CN117949794B - Cable partial discharge fault detection method - Google Patents
Cable partial discharge fault detection method Download PDFInfo
- Publication number
- CN117949794B CN117949794B CN202410355793.2A CN202410355793A CN117949794B CN 117949794 B CN117949794 B CN 117949794B CN 202410355793 A CN202410355793 A CN 202410355793A CN 117949794 B CN117949794 B CN 117949794B
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- convolution
- convolution stage
- sample
- mixed
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 10
- 230000007547 defect Effects 0.000 claims abstract description 24
- 238000004519 manufacturing process Methods 0.000 claims abstract description 15
- 230000007246 mechanism Effects 0.000 claims abstract description 13
- 238000005520 cutting process Methods 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 13
- 238000000034 method Methods 0.000 claims description 11
- 238000011176 pooling Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 11
- 238000009413 insulation Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 3
- 230000006872 improvement Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 6
- 239000002245 particle Substances 0.000 description 3
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 239000002923 metal particle Substances 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 229910001111 Fine metal Inorganic materials 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 239000003822 epoxy resin Substances 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 239000011810 insulating material Substances 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 229920000647 polyepoxide Polymers 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Testing Relating To Insulation (AREA)
Abstract
The invention belongs to the technical field of cable production and application, and particularly relates to a cable partial discharge fault detection method. According to the invention, PRPD gray level images of the discharge defects are obtained by utilizing simulated four types of cable partial discharge defects, then, the number of samples is increased by cutting and matching with a linear superposition mode so as to improve the accuracy of subsequent classification, meanwhile, the input samples are weighted by matching with a multi-space mixed attention mechanism, so that the model is self-adaptively focused on the characteristic areas of four different types of partial discharge, and further, the division is better, finally, the loss function is improved, the weight parameters are punished by utilizing a regular coefficient, the overfitting is avoided, the required classification is finally obtained, and the help is provided for the improvement and adjustment of the subsequent production process.
Description
Technical Field
The invention belongs to the technical field of cable production and application, and particularly relates to a cable partial discharge fault detection method.
Background
The power cable is little influenced by natural factors such as weather and the like in the process of conveying electricity, and has low maintenance cost, so that the power cable is widely applied to power transmission engineering in China. For this reason, the performance of the cable is related to the stability of the power system.
In the existing power cable design and manufacturing process, many problems occur such that the power cable insulation is more or less damaged, thereby causing partial discharge. For example, during production and operation, the power cable may be subjected to operations such as cutting or rubbing, and fine metal particles may be generated during these operations and remain inside the power cable. After the power cable is electrified, an electric field is formed around the power cable, and the residual metal particles can obtain energy under the action of the electric field, so that jump or displacement can occur to trigger partial discharge; or air may be present inside the insulating layer during manufacture and voids may be present between the insulating material and other layers during shrinkage, both of which may cause air gaps to be present inside the power cable, thereby inducing partial discharge.
Currently, there are mainly four types of partial discharges that can exist during the production of cables: particle discharge, tip discharge, creeping discharge, and air gap discharge. These four modes of partial discharge are all caused by different production reasons. In the production of the existing power cable, a pulse current method is mainly adopted to detect whether partial discharge exists, but for the type of the partial discharge, a complex mode is needed to determine, so that how to rapidly determine the type of the partial discharge, and aiming at the reason of the generation of the partial discharge, the improvement and adjustment of the production process are problems to be solved by the current cable production enterprises.
Disclosure of Invention
Aiming at the technical problem that the type of the cable with partial discharge is rapidly determined after the existing cable is produced, the invention provides the cable partial discharge fault detection method which is simple in method, convenient to operate and capable of rapidly determining the type of the partial discharge.
