CN118211130A - GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method - Google Patents
GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method Download PDFInfo
- Publication number
- CN118211130A CN118211130A CN202410624069.5A CN202410624069A CN118211130A CN 118211130 A CN118211130 A CN 118211130A CN 202410624069 A CN202410624069 A CN 202410624069A CN 118211130 A CN118211130 A CN 118211130A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- convolution
- partial discharge
- discharge defect
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000005457 optimization Methods 0.000 claims abstract description 5
- 238000010606 normalization Methods 0.000 claims abstract description 4
- 238000005070 sampling Methods 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 10
- 230000004927 fusion Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000013256 Gubra-Amylin NASH model Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 abstract description 2
- 238000006243 chemical reaction Methods 0.000 abstract 1
- 238000010276 construction Methods 0.000 abstract 1
- 238000004088 simulation Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 5
- 238000003909 pattern recognition Methods 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/15—Correlation function computation including computation of convolution operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Mathematical Optimization (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Computing Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses a GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method, and belongs to the field of transformer ultrahigh frequency partial discharge signal data enhancement. The method specifically comprises the steps of data acquisition, normalization, wavelet soft threshold denoising, data cleaning, time-frequency graph conversion and data set construction of an ultrahigh frequency partial discharge defect signal; and the improved depth separable convolution structure SMixedConv is utilized to construct a data enhancement module DSWGAN-GP, and the data feature learning capacity of the model is improved through iterative training. Balance and expansion optimization of a small sample unbalanced data set is achieved. The model has smaller model parameters, the feature simulation capability of the model is improved by introducing the mixed convolution mode, feature learning and convergence can be completed in a smaller training frequency, and the model has better data learning capability and smaller model parameters and has practical engineering application value.
Description
Technical Field
The invention relates to the field of transformer ultrahigh frequency partial discharge signal mode identification, in particular to a GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method.
Background
The detection of partial discharge defects is used as a core method for evaluating the insulation state of power equipment such as transformers, and the technical level of detection of partial discharge defects is related to the safety operation and personal safety of the power equipment such as transformers, cables, switch cabinets and the like. Due to randomness of partial discharge signal discharge, partial discharge pattern recognition can be caused to have the problems of few samples and unbalanced data due to the collected partial discharge defect signal data set in the model training stage, and the problem greatly influences the model training efficiency and recognition accuracy of partial discharge pattern recognition. Current techniques for data enhancement and pattern recognition in combination with generation of countermeasure models in partial discharge pattern recognition have attracted a large number of scholars and experts to conduct research. Most of the researches focus on how to improve the training stability of the generated countermeasure model, and do not consider the problem that the size and parameter amount of the model need to be optimized while effectively learning the data characteristics in practical engineering application.
Therefore, those skilled in the art are dedicated to develop a GAN-based method for enhancing ultrahigh frequency partial discharge defect data of a transformer, so as to improve accuracy of partial discharge pattern recognition, reduce misjudgment probability, optimize a model structure of a data enhancement model, reduce parameter scale, and improve engineering application value of the model.
Disclosure of Invention
The invention aims to solve the problems and provide a GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows: the method specifically comprises the following steps:
1. A GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method is characterized by comprising the following steps of: the method specifically comprises the following steps:
step 1, collecting ultrahigh frequency partial discharge defect signal data of a transformer through equipment such as a receiver, an antenna and the like;
Step 2, carrying out data preprocessing on the partial discharge defect signal data, wherein the data preprocessing comprises data normalization, denoising and pulse signal extraction, converting a single pulse signal into a time-frequency diagram, constructing a partial discharge defect signal data set, and dividing the partial discharge defect signal data set into a training set and a test set according to a proportion;
Step 3, introducing Wasserstein distance and gradient punishment mechanism on the traditional GAN model structure to improve training stability of the model, constructing a data enhancement model DSWGAN-GP by combining an improved depth separable structure SMixedConv, training by adopting the data set in the step 2, updating model weight parameters, improving feature learning capacity of a generator model, and evaluating the generating capacity of the model after training is finished to obtain a data enhancement model DSWGAN-GP with optimal generating capacity;
Specifically, the specific details of building the data enhancement model DSWGAN-GP in combination with the improved depth separable structure SMixedConv are:
The generator structure in the data enhancement model DSWGAN-GP is divided into five up-sampling modules, and the discriminator structure is divided into five down-sampling modules;
A first up-sampling module of a generator of the data enhancement module adopts transposition convolution to finish preliminary data characteristic extraction and up-sampling; upsampling in the second, third, and fourth upsampling modules using bicubic interpolation and feature extracting the data using the improved depth separable convolution structure SMixedConv; the final upsampling section is completed in a fifth upsampling module using transpose convolution. In the discriminator section, a data enhancement model DSWGAN-GP is constructed using a modified depth separable convolution SMixedConv in the second, third, and fourth downsampling modules of the discriminator instead of the conventional convolution.
