CN117242487A - Separation and identification method for multisource mixed ultrahigh frequency partial discharge spectrum - Google Patents
Separation and identification method for multisource mixed ultrahigh frequency partial discharge spectrum Download PDFInfo
- Publication number
- CN117242487A CN117242487A CN202380010710.XA CN202380010710A CN117242487A CN 117242487 A CN117242487 A CN 117242487A CN 202380010710 A CN202380010710 A CN 202380010710A CN 117242487 A CN117242487 A CN 117242487A
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- prpd
- discharge
- map
- detection
- 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.)
- Pending
Links
- 238000001228 spectrum Methods 0.000 title claims abstract description 26
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000000926 separation method Methods 0.000 title claims abstract description 14
- 238000001514 detection method Methods 0.000 claims abstract description 67
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000013135 deep learning Methods 0.000 claims abstract description 14
- 230000000694 effects Effects 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract 2
- 238000002372 labelling Methods 0.000 claims description 12
- 238000009413 insulation Methods 0.000 claims description 10
- 239000002245 particle Substances 0.000 claims description 10
- 238000012360 testing method Methods 0.000 claims description 10
- 239000007787 solid Substances 0.000 claims description 9
- 239000000725 suspension Substances 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000012790 confirmation Methods 0.000 claims description 3
- 230000010363 phase shift Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 239000004020 conductor Substances 0.000 description 1
- 238000011900 installation process Methods 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1227—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials
- G01R31/1254—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials of gas-insulated power appliances or vacuum gaps
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses a separation and identification method of a multisource mixed type ultrahigh frequency partial discharge spectrum, which comprises the following steps: obtaining a PRPD (partial discharge) map of an ultrahigh frequency partial discharge signal, carrying out standardized processing, constructing and marking a partial discharge sample database, and constructing a PRPD map data set of the partial discharge signal and a noise signal; determining a deep learning network as a detection model, and inputting detection model parameters; dividing a PRPD map data set, inputting the PRPD map data set into a deep learning network detection model for training, adjusting the detection model and training parameters, reserving and evaluating the trained model, and selecting the model with the optimal effect as the PRPD map detection model; and (3) identifying the atlas to be identified, using a selected detection model, analyzing and confirming detection targets one by one, and reserving the final detection and multi-source separation results. The invention solves the problem that when the GIS ultrahigh frequency PRPD map is subjected to discharge type identification by using an image identification algorithm, a plurality of partial discharge signals cannot be detected respectively when the partial discharge signals actually exist.
Description
The invention belongs to the field of partial discharge signal identification, and particularly relates to a separation and identification method of a multi-source mixed ultrahigh frequency partial discharge spectrum based on a digital image target detection and pairing technology.
Since the practical use of gas insulated switchgear (Gas Insulated Switchgear, GIS) devices in the 60 th century, the gas insulated switchgear has been widely used not only in high-voltage and ultra-high-voltage fields but also in ultra-high-voltage fields. However, in a complex working environment or during the manufacturing and installation process, internal defects and safety hazards, such as poor contact of conductors, conductive particles, metal tips, air gaps of insulators, etc., are inevitably generated, which cause various types of partial discharge, and further cause insulation faults of GIS and accidents of electric power systems. Therefore, the ultra-high frequency partial discharge detection is carried out on the running GIS equipment at regular intervals, and partial discharge inside the GIS is analyzed and diagnosed according to the detected partial discharge PRPD (partial discharge pulse-duration) spectrum, so that hidden dangers of defects inside the GIS are discovered and eliminated in time, and the ultra-high frequency partial discharge detection becomes an important component for the running and maintenance of the GIS equipment.
