Transformer fault detection method based on antenna enhanced radio frequency identification sensor

Abstract

The invention relates to a transformer fault detection method based on an antenna enhanced radio frequency identification sensor. The method comprises the steps of performing modal decomposition according to an initial vibration signal of a to-be-detected transformer to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal, and performing Hilbert transform processing according to the similar characteristic modes to determine a Hilbert envelope spectrum of the initial vibration signal, and inputting the Hilbert envelope spectrum into a pre-trained sparse noise reduction auto-encoder SSDA model to obtain features of an initial vibration signal, performing fault classification on the to-be-detected transformer according to the features of the initial vibration signal, and determining a current fault type of the to-be-detected transformer. By adopting the method, whether the fault generated by the transformer to be detected is the known fault or not is not limited, and the accuracy and timeliness of transformer fault detection based on the antenna enhanced radio frequency identification sensor can be improved.

Description

Technical Field

[0001] This application relates to the field of power electronics technology, and in particular, to a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor. Background Art

[0002] With the development of power grid technology, transformers have become core equipment in the power system, and their stable operation is crucial for ensuring the safety and reliability of the power grid. In actual operation, there are various types of faults that transformers encounter. Therefore, it is particularly important to detect the faults occurring in transformers.

[0003] In related technologies, the operating state of a transformer to be measured, that is, the fault state or non-fault state, is mainly determined based on data such as the temperature, oil quality, vibration, and gas analysis of the transformer to be measured.

[0004] However, in the process of detecting transformer faults in related technologies, there will be a problem of inaccurate detection results. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor that can improve the accuracy of transformer fault detection based on an antenna-enhanced radio frequency identification sensor.

[0006] In a first aspect, this application provides a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor, including:

[0007] Performing modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal;

[0008] Performing Hilbert transform processing on each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal;

[0009] Inputting the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder SSDA model to obtain the characteristics of the initial vibration signal;

[0010] According to the characteristics of the initial vibration signal, classifying the faults of the transformer to be measured to determine the current fault type of the transformer to be measured.

[0011] In one embodiment, performing modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal includes:

[0012] Obtaining a plurality of simulated vibration signals obtained by adding a noise signal to the initial vibration signal;

[0013] Perform empirical mode decomposition on each simulated vibration signal to obtain multiple intrinsic modes corresponding to the initial vibration signal;

[0014] Perform similarity processing on each intrinsic mode and the initial vibration signal to obtain the similarity values between each intrinsic mode and the initial vibration signal;

[0015] Select the intrinsic modes with similarity values greater than a preset threshold from each intrinsic mode as similar characteristic modes.

[0016] In one embodiment, perform Hilbert transform processing according to each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal, including:

[0017] Perform Hilbert transform processing on each similar characteristic mode to obtain the time-domain envelope spectrum of each similar characteristic mode;

[0018] Perform fast Fourier transform on each time-domain envelope spectrum to obtain the frequency-domain envelope spectrum of each similar characteristic mode;

[0019] Combine the frequency-domain envelope spectra of each to obtain the Hilbert envelope spectrum of the initial vibration signal.

[0020] In one embodiment, the construction process of the SSDA model includes:

[0021] Obtain the corresponding multiple sample Hilbert envelope spectra according to the vibration signal training sets of multiple sample transformers;

[0022] Input each sample Hilbert envelope spectrum into the initial SSDA model to obtain a candidate SSDA model;

[0023] In the case where the loss function value of the candidate SSDA model is greater than the preset loss value or the candidate SSDA model does not meet the convergence condition, use the quantum particle swarm QPSO algorithm to optimize the candidate SSDA model according to each sample Hilbert envelope spectrum to obtain a trained SSDA model.

[0024] In one embodiment, obtaining the corresponding multiple sample Hilbert envelope spectra according to the vibration signal training sets of multiple sample transformers includes:

[0025] Perform modal decomposition on the vibration signal training set to obtain multiple similar characteristic mode sets corresponding to the vibration signal training set;

[0026] Perform Hilbert transform processing according to each similar characteristic mode set to determine the multiple sample Hilbert envelope spectra of the vibration signal training set.

[0027] In one embodiment, the above method further includes:

[0028] In the case of failure in the fault classification of the transformer to be measured, add the initial vibration signal to the vibration signal training set to obtain a new training set;

[0029] Use the new training set to optimize the SSDA model to obtain an updated SSDA model;

[0030] Input the Hilbert envelope spectrum into the updated SSDA model to complete the fault classification of the transformer to be measured.

[0031] In one embodiment, the above method further includes:

[0032] Obtain the original vibration signal of the transformer to be measured;

[0033] Preprocess the original vibration signal to obtain the initial vibration signal of the transformer to be measured.

[0034] In a second aspect, the present application further provides a transformer fault detection device based on an antenna-enhanced radio frequency identification sensor, including:

[0035] A modal decomposition module, configured to perform modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal;

[0036] A transformation processing module, configured to perform Hilbert transform processing according to each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal;

[0037] A feature acquisition module, configured to input the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder SSDA model to obtain the features of the initial vibration signal;

[0038] A fault classification module, configured to perform fault classification on the transformer to be measured according to the features of the initial vibration signal to determine the current fault type of the transformer to be measured.

[0039] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method in any embodiment of the first aspect are implemented.

[0040] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method in any embodiment of the first aspect are implemented.

[0041] In a fifth aspect, the present application further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the method in any embodiment of the first aspect are implemented.