In order to achieve the above purpose, the invention adopts the following technical scheme: the invention provides a cable partial discharge fault detection method, which comprises the following steps:
a. firstly, building a cable partial discharge test platform, manufacturing four common partial discharge defects and collecting partial discharge original signals;
b. extracting power frequency phase of pulse of effective partial discharge original signal Discharge amountAnd the number of dischargesConstructing four PRPD gray-scale patterns of four insulation defect types;
c. the method comprises the steps of randomly cutting out pictures of a part of areas from PRPD gray images of the same type, linearly superposing the pictures with the same areas of the other three PRPD gray images to obtain a generated sample, and forming a training sample set by the generated sample and a real sample;
d. constructing a depth residual error network model structure, dividing a training sample set into a training set and a testing set according to the proportion of 7:3, and training and testing the depth residual error network model structure;
e. finally, identifying the partial discharge defect by using the trained depth residual error network model structure;
In the step d, a depth residual error network model structure is constructed, wherein the depth residual error network model structure comprises an input layer, a first convolution stage, a multi-space mixed attention mechanism and a maximum pooling layer which are sequentially connected in sequence, a second convolution stage, a third convolution stage, a fourth convolution stage, a fifth convolution stage, a serial mixed attention mechanism, an average pooling layer, a full connection layer and an output layer, the multi-space mixed attention mechanism is that feature information extracted from a space domain is multiplied by original input features to obtain space domain mixed information, feature information extracted from a channel domain is multiplied by original input to obtain mixed information of the channel domain, then the mixed information of the channel domain is subjected to the space domain to extract feature information to obtain mixed attention feature information, and finally the obtained space domain mixed information is multiplied by the mixed attention feature information to obtain complete multi-space mixed attention feature information.
Preferably, in the step d, the loss function of the depth residual network model structure is:
wherein, the method comprises the steps of, wherein, For the number of samples to be taken,In the case of a sample,Is of the sample typeIs used to determine the predicted value of (c),For the sampleIs used to determine the actual value of (c) in the (c),For the coefficients of the regular term(s),Is a weight parameter.
Preferably, the input layer, the first convolution stage, the maximum pooling layer, the second convolution stage, the third convolution stage, the fourth convolution stage, the fifth convolution stage, the average pooling layer, the full connection layer, and the output layer are arranged according to a ResNet structure.
Preferably, the residual structures of the second convolution stage, the third convolution stage, the fourth convolution stage and the fifth convolution stage are all convolution layers with a convolution kernel 1*1 and a step length of 2 added on the short circuit.
Compared with the prior art, the invention has the advantages and positive effects that:
1. The invention provides a cable partial discharge fault detection method, which utilizes simulated four types of cable partial discharge defects to obtain PRPD gray level images of the discharge defects, then increases the number of samples by cutting and matching with a linear superposition mode so as to improve the accuracy of subsequent classification, and simultaneously weights input samples by matching with a multi-space mixed attention mechanism so that a model adaptively pays attention to characteristic areas of four different types of partial discharge, thereby better classifying, and finally, the loss function is improved, punishment is carried out on weight parameters by utilizing a regular coefficient so as to avoid overfitting, finally, the required classification is obtained, and the method provides help for the improvement and adjustment of the subsequent production process.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort to a person skilled in the art.
Fig. 1 is a schematic diagram of a cable partial discharge fault detection method provided in embodiment 1;
fig. 2 is a schematic structural diagram of a depth residual network model structure provided in embodiment 1;
fig. 3 is a residual block diagram of a depth residual network model structure provided in embodiment 1;
Fig. 4 is a block diagram of the multi-spatial mixed attention mechanism provided in embodiment 1.
Detailed Description
In order that the above objects, features and advantages of the application will be more clearly understood, a further description of the application will be rendered by reference to the appended drawings and examples. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced otherwise than as described herein, and therefore the present invention is not limited to the specific embodiments of the disclosure that follow.
In embodiment 1, as shown in fig. 1 to 4, the present embodiment aims to provide a method for rapidly detecting a partial discharge fault of a cable, so as to facilitate determination of a type of partial discharge, thereby correcting a problem generated in a production process.
Therefore, the method for rapidly detecting the cable partial discharge fault provided by the embodiment firstly builds a cable partial discharge test platform, and makes four common partial discharge defects and collects partial discharge original signals, and the four partial discharge defect construction methods mainly make defects in an artificial mode, for example, air gap discharge is to prick a plurality of micropores on the surface of an insulating layer, so that a small amount of air enters the inside of the insulating layer, and then the insulating layer is sealed by epoxy resin, and particle discharge is an artificial construction defect in a mode of attaching metal chips and the like on the main insulating surface at one end of the cable, so that different reasons of the partial discharge defects are formed.