The type identification of partial discharge is an important link of GIS partial discharge detection, and is mainly based on a partial discharge PRPD map. At present, for GIS partial discharge type identification, besides manual identification according to experience by overhaulers, a plurality of research institutions research and develop an artificial intelligent identification method based on an image identification technology so as to realize automation of partial discharge identification. Such methods have the following problems: the partial discharge spectrum data can only be identified according to the integral characteristics of the spectrum image, and only one partial discharge type is output as a final identification result. In actual situations, the conditions of coexistence of a plurality of partial discharge signals in the GIS, coexistence of an internal partial discharge signal and an external noise signal and the like often occur, and the PRPD pattern is presented as a mixed superposition image of a plurality of partial discharge signal patterns, so that the pattern is 'four unlike', and the identification error is caused. Even if the type of a certain signal source is correctly identified, other coexisting signal sources must be omitted.
Disclosure of Invention
The invention aims to provide a separation and identification method of a multi-source mixed type ultrahigh frequency partial discharge spectrum, which is used for solving the problem that when the type identification is carried out on a GIS ultrahigh frequency PRPD spectrum by the existing method, the spectrum coexisting with a plurality of signal sources cannot be detected and identified one by one.
A separation and identification method of a multisource mixed ultrahigh frequency partial discharge spectrum comprises the following steps:
step one: a special high frequency partial discharge detection device is used for obtaining a PRPD map, and the PRPD map is processed and converted into a standard gray level image, so that a partial discharge map sample database is constructed;
step two: labeling a sample database, carrying out frame selection on the map areas corresponding to each signal source in each sample image one by one, and labeling the corresponding signal types of the map areas to form a PRPD map data set;
step three: determining a deep learning network as a detection model, and inputting detection model parameters;
step four: dividing a PRPD map data set, inputting the PRPD map data set into a deep learning network detection model for training, adjusting the detection model and training parameters, and reserving the trained model;
step five: evaluating the trained model, and selecting the model with the optimal effect as a PRPD (partial discharge detector) spectrum detection model of the partial discharge signal;
step six: detecting discharge signals of the map to be detected, and detecting targets of various partial discharges and noise interference in the corresponding gray level map;
step seven: and analyzing and confirming detection targets one by one, and reserving the detection targets meeting the standard as the final detection and multi-source separation results.
Preferably, in the first step, a PRPD map is obtained by using an ultrahigh frequency partial discharge detection device, and the PRPD map includes: independent partial discharge signals and interference noise signals, and several signals of mixed partial discharge and interference noise;
the PRPD pattern is converted into a gray pattern, the gray value of each pixel is an integer in which the pulse number at the position is normalized to the range of [0,255], namely, the maximum value of the pulse number is normalized to 255, and the other pulse numbers are normalized to the range of [0,254] in the same proportion.
Preferably, in the second step, labeling of the partial discharge sample data is performed by an image labeling tool "labelme", wherein: for each map in the partial discharge sample database, marking the areas of each partial discharge signal and each noise signal in the PRPD map and the respective signal types by adopting rectangular boxes; the suspension discharge, solid insulation discharge and particle discharge signals are gathered in two clusters with 180-degree phase deviation in a PRPD map, and are respectively marked by adopting 2 rectangular frames;
the PRPD profile dataset comprises: suspension discharge, solid insulation discharge, particle discharge, tip discharge and interference noise.
Preferably, in the third step, the deep learning detection model is based on YOLOv3 detection algorithm.
Preferably, in the third step, dark-53 is used as a backbone network of the YOLOv3 detection algorithm.
Preferably, in the third step, the deep learning network detection model includes feature maps of 13×13, 26×26 and 52×52 pixels, and the feature maps use a size of K-means algorithm to cluster an average size of 9 classes of prior frames.
Preferably, in the fourth step, the PRPD map data set includes: training set, verification set and test set, wherein: the ratio of the training set to the verification set to the test set is 6:2:2.
preferably, in the fourth step, parameters of batch (the number of pictures trained by inputting the detection model each time) and an initial learning rate are set, and the learning rate is dynamically adjusted according to the change of the loss function;
the conditions for storing the trained model are as follows: the value of the loss function fluctuates around a small range and does not drop any more.