[0042] The above-mentioned transformer fault detection method based on an antenna-enhanced radio frequency identification sensor includes: performing modal decomposition on the initial vibration signal of the transformer to be measured to obtain multiple similar characteristic modes corresponding to the initial vibration signal, performing Hilbert transform processing on each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal, inputting the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder (SSDA) model to obtain the characteristics of the initial vibration signal, and classifying the faults of the transformer to be measured according to the characteristics of the initial vibration signal to determine the current fault type of the transformer to be measured. Using the above method has no limitation on whether the faults generated by the transformer to be measured are known faults. It can not only detect known faults but also detect unknown faults. This process can not only improve the wide applicability of transformer fault detection based on an antenna-enhanced radio frequency identification sensor but also improve the accuracy and timeliness of transformer fault detection based on an antenna-enhanced radio frequency identification sensor, reduce the risks and maintenance costs of the transformer, and thus improve the reliability and safety of the power system. In addition, this method helps to detect potential faults and problems of the transformer in advance, reduce the downtime and maintenance costs of the transformer; at the same time, the above method does not require manual participation, thereby being able to accelerate the speed and efficiency of transformer fault detection based on an antenna-enhanced radio frequency identification sensor. Description of the Drawings

[0043] To more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0044] Figure 1 It is an application environment diagram of a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in an embodiment;

[0045] Figure 2 It is a schematic flowchart of a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in an embodiment;

[0046] Figure 3 It is a schematic flowchart of a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in another embodiment;

[0047] Figure 4 It is a schematic flowchart of a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in another embodiment;

[0048] Figure 5 It is a schematic flowchart of a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in another embodiment;

[0049] Figure 6 It is a schematic flowchart of a transformer fault detection method based on an antenna - enhanced radio - frequency identification sensor in another embodiment;

[0050] Figure 7 It is a schematic flowchart of a transformer fault detection method based on an antenna - enhanced radio - frequency identification sensor in another embodiment;

[0051] Figure 8 It is a schematic flowchart of a transformer fault detection method based on an antenna - enhanced radio - frequency identification sensor in another embodiment;

[0052] Figure 9 It is a block diagram of the structure of a transformer fault detection device based on an antenna - enhanced radio - frequency identification sensor in one embodiment;

[0053] Figure 10 It is an internal structure diagram of a computer device in one embodiment. Detailed implementation manners

[0054] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0055] The transformer fault detection method based on an antenna - enhanced radio - frequency identification sensor provided by the embodiments of the present application can be applied to an application environment as shown in Figure 1 The transformer fault detection system based on an antenna - enhanced radio - frequency identification sensor includes an antenna - enhanced radio - frequency identification (RFID) sensor, an RFID reader, and a transformer to be measured. Among them, the antenna - enhanced RFID sensor communicates with the RFID reader through a network, and this communication method can be connection methods such as Bluetooth, mobile data, and Wifi. At the same time, the antenna - enhanced RFID sensor may include a dual - antenna device, a power manager, a micro - controller (i.e., the MCU module), a vibration sensor module, and an RFID tag chip; the power manager may be composed of a two - stage boost rectifier and a low - dropout regulator in sequence, and perform boost processing and voltage regulation processing on the received voltage based on the LDO voltage rule. Among them, the transformer to be measured may be, but is not limited to, an on - load tap - changing transformer, an off - load tap - changing transformer, a copper - winding transformer, an aluminum - winding transformer, a core - type transformer, or a shell - type transformer.

[0056] It should be noted here that the dual-antenna device may include an energy harvesting antenna and a communication antenna (i.e., an RFID tag antenna); the operating frequency band of the energy harvesting antenna is 915 MHz, which is used to collect radio frequency energy from the radio frequency signal transmitted by the RFID reader, so as to send the radio frequency energy to the power manager for the power manager to reuse and supply power to the microcontroller, the vibration sensor module and the RFID tag chip; the communication antenna is used to collect the original vibration signal of the transformer to be measured, and transmit the original vibration signal to the RFID tag chip through the power manager, the microcontroller and the vibration sensor module for storage in the ID area of the RFID tag chip; the RFID tag chip is used to transmit the original vibration signal to the RFID reader. Among them, the above boost rectifier may include a zero-bias diode for converting radio frequency energy into a DC voltage; the above low-dropout regulator is used to convert the DC voltage output by the two-stage boost rectifier into a stable output voltage to supply power to the microcontroller, the vibration sensor module and the RFID tag chip, reducing the power consumption of the antenna-enhanced RFID sensor.

[0057] At the same time, the above microcontroller may be a low-power microcontroller, which is used to manage the original vibration signal transmitted by the power manager. In the embodiment of the present application, the vibration sensor module may be arranged on the outer wall of the oil tank of the transformer to be measured; the vibration sensor module may be composed of a 3-axis accelerometer, with a sampling frequency of 3.2 kHz and a power consumption of 22 , and the frequency range is 100 - 800 Hz, which is used to obtain the vibration signal in the X-axis direction of the original vibration signal to reduce redundant information. In addition, the RFID tag chip has an I²C interface and a UHF interface for communicating with the RFID reader.

[0058] In an exemplary embodiment, as Figure 2 shown, a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor is provided. Taking the RFID reader in Figure 1 as an example for description, this method can be implemented through the following steps:

[0059] S100. Perform modal decomposition on the initial vibration signal of the transformer to be measured to obtain multiple similar characteristic modes corresponding to the initial vibration signal.

[0060] Specifically, the RFID reader can obtain the initial vibration signal of the transformer to be measured, and perform modal decomposition on the initial vibration signal of the transformer to be measured by using the empirical mode decomposition algorithm to obtain multiple similar characteristic modes corresponding to the initial vibration signal.