The specific steps of the improved depth separable convolution include: optimizing a single convolution kernel size channel-by-channel convolution structure of the depth separable convolution, introducing a mixed convolution mode, dividing an input channel and an output channel into different groups according to a mean value division mode, mapping the input channel and the output channel according to the number of the mixed convolution groups, carrying out convolution on feature vectors by using different convolution kernel sizes in each group, obtaining multi-scale feature information by carrying out convolution on the feature vectors by using the convolution kernel sizes, and completing feature information fusion by carrying out point-by-point convolution. The large-size convolution kernel can pay attention to global information of data, the small-size convolution kernel can pay attention to local details of the data, model parameters can be effectively reduced by combining a convolution mode of channel grouping, and multi-scale feature information extraction and fusion are achieved.
Wherein the mathematical formula of the group convolution is:
where i, j is the spatial position index of the output feature map, m is the intra-group input channel index, k is the output channel index, p, q is the spatial position index of the convolution kernel, filter g is the convolution kernel in the g-th group, and the size is (C in/G, Cin /G,K,K）,bg,k is the offset of the k-th output channel in the g-th group, input g is the input feature vector in the g-th group, output g is the output feature vector in the g-th group.
And 4, calling the trained model through the data enhancement module to generate defect data, expanding the data, balancing the data set and realizing the optimization enhancement of the partial discharge defect data.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention can solve the problems of small defect data quantity and unbalanced data among different defect types faced by partial discharge in actual engineering, provides a data basis for training of a deep learning architecture of a large model, can more fully mine the performance advantage of the large model, improves the accuracy of partial discharge identification, and reduces the false judgment probability;
2. The network parameters of the generated model are reduced by combining the depth separable convolution and the bicubic interpolation upsampling method, the multi-scale feature extraction function is added, the feature learning capacity of the generator is improved, the partial discharge defect data features are learned in a smaller training round under the condition that the convergence of the generated countermeasure model is ensured, and then the model optimization method can be further explored through the research foundation, so that the method is better for engineering actual service.
3. The data enhancement model has smaller parameter quantity, effectively reduces the model scale on the basis of realizing data expansion, can effectively reduce the hardware pressure in practical application, and has more engineering practical value.
Drawings
FIG. 1 is a flow chart of a method for enhancing partial discharge defect signal data according to the present invention;
FIG. 2 is a waveform and a time-frequency chart of different types of partial discharge defect signals extracted by the invention;
FIG. 3 is a diagram showing the comparison of SMixedConv and DSConv structures according to the present invention;
FIG. 4 is a block diagram of a data enhancement model DSWGAN-GP in the present invention;
FIG. 5 is a training loss diagram of the data enhancement model DSWGAN-GP in the present invention;
FIG. 6 is a graph showing the effect of pictures generated at different training stages of DSWGAN-GP model in the present invention;
Detailed Description
The invention is further described in connection with the following detailed description, in order to make the technical means, the creation characteristics, the achievement of the purpose and the effect of the invention easy to understand.