Preferably, in the fifth step, the model with the optimal effect is: the highest mAP trained model on the test set.
Preferably, in the seventh step, the specific analysis and confirmation of the detection targets one by one is: for the objects to be tested of suspension discharge, solid insulation discharge and particle discharge, in the area which is 180 degrees away from the phase shift of the objects to be tested and has the amplitude within 10dB, if the similar objects exist, the objects are paired and then used as the same detected partial discharge signal; if the similar object does not exist, discarding the object.
The invention has the following technical effects:
1. the problem that when the discharge type identification is carried out on the GIS ultrahigh frequency PRPD map by using an image identification algorithm, a plurality of partial discharge signals cannot be detected respectively when the partial discharge signals actually exist is solved;
2. not only can accurately carry out multi-source separation on the multi-partial discharge mixed signal spectrum, but also can respectively carry out type identification on separated signals;
3. the method changes the traditional GIS discharge signal detection mode, realizes automatic detection and multi-source separation, adopts a K-means clustering algorithm to improve the parameters of the initial candidate frames, effectively improves the recognition speed, and can accurately and effectively monitor the GIS discharge signals in real time by more than 95% of mAP on the test set.
FIG. 1 is a schematic flow chart of the present invention;
FIG. 2 is a schematic diagram of PRPD pattern normalization and dataset annotation according to the present invention;
in the figure: (a) - (e) is the gray graphic intent of the PRPD map normalized, and (f) - (j) are the dataset annotation schematic;
FIG. 3 is a schematic diagram of a network structure of the YOLOv3 target detection algorithm of the present invention;
FIG. 4 is a schematic diagram of the detection flow of the map to be detected according to the present invention.
The technical scheme of the invention is further described below with reference to the embodiment and the attached drawings.
The PRPD pattern, namely a phase resolution partial discharge pattern (Phase Resolved Partial Discharge), is a statistical pattern formed by collecting a plurality of partial discharge ultrahigh frequency pulses within a period of time by a partial discharge detection instrument and simultaneously collecting and recording the power frequency phase (0-360 degrees) of the power grid when each pulse is collected. The map is a scattered two-dimensional coordinate plane map, the abscissa is the power frequency phase (0-360 degrees) of the power grid of the pulse, the ordinate is the power of the pulse (the range is generally-80-0 dBm), and the value of each scattered point is the number of the pulses of the coordinate in the acquisition time period.
The embodiment provides a separation and identification method of a multi-source mixed ultrahigh frequency partial discharge spectrum, which comprises the following steps:
step one: a special high frequency partial discharge detection device is used for obtaining a PRPD map, and the PRPD map is processed and converted into a standard gray level image, so that a partial discharge map sample database is constructed;
step two: labeling a sample database, carrying out frame selection on the map areas corresponding to each signal source in each sample image one by one, and labeling the corresponding signal types of the map areas to form a PRPD map data set;
step three: determining a deep learning network as a detection model, and inputting detection model parameters;
step four: dividing a PRPD map data set, inputting the PRPD map data set into a deep learning network detection model for training, adjusting the detection model and training parameters, and reserving the trained model;
step five: evaluating the trained model, and selecting the model with the optimal effect as a PRPD (partial discharge detector) spectrum detection model of the partial discharge signal;
step six: detecting discharge signals of the map to be detected, and detecting targets of various partial discharges and noise interference in the corresponding gray level map;
step seven: and analyzing and confirming detection targets one by one, and reserving the detection targets meeting the standard as the final detection and multi-source separation results.