[0061] In addition, the RFID reader can pre-train a modal decomposition algorithm model, and then input the initial vibration signal of the transformer to be measured into the modal decomposition algorithm model. After performing modal decomposition on the initial vibration signal of the transformer to be measured, multiple similar characteristic modes corresponding to the initial vibration signal are output.

[0062] S200. Perform Hilbert transform processing according to each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal.

[0063] In practical applications, the RFID reader can perform Hilbert transform processing on each similar characteristic mode to obtain the Hilbert envelope spectrum of the initial vibration signal.

[0064] S300. Input the Hilbert envelope spectrum into the pre-trained sparse denoising autoencoder SSDA model to obtain the characteristics of the initial vibration signal.

[0065] Among them, the sparse denoising autoencoder SSDA model can include an input layer, an encoder, a sparse constraint layer, a denoising layer, a decoder, and an output layer; both the encoder and the decoder can be composed of multiple hidden layers.

[0066] Specifically, the RFID reader can input the Hilbert envelope spectrum into the pre-trained sparse denoising autoencoder SSDA model, and the SSDA model outputs the characteristics of the initial vibration signal.

[0067] S400. Classify the faults of the transformer to be measured according to the characteristics of the initial vibration signal to determine the current fault type of the transformer to be measured.

[0068] In practical applications, the RFID reader can classify the characteristics of the initial vibration signal to complete the fault classification of the transformer to be measured according to the characteristic classification result and determine the current fault type of the transformer to be measured.

[0069] In addition, the RFID reader can also train an algorithm model, and then input the characteristics of the initial vibration signal into the algorithm model. The algorithm model classifies the faults of the transformer to be measured according to the characteristics of the initial vibration signal and then outputs the current fault type of the transformer to be measured.

[0070] In an embodiment of the present application, the RFID reader can randomly select multiple features from the features of the initial vibration signal as the initial cluster centers in the fault classification process, calculate the distances from each feature to all the initial cluster centers, allocate each feature to the initial cluster center closest to it, further update each initial cluster center to the mean value of all the feature points within the cluster it belongs to, and repeat the iteration until the updated cluster centers no longer change or reach the maximum number of iterations. Finally, a clustering quality evaluation index can be used to evaluate the clustering quality of each feature. Optionally, the above clustering quality evaluation index can be the Calinski-Harabasz index, the Davies-Bouldin index, etc. However, in the embodiment of the present application, the clustering quality evaluation index can be the silhouette coefficient; the silhouette coefficient can be expressed by the following formula (1):

[0071] (1)

[0072] wherein, represents the average distance between the feature in any class and other features within the cluster it belongs to, represents the average distance between the feature in any class and all the features within the closest other cluster.

[0073] It should be noted here that if , it means that the feature is very close to the cluster it belongs to and has a good separation from other clusters; if , it means that the feature may be misallocated to a certain cluster, and the feature should be reclassified.

[0074] In the technical solution of the embodiment of the present application, modal decomposition is performed on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal. Hilbert transform processing is performed according to each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal. The Hilbert envelope spectrum is input into a pre-trained sparse denoising autoencoder SSDA model to obtain the characteristics of the initial vibration signal, and the fault classification of the transformer to be measured is performed according to the characteristics of the initial vibration signal to determine the current fault type of the transformer to be measured. The above method has no limitation on whether the fault generated by the transformer to be measured is a known fault. It can not only detect known faults but also detect unknown faults. This process can not only improve the wide applicability of transformer fault detection based on antenna-enhanced RFID sensors, but also improve the accuracy and timeliness of transformer fault detection based on antenna-enhanced RFID sensors, reduce the risk and maintenance cost of transformers, and thus improve the reliability and safety of the power system. In addition, this method helps to detect potential faults and problems of transformers in advance, reduce the downtime and maintenance cost of transformers. At the same time, the above method does not require manual participation, thereby being able to accelerate the speed and efficiency of transformer fault detection based on antenna-enhanced RFID sensors.

[0075] The process of obtaining a plurality of similar characteristic modes corresponding to the initial vibration signal by performing modal decomposition on the initial vibration signal of the transformer to be measured as described above will be described below. In one embodiment, as Figure 3 shown, the steps in S100 above can be implemented in the following manner:

[0076] S110. Obtain a plurality of simulated vibration signals obtained by adding a noise signal to the initial vibration signal.

[0077] Among them, during the modal decomposition process, m decomposition tests can be performed. During each decomposition test, a noise signal can be added to the initial vibration signal to obtain the corresponding simulated vibration signal. Correspondingly, the RFID reader can obtain a plurality of simulated vibration signals obtained by adding a noise signal to the initial vibration signal during the m decomposition tests.

[0078] Optionally, the noise signals added to the initial vibration signal during each decomposition test can be the same or different; the noise signal can be. In practical applications, the noise signal added to the initial vibration signal during each decomposition test can be non-Gaussian noise, colored noise, additive noise, etc. In the embodiment of the present application, the noise signal is taken as random Gaussian white noise as an example for illustration.

[0079] It should be noted here that the simulated vibration signal obtained during any decomposition test can be expressed by the following formula (2):

[0080] (2)

[0081] Among them, represents the noise signal added in the m-th decomposition test, represents the total number of decomposition tests, and t represents the lengths of the initial vibration signal, the noise signal, and the simulated vibration signal.

[0082] S120. Perform empirical mode decomposition on each simulated vibration signal to obtain multiple intrinsic modes corresponding to the initial vibration signal.

[0083] Specifically, the RFID reader can adopt the ensemble empirical mode decomposition (EEMD) algorithm to perform empirical mode decomposition on each simulated vibration signal to obtain multiple intrinsic modes corresponding to the initial vibration signal.