The embodiment of the invention discloses a GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method, which comprises the following steps as shown in fig. 1:
step 1: the method comprises the steps of collecting ultrahigh frequency partial discharge defect signal data of a transformer through equipment such as a receiver, an antenna and the like;
Step 2: performing data preprocessing on the partial discharge defect signal data, wherein the data preprocessing comprises data normalization, denoising and pulse signal extraction, converting a single pulse signal into a time-frequency diagram, and constructing a partial discharge defect signal data set;
The denoising method adopts a wavelet soft threshold method to denoise, the wavelet is selected as db6, the decomposition times are 6, and the threshold is set to be 0.08; pulse extraction is carried out on the denoised signals, the occurrence position of the partial discharge pulse is judged by adopting a double-threshold method, a complete pulse waveform is obtained by taking points before and after the point, data cleaning is carried out on a waveform data set, and unqualified data and extra pulse interference are removed; the time-frequency diagram of the pulses with different defect partial discharge categories is obtained by wavelet transformation, as shown in fig. 2, different waveforms of the pulses with different partial discharge can be observed, and different frequency ranges are provided in the wavelet time-frequency diagram, the characteristics can be used as a classifying basis for distinguishing the different pulses, a time-frequency diagram data set with different categories is constructed according to pulse data, and the time-frequency diagram data set with different categories is prepared according to the following steps of 4:1, dividing a training set and a testing set in proportion;
Step 3, introducing Wasserstein distance and gradient punishment mechanism on the traditional GAN model structure to improve training stability of the model, constructing a data enhancement model DSWGAN-GP by combining an improved depth separable structure SMixedConv, training by adopting the data set in the step 2, updating model weight parameters, improving feature learning capacity of a generator model, and evaluating the generating capacity of the model after training is finished to obtain a data enhancement model DSWGAN-GP with optimal generating capacity;
Specifically, the specific details of building the data enhancement model DSWGAN-GP in combination with the improved depth separable structure SMixedConv are:
According to the illustration of fig. 4, the generator structure in the data enhancement model DSWGAN-GP is divided into five upsampling modules and the discriminator structure into five downsampling modules;
A first up-sampling module of a generator of the data enhancement module adopts transposition convolution to finish preliminary data characteristic extraction and up-sampling; upsampling in the second, third, and fourth upsampling modules using bicubic interpolation and feature extracting the data using the improved depth separable convolution structure SMixedConv; the final upsampling section is completed in a fifth upsampling module using transpose convolution. In the discriminator section, a data enhancement model DSWGAN-GP is constructed using a modified depth separable convolution SMixedConv in the second, third, and fourth downsampling modules of the discriminator instead of the conventional convolution.
The specific steps of the improved depth separable convolution include: optimizing a single convolution kernel size channel-by-channel convolution structure of the depth separable convolution, introducing a mixed convolution mode, dividing an input channel and an output channel into different groups according to a mean value division mode, mapping the input channel and the output channel according to the number of the mixed convolution groups, carrying out convolution on feature vectors by using different convolution kernel sizes in each group, obtaining multi-scale feature information by carrying out convolution on the feature vectors by using the convolution kernel sizes, and completing feature information fusion by carrying out point-by-point convolution. The large-size convolution kernel can pay attention to global information of data, the small-size convolution kernel can pay attention to local details of the data, model parameters can be effectively reduced by combining a convolution mode of channel grouping, and multi-scale feature information extraction and fusion are achieved.
Wherein the mathematical formula of the group convolution is:
(1)
where i, j is the spatial position index of the output feature map, m is the intra-group input channel index, k is the output channel index, p, q is the spatial position index of the convolution kernel, filter g is the convolution kernel in the g-th group, and the size is (C in/G, Cin /G,K,K）,bg,k is the offset of the k-th output channel in the g-th group, input g is the input feature vector in the g-th group, output g is the output feature vector in the g-th group.
As can be seen from the comparison of Table 1, the total model parameters of the up-sampling part in the data enhancement module are reduced from 3,572,736 to 1,009,984 by about 3.5 times after the improvement of the invention. As can be seen from FIG. 6, the data enhancement model of the present invention can capture the key features of the partial discharge defect signal data through the five up-sampling modules in the generator and the five down-sampling modules in the discriminator in a smaller training frequency, and compared with the conventional model, the data enhancement model can generate matrix noise interference in a smaller training frequency, and requires a larger number of iterations to eliminate the situation, and the improved model has a better generation effect, has fewer parameters and has a better engineering application value.
Table 1 model parameter comparison
| Model | Up-sampling module total parameter | Model total parameters |
| WGAN-GP | 3,572,736 | 3,574,656 |
| DSWGAN-GP | 1,009,984 | 1,011,904 |
Observing the convergence condition of the generator and the discriminator through the change condition of the loss function during model training, and stopping training after training iteration for a certain number of times is needed when the model converges;
Model performance evaluation: and calling the trained model to input random noise into the model to obtain the data-enhanced partial discharge defect signal data, and evaluating the training effect of the generator by calculating the generator performance evaluation indexes such as SSIM values between the generated partial discharge defect signal data and the original data. In the experiment, model weights after training convergence are loaded into a model for data enhancement effect evaluation, two defect types are selected as experimental objects, 50 time-frequency graphs are generated respectively, SSIM value calculation is carried out on the data in an original data set, the model generation capacity is evaluated, the comparison result is shown in a table 2, the SSIM value between the data of the two defect types subjected to data enhancement through the method and the original data is more than 99%, and the method can be used as an example of data expansion.