In a further implementation manner of this embodiment, in step one, a PRPD spectrum is obtained using an ultrahigh frequency partial discharge detection apparatus, where the PRPD spectrum includes: four kinds of independent partial discharge signal patterns, various external interference signal patterns and various partial discharge and external interference mixed signal patterns;
converting the PRPD pattern into gray patterns, and obtaining gray patterns corresponding to the five tens of thousands of PRPD patterns in total, wherein ten thousands of gray patterns of four independent partial discharge signals are respectively obtained, and ten thousands of gray patterns of mixed partial discharge signals are obtained; the horizontal and vertical axes are quantized to 416 parts, respectively, and the gray value of each pixel is an integer in which the number of pulses at that position is normalized to the range of [0,255], i.e., the maximum number of pulses is normalized to 255, and the other number of pulses is normalized to the range of [0,254] in the same ratio.
In a further implementation manner of the embodiment, in the second step, labeling of the partial discharge sample data is performed by using an image labeling tool "labelme" to generate an xml file containing labeling information, where: for each map in the partial discharge sample database, marking the areas of each partial discharge signal and each noise signal in the PRPD map and the respective signal types by adopting rectangular boxes; the signals of suspension discharge, solid insulation discharge and particle discharge are gathered in two clusters with 180-degree phase deviation in a PRPD map, and are respectively marked by adopting 2 rectangular frames;
the PRPD profile dataset includes: suspension discharge, solid insulation discharge, particle discharge, tip discharge and interference noise.
In a further implementation manner of this embodiment, in the third step, the deep learning detection model is based on the YOLOv3 detection algorithm, the size parameter of the input picture is 416×416 pixels, yolo is a target detection algorithm, the task of target detection is to find the object from the picture and give its category and position, it applies a single Convolutional Neural Network (CNN) to the whole image, divides the image into grids, and predicts the class probability and bounding box of each grid.
In a further implementation manner of this embodiment, in the third step, the dark-53 is used as the backbone network of the YOLOv3 detection algorithm, the dark-53 includes a 53-layer convolution, the first 52-layer convolution is set as the main network, and the last layer is the full connection layer.
In a further implementation manner of this embodiment, in the third step, the deep learning network detection model includes feature maps of 13×13, 26×26, and 52×52 pixels, and the feature maps use a size that is an average size of 9 classes of prior frames clustered by a K-means algorithm.
The number of prior boxes is 13×13×3+26×26×3+52×52×3= 10647, each prediction is a 10-dimensional vector (4+1+5=10), and the 10-dimensional vector includes: frame coordinates (4 values), frame confidence (1 value), probability of object class (4 discharge signals+1 noise signal);
the depth of the final output of each feature map is: 3 (4+1+5) =30;
the loss function includes: center coordinate error, wide-high coordinate error, confidence error, and class error.
In a further implementation manner of the present embodiment, in step four, the PRPD map data set includes: training set, verification set and test set, wherein: the training set, the verification set and the test set have the proportion of 6:2:2, the numbers are 30000, 10000 and 10000 respectively.
In a further implementation manner of the embodiment, in the fourth step, the batch is set to 64, which means that the network parameter is updated by forward propagation once every 64 samples are trained, the total iteration number is set to 400000, the initial learning rate is 0.001, the learning rate is dynamically adjusted according to the change of the loss function, the learning rate decays ten times when iterating to 200000 times, and the learning rate decays ten times again on the basis of the previous learning rate when iterating to 300000 times;
the trained model conditions are stored as follows: the value of the loss function fluctuates around a small range and does not drop any more, keeping the number of models at 5.
In a further implementation manner of the embodiment, in the fifth step, the five trained models are tested on the test set, the mAP (average accuracy of five targets) is calculated, and the model with the highest mAP is selected as the final detection model.