[0084] In the embodiment of the present application, the process of performing empirical mode decomposition on the simulated vibration signal obtained from the m-th decomposition test can be expressed by the following formula (3):

[0085] (3)

[0086] Among them, represents the remaining component in the initial vibration signal during the m-th decomposition test process, represents the number of working conditions generated by the transformer under test during each decomposition test process, is the i-th type of sub-intrinsic mode IMF corresponding to the working condition of the transformer under test during the m-th decomposition test process. In the embodiment of the present application, it is described by taking the number of working conditions generated by the transformer under test during each decomposition test process as equal as an example. Optionally, the working conditions of the transformer under test can be conditions such as faults or noises.

[0087] Furthermore, the mean value of the sub-intrinsic modes corresponding to each type of working condition can be calculated to obtain multiple intrinsic modes corresponding to the initial vibration signal; the intrinsic mode IMFs obtained by calculating the mean value of the sub-intrinsic modes corresponding to the i-th type of working condition can be expressed by the following formula (4):

[0088] (4)

[0089] S130. Perform similarity processing on each intrinsic mode and the initial vibration signal to obtain the similarity value between each intrinsic mode and the initial vibration signal.

[0090] Specifically, the RFID reader can perform similarity processing on each intrinsic mode and the initial vibration signal using a similarity calculation method to obtain the similarity value between each intrinsic mode and the initial vibration signal. Among them, the above similarity calculation method can be methods such as Euclidean distance, Manhattan distance, cosine similarity, Pearson correlation coefficient, Hamming distance, Pearson correlation coefficient, etc.

[0091] In practical applications, the RFID reader can perform similarity processing on each intrinsic mode and the initial vibration signal to obtain the similarity value between each intrinsic mode and the initial vibration signal.

[0092] In the embodiment of the present application, calculating the intrinsic mode and the initial vibration signal the similarity value between can be expressed by formula (5) as:

[0093] (5)

[0094] Wherein, to represents a part of the length in length t.

[0095] S140. Screen out the intrinsic modes with similarity values greater than the preset threshold from each intrinsic mode as similar characteristic modes.

[0096] Specifically, the RFID reader can determine whether the similarity value between each intrinsic mode and the initial vibration signal is greater than the preset threshold, and screen out the intrinsic modes with similarity values greater than the preset threshold from each intrinsic mode as similar characteristic modes according to the judgment result.

[0097] Optionally, the above preset threshold can be determined by the user's self - definition or can also be determined according to historical experience values.

[0098] In the technical solution of the embodiment of the present application, multiple simulated vibration signals obtained by adding noise signals to the initial vibration signal are acquired, empirical mode decomposition is performed on each simulated vibration signal to obtain multiple intrinsic modes corresponding to the initial vibration signal, similarity processing is performed on each intrinsic mode and the initial vibration signal to obtain the similarity value between each intrinsic mode and the initial vibration signal, and the intrinsic modes with similarity values greater than the preset threshold are screened out from each intrinsic mode as similar characteristic modes; the above method can screen out a part of the intrinsic modes from the multiple intrinsic modes corresponding to the initial vibration signal of the transformer to be measured for further processing, thereby being able to reduce the amount of data for subsequent processing, speed up the transformer fault detection based on the antenna - enhanced radio frequency identification sensor, and reduce the complexity of the transformer fault detection based on the antenna - enhanced radio frequency identification sensor.

[0099] In one embodiment, such asFigure 4 As shown in Figure 4 , the steps of performing Hilbert transform processing on each similar characteristic mode in the above S200 to determine the Hilbert envelope spectrum of the initial vibration signal may include:

[0100] S210. Perform Hilbert transform processing on each similar characteristic mode to obtain the time-domain envelope spectrum of each similar characteristic mode.

[0101] In the embodiments of the present application, for any similar characteristic mode, the process of performing Hilbert transform processing on the similar characteristic mode can be expressed by the following formula (6):

[0102] (6)

[0103] Where represents the Hilbert transform, represents the th similar characteristic mode, represents the th time-domain envelope spectrum of the similar characteristic mode.

[0104] S220. Perform fast Fourier transform on each time-domain envelope spectrum to obtain the frequency-domain envelope spectrum of each similar characteristic mode.

[0105] S230. Combine the frequency-domain envelope spectra to obtain the Hilbert envelope spectrum of the initial vibration signal.

[0106] In practical applications, the RFID reader can splice the frequency-domain envelope spectra into a one-dimensional feature vector to complete the combination to obtain the Hilbert envelope spectrum RHES of the initial vibration signal.

[0107] The technical solution in the embodiments of the present application performs Hilbert transform processing on each similar characteristic mode to obtain the time-domain envelope spectrum of each similar characteristic mode, performs fast Fourier transform on each time-domain envelope spectrum to obtain the frequency-domain envelope spectrum of each similar characteristic mode, and combines the frequency-domain envelope spectra to obtain the Hilbert envelope spectrum of the initial vibration signal; the above method can process each similar characteristic mode to obtain the corresponding Hilbert envelope spectrum, and this process can effectively remove the interference of noise signals on the analysis of the original vibration signal and improve the accuracy of subsequent fault detection results.

[0108] The construction process of the above SSDA model will be described below. In one embodiment, before performing the steps in the above S300, as Figure 5 shown, the above method may further include the following steps:

[0109] S310. Obtain corresponding multiple sample Hilbert envelope spectra according to the vibration signal training sets of multiple sample transformers.

[0110] Among them, the multiple sample transformers can be transformers of multiple different types; the vibration signal training set can include the initial vibration signals of multiple sample transformers. Further, the RFID reader can pre-train an algorithm model, and then input the initial vibration signals of each sample transformer in the vibration signal training set into the algorithm model, and the corresponding multiple sample Hilbert envelope spectra are output.