By constructing a cable partial discharge test platform, the partial discharge original data under four insulation defects are obtained.
As the variability between PRPD spectra is more pronounced. For this reason, in the present embodiment, the power frequency phase of the pulse of the effective partial discharge original signal is extractedDischarge amountAnd the number of dischargesFour PRPD gray-scale maps of four insulation defect types are constructed, and if the voltages applied to the same insulation defect are different in test, the partial discharge caused by the insulation defect is also different, so that the four PRPD gray-scale maps of the same type of partial discharge defect are obtained by adjusting different voltages, and the problem of under fitting is avoided.
To further increase the number of samples, in the present embodiment, the partial discharge signal in consecutive 200 ac cycles is converted into one PRPD gray scale map. Phase of power frequencyThe axis is divided into 360 sections, and the discharge amountThe axis is divided into 128 intervals, then the wholeThe plane is divided into cells of 128 x 360, and the discharge times in each cell are counted for each sample signal. Then normalizing the discharge times, and finally counting the discharge pointsRepresented by gray scale images, 700 images can thus be acquired for each type.
Therefore, in this embodiment, for the characteristics of the partial discharge PRPD gray-scale image, a picture of a part of the region is arbitrarily cut out from the PRPD gray-scale image of the same type, and then the picture is linearly overlapped with the same region of the other three PRPD gray-scale images to obtain a generated sample, so that the generated sample does not generate a mixed condition, and meanwhile, if only one region is cut out, the sample amount can be increased by 6 times, so that huge sample data can be obtained through cutting of different regions, thereby avoiding the occurrence of the problem of under fitting. And finally, the generated sample and the real sample together form a training sample set.
The depth residual network model can autonomously extract and learn the characteristics in the sample, avoids the subjective influence of manually extracting the characteristics, and has the characteristics of strong robustness and easy training and optimization.
In this embodiment, the depth residual network model adopts the existing ResNet structure, and considering the characteristics of the sample, in this embodiment, the existing ResNet structure is improved to improve the classification effect. In the conventional ResNet structure, the input layer, the first convolution stage, the maximum pooling layer, the second convolution stage, the third convolution stage, the fourth convolution stage, the fifth convolution stage, the average pooling layer, the full connection layer and the output layer are sequentially connected, and in this embodiment, the output number of the output layer is 4.
In this embodiment, in order to weight different regions of the input feature map, the model adaptively focuses on detailed information of the different regions, so as to improve accuracy of classification of partial discharge defects, and attention mechanisms are added after a first convolution stage and a fifth convolution stage of the ResNet model, respectively.
Because the first convolution stage does not contain residual blocks, convolution, regularization, activation function and maximum pooling are mainly performed on the input, background information is not needed to be considered in the PRPD gray level diagram, and for obtaining important information in the feature diagram, in this embodiment, a multi-space mixed attention mechanism is designed, namely, the feature information extracted by a space domain is multiplied by the original input feature to obtain space domain mixed information, the feature information extracted by a channel domain is multiplied by the original input to obtain mixed information of the channel domain, then the mixed information of the channel domain is subjected to the space domain extraction feature information to obtain mixed attention feature information, and finally, the obtained space domain mixed information is multiplied by the mixed attention feature information to obtain complete multi-space mixed attention feature information.
In this way, the characteristic information of two dimensions is realized by utilizing the prior spatial domain mixed information and the channel domain mixed information without mutual interference, and then the local information is more emphasized by adding the characteristic information extracted by the spatial domain, so that the extraction of the local characteristics is ensured.
After five-layer convolution layer operation, different areas of the input feature map are weighted and processed by using a serial mixed attention mechanism (CBAM), so that the model adaptively focuses on the detailed information of the defect area, and the accuracy of defect classification is improved.
Meanwhile, in order to extract image features from multiple dimensions, the embodiment also provides a specific existing residual structure, specifically, a convolution layer with a convolution kernel 1*1 and a step length of 2 is added on the input feature x, so that the convolution layer can change the dimension of the input feature map, and the residual network can extract image features from multiple dimensions, thereby forming better cooperation with the multiple space mixed attention mechanism.