In a further implementation manner of the present embodiment, in step seven, the specific analysis and confirmation detection targets are: for targets to be detected of the tip discharge signal and the interference noise signal, detecting only two targets, and reserving detection results; for the objects to be tested of suspension discharge, solid insulation discharge and particle discharge, in the area (corresponding to 208 pixels horizontally shifted and 52 pixels longitudinally shifted on a gray level diagram) which is 180 degrees away from the phase of the object and has the amplitude within 10dB, if the similar objects exist, the same detected partial discharge signals are obtained after pairing; if the similar object does not exist, discarding the object.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
- The separation and identification method of the multisource mixed type ultrahigh frequency partial discharge spectrum is characterized by comprising the following steps of:step one: acquiring a PRPD (partial discharge detector) map of a GIS partial discharge signal, and carrying out standardized processing on the PRPD map to construct a partial discharge sample database;step two: labeling a partial discharge sample database, and constructing a PRPD map data set of discharge signals and noise signals;step three: determining a deep learning network as a detection model, and inputting detection model parameters;step four: dividing a PRPD map data set, inputting the PRPD map data set into a deep learning network detection model for training, adjusting the detection model and training parameters, and reserving the trained model;step five: evaluating the trained model, and selecting the model with the optimal effect as a PRPD (partial discharge detector) spectrum detection model of the partial discharge signal;step six: detecting discharge signals of the map to be detected, and detecting targets of various partial discharges and noise interference in the corresponding gray level map;step seven: analyzing and confirming detection targets one by one, and reserving the detection targets meeting the standard as final detection and multi-source separation results;the PRPD profile dataset comprises: a target to be measured of suspension discharge, solid insulation discharge, particle discharge, tip discharge, interference noise; in the second step, labeling of the partial discharge sample database is performed through an image labeling tool, wherein: for each map in the partial discharge sample database, marking the positions of partial discharge signals and interference noise signals in the PRPD map and the signal types by adopting rectangular frames; the suspension discharge, solid insulation discharge and particle discharge signals are gathered in two clusters with 180-degree phase deviation in a PRPD map, and are respectively marked by adopting 2 rectangular frames;in the seventh step, the specific analysis and confirmation of the detection targets one by one is as follows: for the objects to be tested of suspension discharge, solid insulation discharge and particle discharge, in the area which is 180 degrees away from the phase shift of the objects to be tested and has the amplitude within 10dB, if the similar objects exist, the objects are paired and then used as the same detected partial discharge signal; if the similar object does not exist, discarding the object.
- The method for separating and identifying the multi-source mixed ultrahigh frequency partial discharge spectrum according to claim 1, wherein in the first step, a PRPD spectrum is obtained by using an ultrahigh frequency partial discharge detection device, and the PRPD spectrum includes: independent partial discharge signals and interference noise signals, and several signals of mixed partial discharge and interference noise;the PRPD pattern is converted into a gray pattern, and the range of gray values normalized by the number of pulses at that position for each pixel is [0,255].
- The method for separating and identifying multi-source mixed ultrahigh frequency partial discharge patterns according to claim 1, wherein in the third step, the deep learning detection model is based on a YOLOv3 detection algorithm.
- The method for separating and identifying multi-source mixed ultrahigh frequency partial discharge patterns according to claim 3, wherein in the third step, dark-53 is adopted as a backbone network of a YOLOv3 detection algorithm.
- The method for separating and identifying the multi-source mixed ultrahigh frequency partial discharge spectrum according to claim 4, wherein in the third step, the deep learning network detection model comprises a feature map of 3 types of pixels, and the size used by the feature map is the average size of 9 types of prior frames clustered by a K-means algorithm.
- The method for separating and identifying a multi-source hybrid ultrahigh frequency partial discharge spectrum according to claim 1, wherein in the fourth step, the PRPD spectrum dataset includes: training set, verification set and test set, wherein: the ratio of the training set to the verification set to the test set is 6:2:2.
- the method for separating and identifying the multi-source mixed ultrahigh frequency partial discharge patterns according to claim 1, wherein in the fourth step, the number of pictures and the initial learning rate for training each time of inputting the detection model are set, and the learning rate is dynamically adjusted according to the change of the loss function;the trained model conditions are as follows: the value of the loss function fluctuates around a small range and does not drop any more.