[0111] In one embodiment, as Figure 6 shown, the step of obtaining the corresponding multiple sample Hilbert envelope spectra according to the vibration signal training set of multiple sample transformers in S310 above may include:

[0112] S311. Perform modal decomposition on the vibration signal training set to obtain multiple similar characteristic mode sets corresponding to the vibration signal training set.

[0113] In practical applications, the RFID reader can perform modal decomposition on the initial vibration signals of each sample transformer in the vibration signal training set to obtain multiple similar characteristic modes corresponding to the initial vibration signals of each sample transformer. Among them, the initial vibration signal of the sample transformer can correspond to multiple similar characteristic modes and constitute a similar characteristic mode set.

[0114] It should be noted here that the modal decomposition here is similar to the modal decomposition process in S100 above, and the embodiments of the present application will not elaborate on this.

[0115] S312. Perform Hilbert transform processing according to each similar characteristic mode set to determine the multiple sample Hilbert envelope spectra of the vibration signal training set.

[0116] Further, the RFID reader can perform Hilbert transform processing according to each similar characteristic mode set to obtain the multiple sample Hilbert envelope spectra of the vibration signal training set. Among them, the Hilbert transform processing here is similar to the Hilbert transform processing process in S200 above, and the embodiments of the present application will not elaborate on this.

[0117] S320. Input each sample Hilbert envelope spectrum into the initial SSDA model to obtain a candidate SSDA model.

[0118] In practical applications, the RFID reader can initialize the parameters of the pre-constructed original SSDA model to obtain the initial SSDA model.

[0119] At the same time, the RFID reader can input each sample Hilbert envelope spectrum into the initial SSDA model and train the initial SSDA model once, that is, update the model parameters once to obtain the corresponding candidate SSDA model.

[0120] S330. When the loss function value of the candidate SSDA model is greater than the preset loss value or the candidate SSDA model does not meet the convergence condition, the quantum particle swarm optimization (QPSO) algorithm is used to optimize the candidate SSDA model according to the Hilbert envelope spectrum of each sample to obtain a trained SSDA model.

[0121] Specifically, the RFID reader can calculate the loss function value of the candidate SSDA model according to the initial vibration signals of each sample transformer and the corresponding results output by the candidate SSDA model by using the loss function. When it is determined that the loss function value of the candidate SSDA model is greater than the preset loss value or the candidate SSDA model does not meet the convergence condition, the QPSO algorithm can continue to be used to optimize the candidate SSDA model according to the Hilbert envelope spectrum of each sample, that is, to perform multiple iterations on the candidate SSDA model until the loss function value of the candidate SSDA model after multiple iterations is greater than the preset loss value or the candidate SSDA model after iteration meets the convergence condition, and the candidate SSDA model after multiple iterations is determined as the trained SSDA model.

[0122] Among them, in the process of optimizing the candidate SSDA model according to the Hilbert envelope spectrum of each sample by using the QPSO algorithm, the minimization of the loss function can be used as the objective function of the QPSO algorithm, and the parameters of the candidate SSDA model can be used as the positions of the particles in the QPSO algorithm to achieve model optimization.

[0123] Optionally, the above loss function can be mean square error, cross-entropy loss function, KL divergence, logarithmic loss, etc. In the embodiments of the present application, the above loss function can be expressed by the following formula (7):

[0124] (7)

[0125] Among them, represents the reconstruction error, represents the initial vibration signal of the l-th sample transformer, represents the output result of the candidate SSDA model corresponding to the l-th sample transformer, represents the weight regularization, represents the weight decay coefficient, represents the sparse constraint term, represents and the divergence between represents the target sparsity, represents the average activation value of the -th unit in the hidden layer of the candidate SSDA model, Represents the number of hidden layers in the candidate SSDA model.

[0126] It should be noted here that in the case where the loss function value of the candidate SSDA model is less than the preset loss value or the candidate SSDA model reaches the convergence condition, the candidate SSDA model can be determined as the optimal SSDA model.

[0127] In the embodiment of the present application, if the Hilbert envelope spectrum of any sample is X, then X is input into the initial SSDA model, and the result h obtained by the encoder in the initial SSDA model mapping X to the hidden layer can be expressed by formula (8) as:

[0128] (8)

[0129] Where, Represents the weight matrix, Represents the bias, Represents the Sigmoid activation function, .

[0130] At the same time, the output result of the output layer of the candidate SSDA model can be expressed by formula (9) as:

[0131] (9)

[0132] Where, Represents the decoding weight, Represents the decoding bias, Represents the activation function.

[0133] The technical solution in the embodiment of the present application can train the initial SSDA model to obtain an optimal SSDA model with higher stability and accuracy, so that in the subsequent application process, the optimal SSDA model can be used to achieve feature extraction and improve the accuracy of the extracted features.

[0134] In some scenarios, the fault features currently generated by the transformer to be tested cannot be detected by the pre-trained SSDA model. In this case, the pre-trained SSDA model needs to be optimized again to ensure that the finally optimized SSDA model can detect normally. The process of optimizing the SSDA model again is described below. In one embodiment, after performing the steps in S300 above, as Figure 7 shown, the above method may further include:

[0135] S500. In the case where the fault classification of the transformer to be tested fails, add the initial vibration signal to the vibration signal training set to obtain a new training set.

[0136] It should be noted here that in the case where the failure of the fault classification of the transformer to be measured is determined, it indicates that the SSDA model fails to extract features, that is, the SSDA model cannot extract the current fault features of the transformer to be measured. At this time, the RFID reader can optimize the SSDA model again.