Table 2 calculation of SSIM value evaluation model Generation Capacity
| Defect type | Number of generations | SSIM value |
| Metal discharge | 50 | 99.989% |
| Creeping discharge | 50 | 99.992 % |
And 4, calling the trained model through the data enhancement module to generate defect data, expanding the data, balancing the data set and realizing the optimization enhancement of the partial discharge defect data.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.
Claims (1)
1. A GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method is characterized by comprising the following steps of: the method specifically comprises the following steps:
step 1, collecting ultrahigh frequency partial discharge defect signal data of a transformer through equipment such as a receiver, an antenna and the like;
Step 2, carrying out data preprocessing on the partial discharge defect signal data, wherein the data preprocessing comprises data normalization, denoising and pulse signal extraction, converting a single pulse signal into a time-frequency diagram, constructing a partial discharge defect signal data set, and dividing the partial discharge defect signal data set into a training set and a test set according to a proportion;
Step 3, introducing Wasserstein distance and gradient punishment mechanism on the traditional GAN model structure to improve training stability of the model, constructing a data enhancement model DSWGAN-GP by combining an improved depth separable structure SMixedConv, training by adopting the data set in the step 2, updating model weight parameters, improving feature learning capacity of a generator model, and evaluating the generating capacity of the model after training is finished to obtain a data enhancement model DSWGAN-GP with optimal generating capacity;
Specifically, the specific details of building the data enhancement model DSWGAN-GP in combination with the improved depth separable structure SMixedConv are: the generator structure in the data enhancement model DSWGAN-GP is divided into five up-sampling modules, and the discriminator structure is divided into five down-sampling modules;
A first up-sampling module of a generator of the data enhancement module adopts transposition convolution to finish preliminary data characteristic extraction and up-sampling; upsampling in the second, third, and fourth upsampling modules using bicubic interpolation and feature extracting the data using the improved depth separable convolution structure SMixedConv; a final upsampling section is completed in a fifth upsampling module using transpose convolution; in the discriminator section, a data enhancement model DSWGAN-GP is constructed using a modified depth separable convolution structure SMixedConv in the second, third, and fourth downsampling modules of the discriminator instead of the conventional convolution structure;
The specific steps of the improved depth separable convolution include: optimizing a single convolution kernel size channel-by-channel convolution structure of the depth separable convolution, introducing a mixed convolution mode, dividing an input channel and an output channel into different groups according to a mean division mode, mapping the input channel and the output channel according to the number of the mixed convolution groups, carrying out convolution on feature vectors by using different convolution kernel sizes in each group, obtaining multi-scale feature information by using convolution kernel feature vectors of different sizes, and completing feature information fusion by point-by-point convolution; the large-size convolution kernel can pay attention to global information of the data, the small-size convolution kernel can pay attention to local details of the data, model parameters can be effectively reduced by combining a convolution mode of channel grouping, and multi-scale feature information extraction and fusion are realized;
Wherein the mathematical formula of the group convolution is:
Where i, j is the spatial position index of the output feature map, m is the intra-group input channel index, k is the output channel index, p, q is the spatial position index of the convolution kernel, filter g is the convolution kernel in the g-th group, the size is (C in/G, Cin /G,K,K）,bg,k is the offset of the k-th output channel in the g-th group, input g is the input feature vector in the g-th group, output g is the output feature vector in the g-th group;
and 4, calling the trained model through the data enhancement module to generate defect data, expanding the data, balancing the data set and realizing the optimization enhancement of the partial discharge defect data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410624069.5A CN118211130B (en) | 2024-05-20 | 2024-05-20 | GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410624069.5A CN118211130B (en) | 2024-05-20 | 2024-05-20 | GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118211130A true CN118211130A (en) | 2024-06-18 |