- The method for separating and identifying the multi-source mixed ultrahigh frequency partial discharge spectrum according to claim 1, wherein in the fifth step, the model with the optimal effect is as follows: the highest mAP trained model on the test set.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210721508.5A CN115187527B (en) | 2022-06-27 | 2022-06-27 | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum |
| PCT/CN2023/101087 WO2023213332A1 (en) | 2022-06-27 | 2023-06-19 | Separation and identification method for multi-source hybrid ultra-high frequency partial discharge diagram |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117242487A true CN117242487A (en) | 2023-12-15 |
Family
ID=83515178
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210721508.5A Active CN115187527B (en) | 2022-06-27 | 2022-06-27 | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum |
| CN202380010710.XA Pending CN117242487A (en) | 2022-06-27 | 2023-06-19 | Separation and identification method for multisource mixed ultrahigh frequency partial discharge spectrum |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210721508.5A Active CN115187527B (en) | 2022-06-27 | 2022-06-27 | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum |
Country Status (2)
| Country | Link |
|---|---|
| CN (2) | CN115187527B (en) |
| WO (1) | WO2023213332A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117630611A (en) * | 2024-01-22 | 2024-03-01 | 南京卓煊电力科技有限公司 | Full-bandwidth high-frequency partial discharge PRPD spectrogram capturing and generating method and system |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115187527B (en) * | 2022-06-27 | 2023-04-07 | 上海格鲁布科技有限公司 | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum |
| CN116091501B (en) * | 2023-04-07 | 2023-06-23 | 武汉纺织大学 | Method, device, equipment and medium for identifying partial discharge type of high-voltage electrical equipment |
| CN117607642B (en) * | 2024-01-23 | 2024-03-26 | 中铁电气化铁路运营管理有限公司 | Rail transit data analysis method and system |
| CN117786521B (en) * | 2024-02-23 | 2024-04-19 | 兰州交通大学 | Signal classification method based on intelligent frequency spectrum sensing |
Family Cites Families (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103558529B (en) * | 2013-11-14 | 2016-11-23 | 国家电网公司 | A kind of mode identification method of three-phase cartridge type supertension GIS partial discharge altogether |
| CN104198899B (en) * | 2014-08-04 | 2017-01-25 | 西安交通大学 | Partial discharge type identifying method for transformer under multi-discharge source situation |
| CN105807190B (en) * | 2014-12-29 | 2019-06-18 | 国家电网公司 | A kind of GIS partial discharge SHF band electric detection method |
| CN106556781A (en) * | 2016-11-10 | 2017-04-05 | 华乘电气科技（上海）股份有限公司 | Shelf depreciation defect image diagnostic method and system based on deep learning |
| ES2780448A1 (en) * | 2019-02-22 | 2020-08-25 | Ormazabal Corp Technology Aie | Partial discharge recognition method and system for electrical network diagnosis (Machine-translation by Google Translate, not legally binding) |
| KR102389439B1 (en) * | 2019-04-01 | 2022-04-22 | 명지대학교 산학협력단 | RNN based partial discharge detection apparatus and method in gas-insulated switchgear |
| CN110794264B (en) * | 2019-10-14 | 2021-11-19 | 浙江浙能技术研究院有限公司 | Generator partial discharge type identification method based on time domain pulse characteristics |
| CN110850244B (en) * | 2019-11-11 | 2022-03-11 | 国网湖南省电力有限公司 | Local discharge defect time domain map diagnosis method, system and medium based on deep learning |
| CN111142001B (en) * | 2020-01-10 | 2022-04-22 | 三峡大学 | Transformer multi-source partial discharge mode identification method based on parallel characteristic domain |
| CN112198399B (en) * | 2020-09-07 | 2023-12-19 | 红相股份有限公司 | Multi-source electromagnetic wave signal identification method and terminal |
| 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 |
| CN114186589A (en) * | 2021-12-08 | 2022-03-15 | 国网上海市电力公司 | Superconducting cable partial discharge mode identification method based on residual error network Resnet50 |