[0137] Specifically, the RFID reader can add the initial vibration signal of the transformer to be measured to the vibration signal training set to obtain a new training set.

[0138] S600. Optimize the SSDA model using the new training set to obtain an updated SSDA model.

[0139] In practical applications, the RFID reader can optimize the SSDA model using the new training set according to the steps in S310 - S330 above to obtain the optimal updated SSDA model. During the optimization process, the learning rate can be adjusted according to the following formula (10) :

[0140] (10)

[0141] Where, represents the initial learning rate, represents the adjustment factor, represents the number of training rounds.

[0142] At the same time, during the optimization process, the update of the weights can be achieved through the gradient descent of the objective function, and the weights can be updated in a weighted sum manner. The update process can be expressed by the following formula (11) as:

[0143] (11)

[0144] Where, is the improvement of the objective function of.

[0145] S700. Input the Hilbert envelope spectrum into the updated SSDA model to complete the fault classification of the transformer to be measured.

[0146] Furthermore, the RFID reader can input the Hilbert envelope spectrum corresponding to the initial vibration signal of the transformer to be measured into the updated SSDA model to obtain the features corresponding to the initial vibration signal of the transformer to be measured, and then classify the fault of the transformer to be measured according to the features corresponding to the initial vibration signal of the transformer to be measured to determine the current fault type of the transformer to be measured.

[0147] In one embodiment, before performing the steps in S100 above, as Figure 8 shown, the above method may further include:

[0148] S800. Obtain the original vibration signal of the transformer to be measured.

[0149] In practical applications, the RFID reader can receive the original vibration signal of the transformer to be measured sent by the antenna-enhanced RFID sensor in the transformer fault detection based on the antenna-enhanced RFID sensor.

[0150] S900. Preprocess the original vibration signal to obtain the initial vibration signal of the transformer to be measured.

[0151] Furthermore, the RFID reader can preprocess the original vibration signal of the transformer to be measured to obtain an effective initial vibration signal of the transformer to be measured. Optionally, the above preprocessing may include at least one of denoising processing, filtering processing, etc.

[0152] In the technical solution of the embodiment of the present application, in the case of the failure of the fault classification of the transformer to be measured, the initial vibration signal is added to the vibration signal training set to obtain a new training set, and the Hilbert envelope spectrum is input into the updated SSDA model to complete the fault classification of the transformer to be measured; in the case of fault anomaly detection, the above method can optimize and adjust the SSDA model again to ensure that the optimized SSDA model can flexibly cope with the diversity and complexity of fault types, improve the fault category coverage rate, achieve normal detection, improve the wide applicability and success rate of the transformer fault detection based on the antenna-enhanced RFID sensor, and reduce the risk of missed detection.

[0153] In one embodiment, the embodiment of the present application also provides a transformer fault detection method based on an antenna-enhanced RFID sensor, which is applied to an RFID reader. The method includes the following processes:

[0154] (1) Obtain the original vibration signal of the transformer to be measured;

[0155] (2) Preprocess the original vibration signal to obtain the initial vibration signal of the transformer to be measured;

[0156] (3) Obtain a plurality of simulated vibration signals obtained by adding a noise signal to the initial vibration signal;

[0157] (4) Perform empirical mode decomposition on each simulated vibration signal to obtain a plurality of intrinsic modes corresponding to the initial vibration signal;

[0158] (5) Perform similarity processing on each intrinsic mode and the initial vibration signal to obtain a similarity value between each intrinsic mode and the initial vibration signal;

[0159] (6) Screen out the intrinsic modes with similarity values greater than a preset threshold from each intrinsic mode as similar characteristic modes;

[0160] (7) Perform Hilbert transform processing on each similar characteristic mode to obtain the time-domain envelope spectrum of each similar characteristic mode;

[0161] (8) Perform fast Fourier transform on each time-domain envelope spectrum to obtain the frequency-domain envelope spectrum of each similar characteristic mode;

[0162] (9) Combine the frequency-domain envelope spectra to obtain the Hilbert envelope spectrum of the initial vibration signal;

[0163] (10) Input the Hilbert envelope spectrum into the pre-trained sparse denoising autoencoder SSDA model to obtain the features of the initial vibration signal;

[0164] Among them, the construction process of the SSDA model in step (10) above includes:

[0165] (101) Perform modal decomposition on the vibration signal training set to obtain multiple sets of similar characteristic modes corresponding to the vibration signal training set;

[0166] (102) Perform Hilbert transform processing according to each set of similar characteristic modes to determine multiple sample Hilbert envelope spectra of the vibration signal training set;

[0167] (103) Input each sample Hilbert envelope spectrum into the initial SSDA model to obtain a candidate SSDA model;

[0168] (104) In the case that the loss function value of the candidate SSDA model is greater than the preset loss value or the candidate SSDA model does not meet the convergence condition, use the quantum particle swarm QPSO algorithm to optimize the candidate SSDA model according to each sample Hilbert envelope spectrum to obtain a trained SSDA model;

[0169] (11) According to the features of the initial vibration signal, perform fault classification on the transformer under test to determine the current fault type of the transformer under test;

[0170] (12) In the case of failure of the fault classification of the transformer under test, add the initial vibration signal to the vibration signal training set to obtain a new training set;

[0171] (13) Use the new training set to optimize the SSDA model to obtain an updated SSDA model;

[0172] (14) Input the Hilbert envelope spectrum into the updated SSDA model to complete the fault classification of the transformer under test.

[0173] The execution processes of the above (1) to (14) can specifically refer to the description of the above embodiments, and their implementation principles and technical effects are similar, so they will not be elaborated here.