| CN118211130B CN118211130B (en) | 2024-07-12 |
Family
ID=91449115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410624069.5A Active CN118211130B (en) | 2024-05-20 | 2024-05-20 | GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118211130B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037001A (en) * | 2021-10-11 | 2022-02-11 | 中国人民解放军92578部队 | Mechanical pump small sample fault diagnosis method based on WGAN-GP-C and metric learning |
| CN115908842A (en) * | 2022-11-02 | 2023-04-04 | 淮南市潘阳光伏发电有限公司 | Transformer partial discharge data enhancement and identification method |
| CN117332268A (en) * | 2023-10-08 | 2024-01-02 | 上海交通大学 | GIS partial discharge data enhancement method and system for improving GAN |
| CN117496358A (en) * | 2023-12-29 | 2024-02-02 | 中国石油大学(华东) | Neural network construction method for feature enhancement based on laser radar data |
-
2024
- 2024-05-20 CN CN202410624069.5A patent/CN118211130B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037001A (en) * | 2021-10-11 | 2022-02-11 | 中国人民解放军92578部队 | Mechanical pump small sample fault diagnosis method based on WGAN-GP-C and metric learning |
| CN115908842A (en) * | 2022-11-02 | 2023-04-04 | 淮南市潘阳光伏发电有限公司 | Transformer partial discharge data enhancement and identification method |
| CN117332268A (en) * | 2023-10-08 | 2024-01-02 | 上海交通大学 | GIS partial discharge data enhancement method and system for improving GAN |
| CN117496358A (en) * | 2023-12-29 | 2024-02-02 | 中国石油大学(华东) | Neural network construction method for feature enhancement based on laser radar data |
Non-Patent Citations (1)
| Title |
|---|
| YANG LIU 等: "A Study on the Localization of Partial Discharge Using Ngccphat CNN", 2023 3RD INTERNATIONAL CONFERENCE ON NEW ENERGY AND POWER ENGINEERING (ICNEPE), 16 February 2024 (2024-02-16), pages 635 - 639 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118211130B (en) | 2024-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860982B (en) | VMD-FCM-GRU-based wind power plant short-term wind power prediction method | |
| CN113935460A (en) | Intelligent diagnosis method for mechanical fault under class imbalance data set | |
| CN109890043B (en) | Wireless signal noise reduction method based on generative countermeasure network | |
| CN113094993B (en) | Modulation signal denoising method based on self-coding neural network | |
| CN112287784B (en) | Radar signal classification method based on deep convolutional neural network and feature fusion | |
| CN111914705A (en) | Signal generation method and device for improving health state evaluation accuracy of reactor | |
| CN113488060B (en) | Voiceprint recognition method and system based on variation information bottleneck | |
| CN109655266B (en) | Wind turbine generator bearing fault diagnosis method based on AVMD and spectrum correlation analysis | |
| CN111624570B (en) | Radar target identification method based on two-dimensional convolutional neural network | |
| CN103995950A (en) | Wavelet coefficient partial discharge signal noise elimination method based on related space domain correction threshold values | |
| CN112613493A (en) | Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM | |
| CN111010356A (en) | Underwater acoustic communication signal modulation mode identification method based on support vector machine | |
| Qin et al. | Radar waveform recognition based on deep residual network | |
| CN117473414A (en) | Bearing fault position identification method based on low-noise time-frequency image | |
| CN112926728A (en) | Small sample turn-to-turn short circuit fault diagnosis method for permanent magnet synchronous motor | |
| CN106548031A (en) | A kind of Identification of Modal Parameter | |
| CN117390508B (en) | Hydroelectric generating set signal state identification method based on time-shifting multi-scale cosine similarity entropy | |
| CN116863959B (en) | Dolphin sound generating method based on generating countermeasure network | |
| CN118211130B (en) | GAN-based transformer ultrahigh frequency partial discharge defect data enhancement method | |
| CN115356599B (en) | Multi-mode urban power grid fault diagnosis method and system | |
| CN112183260A (en) | One-way valve fault diagnosis method based on total variation noise reduction and RQA | |
| CN116257737A (en) | Novel power transmission line high-frequency fault signal noise reduction method based on automatic encoder | |
| CN102509268B (en) | Immune-clonal-selection-based nonsubsampled contourlet domain image denoising method | |
| CN115563500A (en) | Power distribution equipment partial discharge mode identification method, device and system based on data enhancement technology | |
| CN115169407B (en) | Weak transient signal detection method and system based on time domain depth feature learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |