum step sizes determined based on experience, is the zero-sequence unbalance, is the negative-sequence unbalance, 、 、 are the reference values of zero-sequence unbalance, negative-sequence unbalance, and sampled current respectively, 、 、 are all parameters determining step size adjustment, 、 、 are hyperparameters, is the variance.

6. An electronic device, characterized in that, Including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the following steps are completed: Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image; Use the multi-stage cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively to obtain time-domain features and frequency-domain features, and use the cross-fusion method to fuse the time-domain features and the frequency-domain features to obtain mixed features; Input the time-domain features and frequency-domain features processed by the first-level CNN and the mixed features processed by multiple-level CNN into an improved LSTM model to obtain the error state of the current transformer. The scale factor of the improved LSTM model is an adaptive scale factor, specifically: ； Among them, and are the maximum and minimum step sizes determined based on experience, is the zero-sequence unbalance, is the negative-sequence unbalance, 、 、 are the reference values of zero-sequence unbalance, negative-sequence unbalance, and sampled current respectively, 、 、 are all parameters determining step-size adjustment, 、 、 are hyperparameters, is the variance.

7. A computer-readable storage medium, characterized in that, Used to store computer instructions, when the computer instructions are executed by a processor, the following steps are implemented: Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image; Use a multi-level cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively to obtain time-domain features and frequency-domain features, and use a cross-fusion method to fuse the time-domain features and the frequency-domain features to obtain mixed features; Input the time-domain features and frequency-domain features processed by the first-level CNN and the mixed features processed by multiple-level CNN into an improved LSTM model to obtain the error state of the current transformer. The scale factor of the improved LSTM model is an adaptive scale factor, specifically: ； Among them, and are the maximum and minimum step sizes determined based on experience, is the zero-sequence unbalance, is the negative-sequence unbalance, 、 、 are the reference values of zero-sequence unbalance, negative-sequence unbalance, and sampled current respectively, 、 、 are all parameters that determine step-size adjustment, 、 、 are hyperparameters, is the variance.

8. A computer program product, characterized in that, Includes a computer program, when the computer program is executed by a processor, the following steps are implemented: Collect the current time-series data of the historical operation of the current transformer, process the time-series data to obtain a time-domain image and a frequency-domain image; Use a multi-level cascaded CNN algorithm to extract features from the time-domain image and the frequency-domain image respectively to obtain time-domain features and frequency-domain features, and use a cross-fusion method to fuse the time-domain features and the frequency-domain features to obtain mixed features; Input the time-domain features and frequency-domain features processed by the first-level CNN and the mixed features processed by multiple-level CNN into an improved LSTM model to obtain the error state of the current transformer. The scale factor of the improved LSTM model is an adaptive scale factor, specifically: ； Among them, and are the maximum and minimum step sizes determined based on experience, is the zero-sequence unbalance, is the negative-sequence unbalance, 、 、 are the reference values of zero-sequence unbalance, negative-sequence unbalance, and sampled current respectively, 、 、 are all parameters that determine step size adjustment, 、 、 are hyperparameters, is the variance.

Images