[0174] It should be understood that although the steps in the flowcharts involved in the above various embodiments are shown in sequence according to the indication of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above various embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments, and the execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of the steps or stages in other steps or other steps.

[0175] Based on the same inventive concept, an embodiment of the present application further provides a transformer fault detection device based on an antenna-enhanced radio frequency identification sensor for implementing the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor involved above. The implementation solution provided by this device to solve the problem is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more embodiments of the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided below can refer to the limitations on the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in the above text, and will not be repeated here.

[0176] In one embodiment, Figure 9 is a schematic structural diagram of a transformer fault detection device based on an antenna-enhanced radio frequency identification sensor in an embodiment of the present application. The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided by the embodiment of the present application can be applied to an RFID reader. As Figure 9 shown, the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor in the embodiment of the present application may include: a modal decomposition module 11, a transformation processing module 12, a feature acquisition module 13, and a fault classification module 14, where:

[0177] The modal decomposition module 11 is configured to perform modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal;

[0178] The transformation processing module 12 is configured to perform Hilbert transform processing according to each similar characteristic mode to determine the Hilbert envelope spectrum of the initial vibration signal;

[0179] The feature acquisition module 13 is configured to input the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder SSDA model to obtain the features of the initial vibration signal;

[0180] A fault classification module 14, configured to classify faults of a transformer to be measured according to the characteristics of an initial vibration signal, and determine the current fault type of the transformer to be measured.

[0181] The transformer fault detection device based on an antenna-enhanced RFID sensor provided by an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiments of the transformer fault detection method based on an antenna-enhanced RFID sensor of the present application. The implementation principles and technical effects are similar, and will not be elaborated here.

[0182] In one embodiment, the mode decomposition module 11 includes: a signal adding unit, a mode decomposition unit, a similarity processing unit, and a screening unit, where:

[0183] The signal adding unit is configured to obtain a plurality of simulated vibration signals obtained by adding a noise signal to an initial vibration signal;

[0184] The mode decomposition unit is configured to perform empirical mode decomposition on each simulated vibration signal to obtain a plurality of intrinsic modes corresponding to the initial vibration signal;

[0185] The similarity processing unit is configured to perform similarity processing on each intrinsic mode and the initial vibration signal to obtain a similarity value between each intrinsic mode and the initial vibration signal;

[0186] The screening unit is configured to screen out the intrinsic modes with similarity values greater than a preset threshold from each intrinsic mode as similar characteristic modes.

[0187] The transformer fault detection device based on an antenna-enhanced RFID sensor provided by an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiments of the transformer fault detection method based on an antenna-enhanced RFID sensor of the present application. The implementation principles and technical effects are similar, and will not be elaborated here.

[0188] In one embodiment, the transformation processing module 12 includes: a first transformation processing unit, a second transformation processing unit, and a combination unit, where:

[0189] The first transformation processing unit is configured to perform Hilbert transformation processing on each similar characteristic mode to obtain a time-domain envelope spectrum of each similar characteristic mode;

[0190] The second transformation processing unit is configured to perform fast Fourier transform on each time-domain envelope spectrum to obtain a frequency-domain envelope spectrum of each similar characteristic mode;

[0191] The combination unit is configured to combine each frequency-domain envelope spectrum to obtain a Hilbert envelope spectrum of the initial vibration signal.

[0192] The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided by an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of the present application. The implementation principle and technical effect are similar, and will not be elaborated here.

[0193] In one embodiment, the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor further includes: an envelope spectrum acquisition module, a training module, and a first optimization module, where:

[0194] The envelope spectrum acquisition module is configured to obtain corresponding multiple sample Hilbert envelope spectra according to the vibration signal training sets of multiple sample transformers;

[0195] The training module is configured to input each sample Hilbert envelope spectrum into an initial SSDA model to obtain a candidate SSDA model;

[0196] The first optimization module is configured to, when the loss function value of the candidate SSDA model is greater than a preset loss value or the candidate SSDA model does not meet the convergence condition, optimize the candidate SSDA model according to each sample Hilbert envelope spectrum by using a quantum particle swarm QPSO algorithm to obtain a trained SSDA model.

[0197] The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided by an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of the present application. The implementation principle and technical effect are similar, and will not be elaborated here.

[0198] In one embodiment, the envelope spectrum acquisition module is specifically configured to:

[0199] Perform modal decomposition on the vibration signal training set to obtain multiple corresponding similar characteristic mode sets of the vibration signal training set;

[0200] Perform Hilbert transform processing according to each similar characteristic mode set to determine multiple sample Hilbert envelope spectra of the vibration signal training set.

[0201] The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided by an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of the present application. The implementation principle and technical effect are similar, and will not be elaborated here.

[0202] In one embodiment, the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor includes: a signal addition module, a second optimization module, and a fault classification module, where:

[0203] A signal addition module, configured to add an initial vibration signal to a vibration signal training set to obtain a new training set in case of failure in fault classification of a transformer under test;

[0204] A second optimization module, configured to optimize the SSDA model by using the new training set to obtain an updated SSDA model;

[0205] A fault classification module, configured to input a Hilbert envelope spectrum into the updated SSDA model to complete fault classification of the transformer under test.

[0206] The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided in an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor. The implementation principle and technical effects are similar, and will not be described in detail here.

[0207] In one embodiment, the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor further includes: a signal acquisition unit and a preprocessing unit, where:

[0208] The signal acquisition unit is configured to acquire an original vibration signal of the transformer under test;

[0209] The preprocessing unit is configured to preprocess the original vibration signal to obtain an initial vibration signal of the transformer under test.