Meanwhile, in order to update the weight parameters and the bias better, in this embodiment, the loss function is further adjusted, specifically, the loss function is: wherein, the method comprises the steps of, wherein, For the number of samples to be taken,In the case of a sample,Is of the sample typeIs used to determine the predicted value of (c),For the sampleIs used to determine the actual value of (c) in the (c),For the coefficients of the regular term(s),Is a weight parameter. Because the depth of the model deepens, penalty for weight parameters is added into the loss function model in order to avoid the problem of over fitting, and the model can avoid the phenomenon of over fitting to a certain extent by setting the penalty, so that the generalization performance of the model is improved.
And then, dividing the training sample set into a training set and a testing set according to the ratio of 7:3, and training and testing the depth residual error network model structure.
And finally, identifying the partial discharge defect by using the trained depth residual error network model structure.
By the arrangement, classification accuracy for particle discharge, tip discharge, creeping discharge and air gap discharge respectively reaches 97.23%, 96.78%, 95.89% and 98.32%.
The present invention is not limited to the above-mentioned embodiments, and any equivalent embodiments which can be changed or modified by the technical content disclosed above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above-mentioned embodiments according to the technical substance of the present invention without departing from the technical content of the present invention still belong to the protection scope of the technical solution of the present invention.
Claims (3)
1. The cable partial discharge fault detection method is characterized by comprising the following steps of:
a. firstly, building a cable partial discharge test platform, manufacturing four common partial discharge defects and collecting partial discharge original signals;
b. extracting power frequency phase of pulse of effective partial discharge original signal Discharge amount/>And number of discharges/>Constructing four PRPD gray-scale patterns of four insulation defect types;
c. the method comprises the steps of randomly cutting out pictures of a part of areas from PRPD gray images of the same type, linearly superposing the pictures with the same areas of the other three PRPD gray images to obtain a generated sample, and forming a training sample set by the generated sample and a real sample;
d. constructing a depth residual error network model structure, dividing a training sample set into a training set and a testing set according to the proportion of 7:3, and training and testing the depth residual error network model structure;
e. finally, identifying the partial discharge defect by using the trained depth residual error network model structure;
In the step d, the depth residual error network model structure is constructed, which comprises an input layer, a first convolution stage, a multi-space mixed attention mechanism, a maximum pooling layer, a second convolution stage, a third convolution stage, a fourth convolution stage, a fifth convolution stage, a serial mixed attention mechanism, an average pooling layer, a full connection layer and an output layer which are sequentially connected, wherein the multi-space mixed attention mechanism is that feature information extracted from a space domain is multiplied by original input features to obtain space domain mixed information, feature information extracted from a channel domain is multiplied by original input to obtain mixed information of the channel domain, then the mixed information of the channel domain is subjected to space domain extraction feature information to obtain mixed attention feature information, finally, the obtained space domain mixed information is multiplied by the mixed attention feature information to obtain complete multi-space mixed attention feature information, and in the step d, a loss function of the depth residual error network model structure is as follows:
，
Wherein, For the number of samples,/>For the sample,/>For/>, of the samplePredicted value of/>For sample/>Actual value of/>Is a regular term coefficient,/>Is a weight parameter.
2. The cable partial discharge fault detection method of claim 1, wherein the input layer, the first convolution stage, the max-pooling layer, the second convolution stage, the third convolution stage, the fourth convolution stage, the fifth convolution stage, the average pooling layer, the full-connection layer, and the output layer are arranged in a ResNet structure.
3. The method of claim 2, wherein the residual structures of the second, third, fourth and fifth convolution stages are all convolution layers with a convolution kernel 1*1 and a step size of 2 added to the short circuit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410355793.2A CN117949794B (en) | 2024-03-27 | 2024-03-27 | Cable partial discharge fault detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410355793.2A CN117949794B (en) | 2024-03-27 | 2024-03-27 | Cable partial discharge fault detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117949794A CN117949794A (en) | 2024-04-30 |
| CN117949794B true CN117949794B (en) | 2024-06-04 |
Family
ID=90794749
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410355793.2A Active CN117949794B (en) | 2024-03-27 | 2024-03-27 | Cable partial discharge fault detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117949794B (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000214211A (en) * | 1999-01-26 | 2000-08-04 | Furukawa Electric Co Ltd:The | Partial discharge judging method |
| CN103926512A (en) * | 2014-03-25 | 2014-07-16 | 国家电网公司 | Method for simulating XLPE insulated cable partial discharge and simulation testing platform |
| CN108680836A (en) * | 2018-04-26 | 2018-10-19 | 天津大学 | Multiple physical field based on high-frequency method couples paper oil insulation partial discharge detecting system |
| CN109100627A (en) * | 2018-10-31 | 2018-12-28 | 红相股份有限公司 | A kind of power equipment partial discharges fault diagnostic method based on end-to-end mode |
| CN112149549A (en) * | 2020-09-18 | 2020-12-29 | 国网山东省电力公司泰安供电公司 | GIS partial discharge type identification method based on depth residual error network |
| CN113889198A (en) * | 2021-09-24 | 2022-01-04 | 国网宁夏电力有限公司电力科学研究院 | Transformer fault diagnosis method and equipment based on oil chromatogram time-frequency domain information and residual error attention network |
| KR20220023545A (en) * | 2020-08-21 | 2022-03-02 | 한국전력공사 | Apparatus and method for diagnosing partial discharge |
| CN115223228A (en) * | 2022-08-02 | 2022-10-21 | 合肥工业大学 | Face key point detection method based on mixed domain attention mechanism |
| US11810366B1 (en) * | 2022-09-22 | 2023-11-07 | Zhejiang Lab | Joint modeling method and apparatus for enhancing local features of pedestrians |
| WO2023213332A1 (en) * | 2022-06-27 | 2023-11-09 | 上海格鲁布科技有限公司 | Separation and identification method for multi-source hybrid ultra-high frequency partial discharge diagram |
| CN117274760A (en) * | 2023-10-24 | 2023-12-22 | 南京信息工程大学 | Infrared and visible light image fusion method based on multi-scale mixed converter |
| CN117351273A (en) * | 2023-09-28 | 2024-01-05 | 中国电力科学研究院有限公司 | Causal knowledge-guided power equipment partial discharge fault diagnosis method |
| CN117372769A (en) * | 2023-10-24 | 2024-01-09 | 四川大学 | Cable terminal defect fault identification method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| FR3006769B1 (en) * | 2013-06-11 | 2016-12-02 | Commissariat Energie Atomique | REFLECTOMETRY METHOD FOR IDENTIFYING NON-FREE IMPACTS OF A CABLE |
| DE102013224573B3 (en) * | 2013-11-29 | 2014-10-02 | Hagenuk KMT Kabelmeßtechnik GmbH | Method and device for locating partial discharges in electrical cables |
-
2024
- 2024-03-27 CN CN202410355793.2A patent/CN117949794B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000214211A (en) * | 1999-01-26 | 2000-08-04 | Furukawa Electric Co Ltd:The | Partial discharge judging method |
| CN103926512A (en) * | 2014-03-25 | 2014-07-16 | 国家电网公司 | Method for simulating XLPE insulated cable partial discharge and simulation testing platform |
| CN108680836A (en) * | 2018-04-26 | 2018-10-19 | 天津大学 | Multiple physical field based on high-frequency method couples paper oil insulation partial discharge detecting system |
| CN109100627A (en) * | 2018-10-31 | 2018-12-28 | 红相股份有限公司 | A kind of power equipment partial discharges fault diagnostic method based on end-to-end mode |
| KR20220023545A (en) * | 2020-08-21 | 2022-03-02 | 한국전력공사 | Apparatus and method for diagnosing partial discharge |
| CN112149549A (en) * | 2020-09-18 | 2020-12-29 | 国网山东省电力公司泰安供电公司 | GIS partial discharge type identification method based on depth residual error network |