| CN114386499A (en) * | 2021-12-30 | 2022-04-22 | 重庆邮电大学 | Multi-source partial discharge signal data stream clustering separation method based on GIS |
| CN114417926A (en) * | 2022-01-19 | 2022-04-29 | 山东大学 | Power equipment partial discharge pattern recognition method and system based on deep convolution generation countermeasure network |
| CN115187527B (en) * | 2022-06-27 | 2023-04-07 | 上海格鲁布科技有限公司 | Separation and identification method for multi-source mixed ultrahigh frequency partial discharge spectrum |
-
2022
- 2022-06-27 CN CN202210721508.5A patent/CN115187527B/en active Active
-
2023
- 2023-06-19 CN CN202380010710.XA patent/CN117242487A/en active Pending
- 2023-06-19 WO PCT/CN2023/101087 patent/WO2023213332A1/en unknown
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117630611A (en) * | 2024-01-22 | 2024-03-01 | 南京卓煊电力科技有限公司 | Full-bandwidth high-frequency partial discharge PRPD spectrogram capturing and generating method and system |
| CN117630611B (en) * | 2024-01-22 | 2024-04-12 | 南京卓煊电力科技有限公司 | Full-bandwidth high-frequency partial discharge PRPD spectrogram capturing and generating method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115187527B (en) | 2023-04-07 |
| WO2023213332A1 (en) | 2023-11-09 |
| CN115187527A (en) | 2022-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117242487A (en) | Separation and identification method for multisource mixed ultrahigh frequency partial discharge spectrum | |
| CN108573225B (en) | Partial discharge signal pattern recognition method and system | |
| CN111797890A (en) | Method and system for detecting defects of power transmission line equipment | |
| CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
| CN109685117B (en) | Image measuring method for flow parameters in gas-liquid-solid three-phase system | |
| CN116167668B (en) | BIM-based green energy-saving building construction quality evaluation method and system | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| CN111598942A (en) | Method and system for automatically positioning electric power facility instrument | |
| Chang et al. | Application of pulse sequence partial discharge based convolutional neural network in pattern recognition for underground cable joints | |
| CN112528774B (en) | Intelligent unknown radar signal sorting system and method in complex electromagnetic environment | |
| CN110533007A (en) | A kind of vehicle-mounted strain of bridge influences the identification of line feature intelligent and extracting method | |
| CN109754392A (en) | A kind of device and method that defect image automatically analyzes | |
| CN112668612A (en) | Partial discharge signal clustering analysis method based on grids | |
| CN116416884B (en) | Testing device and testing method for display module | |
| CN112348052A (en) | Power transmission and transformation equipment abnormal sound source positioning method based on improved EfficientNet | |
| Su et al. | New particle formation event detection with Mask R-CNN | |
| CN110672988A (en) | Partial discharge mode identification method based on hierarchical diagnosis | |
| CN114839492A (en) | Method and device for identifying GIS partial discharge type based on MOBILE NETV3 | |
| CN111242028A (en) | Remote sensing image ground object segmentation method based on U-Net | |
| CN114700587A (en) | Missing welding defect real-time detection method and system based on fuzzy reasoning and edge calculation | |
| CN116466408B (en) | Artificial neural network superbedrock identification method based on aeromagnetic data | |
| CN115497006B (en) | Urban remote sensing image change depth monitoring method and system based on dynamic mixing strategy | |
| CN111025100A (en) | Transformer ultrahigh frequency partial discharge signal mode identification method and device | |
| CN115586406A (en) | GIS partial discharge fault diagnosis method and system based on ultrahigh frequency signal | |
| CN109765878A (en) | The aided analysis method of new-energy automobile CAN bus network failure |
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 |