[0210] The transformer fault detection device based on an antenna-enhanced radio frequency identification sensor provided in an embodiment of the present application can be used to execute the technical solutions in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor. The implementation principle and technical effects are similar, and will not be described in detail here.

[0211] For specific limitations on the transformer fault detection device based on an antenna-enhanced radio frequency identification sensor, reference can be made to the limitations on the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor in the foregoing text, which will not be elaborated here. Each module in the above-mentioned transformer fault detection device based on an antenna-enhanced radio frequency identification sensor can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor in the computer device in the form of hardware or be independent of it, or can be stored in the memory in the computer device in the form of software, so as to be called by the processor to execute the operations corresponding to the above-mentioned respective modules.

[0212] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 10As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. Among them, the processor of the computer device is used to provide processing capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store the initial vibration signals of the transformer to be tested. The network interface of the computer device is used to communicate with an external endpoint via a network connection. When the computer program is executed by the processor, it implements a transformer fault detection method based on an antenna-enhanced radio frequency identification sensor.

[0213] Those skilled in the art can understand that Figure 10 the structure shown in is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0214] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, it implements the technical solution in the above-mentioned embodiment of the transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of this application. The implementation principle and technical effect are similar and will not be elaborated here.

[0215] In one embodiment, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the technical solution of the above-mentioned transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of this application. The implementation principle and technical effect are similar and will not be elaborated here.

[0216] In one embodiment, a computer program product is further provided, including a computer program. When the computer program is executed by the processor, it implements the technical solution of the above-mentioned transformer fault detection method based on an antenna-enhanced radio frequency identification sensor of this application. The implementation principle and technical effect are similar and will not be elaborated here.

[0217] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0218] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0219] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A transformer fault detection method based on an antenna-enhanced radio frequency identification sensor, characterized in that, The method includes: Performing modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal; Performing Hilbert transform processing on each of the similar characteristic modes to determine the Hilbert envelope spectrum of the initial vibration signal; Inputting the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder SSDA model to obtain the characteristics of the initial vibration signal; Classifying the faults of the transformer to be measured according to the characteristics of the initial vibration signal to determine the current fault type of the transformer to be measured.

2. The method according to claim 1, wherein The performing modal decomposition on the initial vibration signal of the transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal includes: Obtaining a plurality of simulated vibration signals obtained by adding noise signals to the initial vibration signal; Performing empirical modal decomposition on each of the simulated vibration signals to obtain a plurality of intrinsic modes corresponding to the initial vibration signal; Performing similarity processing on each of the intrinsic modes and the initial vibration signal to obtain the similarity values between each of the intrinsic modes and the initial vibration signal; Selecting the intrinsic modes with similarity values greater than a preset threshold from each of the intrinsic modes as the similar characteristic modes.

3. The method according to claim 1 or 2, characterized in that The performing Hilbert transform processing on each of the similar characteristic modes to determine the Hilbert envelope spectrum of the initial vibration signal includes: Performing Hilbert transform processing on each of the similar characteristic modes to obtain the time-domain envelope spectrum of each of the similar characteristic modes; Performing fast Fourier transform on each of the time-domain envelope spectra to obtain the frequency-domain envelope spectrum of each of the similar characteristic modes; Combining each of the frequency-domain envelope spectra to obtain the Hilbert envelope spectrum of the initial vibration signal.

4. The method according to claim 1 or 2, characterized in that, The construction process of the SSDA model includes: Obtaining a plurality of corresponding sample Hilbert envelope spectra according to the vibration signal training set of a plurality of sample transformers; Inputting each of the sample Hilbert envelope spectra into an initial SSDA model to obtain a candidate SSDA model; In the case where the loss function value of the candidate SSDA model is greater than a preset loss value or the candidate SSDA model does not reach the convergence condition, using the quantum particle swarm QPSO algorithm to optimize the candidate SSDA model according to each of the sample Hilbert envelope spectra to obtain the trained SSDA model.

5. The method according to claim 4, characterized in that The obtaining a plurality of corresponding sample Hilbert envelope spectra according to the vibration signal training set of a plurality of sample transformers includes: Performing modal decomposition on the vibration signal training set to obtain a plurality of corresponding sets of similar characteristic modes of the vibration signal training set; Performing Hilbert transform processing according to each of the sets of similar characteristic modes to determine a plurality of sample Hilbert envelope spectra of the vibration signal training set.

6. The method according to claim 1 or 2, characterized in that, The method further includes: In the case where the fault classification of the transformer to be measured fails, adding the initial vibration signal to the vibration signal training set to obtain a new training set; Using the new training set to perform optimization processing on the SSDA model to obtain an updated SSDA model; Inputting the Hilbert envelope spectrum into the updated SSDA model to complete the fault classification of the transformer to be measured.

7. The method according to claim 1 or 2, characterized in that, The method further includes: acquiring an original vibration signal of the transformer to be measured; performing preprocessing on the original vibration signal to obtain an initial vibration signal of the transformer to be measured.

8. A transformer fault detection device based on an antenna-enhanced radio frequency identification sensor, characterized in that, The device includes: a modal decomposition module, configured to perform modal decomposition on an initial vibration signal of a transformer to be measured to obtain a plurality of similar characteristic modes corresponding to the initial vibration signal; a transformation processing module, configured to perform Hilbert transform processing according to each of the similar characteristic modes to determine a Hilbert envelope spectrum of the initial vibration signal; a feature acquisition module, configured to input the Hilbert envelope spectrum into a pre-trained sparse denoising autoencoder SSDA model to obtain features of the initial vibration signal; a fault classification module, configured to perform fault classification on the transformer to be measured according to the features of the initial vibration signal to determine a current fault type of the transformer to be measured.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images