| CN113889198A (en) * | 2021-09-24 | 2022-01-04 | 国网宁夏电力有限公司电力科学研究院 | Transformer fault diagnosis method and equipment based on oil chromatogram time-frequency domain information and residual error attention network |
| WO2023213332A1 (en) * | 2022-06-27 | 2023-11-09 | 上海格鲁布科技有限公司 | Separation and identification method for multi-source hybrid ultra-high frequency partial discharge diagram |
| CN115223228A (en) * | 2022-08-02 | 2022-10-21 | 合肥工业大学 | Face key point detection method based on mixed domain attention mechanism |
| US11810366B1 (en) * | 2022-09-22 | 2023-11-07 | Zhejiang Lab | Joint modeling method and apparatus for enhancing local features of pedestrians |
| CN117351273A (en) * | 2023-09-28 | 2024-01-05 | 中国电力科学研究院有限公司 | Causal knowledge-guided power equipment partial discharge fault diagnosis method |
| CN117274760A (en) * | 2023-10-24 | 2023-12-22 | 南京信息工程大学 | Infrared and visible light image fusion method based on multi-scale mixed converter |
| CN117372769A (en) * | 2023-10-24 | 2024-01-09 | 四川大学 | Cable terminal defect fault identification method |
Non-Patent Citations (3)
| Title |
|---|
| A GIS Partial Discharge Pattern Recognition Method Based on Improved CBAM-ResNet;Di Hu 等;Journal of Electrical and Computer Engineering;20231212;全文 * |
| 基于深度残差网络的GIS局部放电模式识别技术研究;贾勇勇 等;高压电器;20181116;全文 * |
| 电力物联网下基于卷积神经网络和迁移学习的GIS局部放电模式识别分类方法研究;杨为 等;高压电器;20200916;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117949794A (en) | 2024-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112541389B (en) | Transmission line fault detection method based on EfficientDet network | |
| CN102519846B (en) | Hyperspectrum-based composite insulator hydrophobicity detection method | |
| CN111008641B (en) | Power transmission line tower external force damage detection method based on convolutional neural network | |
| CN114417926A (en) | Power equipment partial discharge pattern recognition method and system based on deep convolution generation countermeasure network | |
| CN107356843A (en) | The partial discharge of transformer method for diagnosing faults of small echo is synchronously extruded based on gradient threshold | |
| CN115187154A (en) | Neural network-based regional power grid oscillation source risk prediction method and system | |
| CN116595459A (en) | Pollution flashover early warning method and system based on electric field signals | |
| CN115586406A (en) | GIS partial discharge fault diagnosis method and system based on ultrahigh frequency signal | |
| CN106570582A (en) | Method and system for building transmission line dancing tripping risk prediction network model | |
| CN115728612A (en) | Transformer discharge fault diagnosis method and device | |
| CN113689053B (en) | Strong convection weather overhead line power failure prediction method based on random forest | |
| CN116089882A (en) | Cable fault prediction processing method and device and electronic equipment | |
| CN117949794B (en) | Cable partial discharge fault detection method | |
| CN114839492A (en) | Method and device for identifying GIS partial discharge type based on MOBILE NETV3 | |
| CN116596920B (en) | Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle | |
| CN111537207B (en) | Data enhancement method for intelligent diagnosis of mechanical fault under small sample | |
| CN117272166A (en) | Distributed optical fiber perimeter security intrusion signal identification method based on cross-model knowledge distillation | |
| CN116244600A (en) | Method, system and equipment for constructing GIS intermittent discharge mode identification model | |
| Yang et al. | GIS partial discharge patterns recognition with spherical convolutional neural network | |
| CN113156278A (en) | CNN network-based GIS partial discharge positioning method | |
| CN113341223A (en) | Method suitable for power grid harmonic analysis and harmonic source positioning | |
| Zhang et al. | Transformer multimodal data and knowledge encoding method based on attention model | |
| Houming et al. | A Detection Method of Self Explosion Defect of Transmission Line Insulator Based on Cascade R-CNN | |
| CN116993327B (en) | Defect positioning system and method for transformer substation | |
| Chen et al. | Research on Cable Terminal Partial Discharge Identification Method Based on Genetic Algorithm Optimized CNN Model |
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 |