CN112329870A - Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm - Google Patents
Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm Download PDFInfo
- Publication number
- CN112329870A CN112329870A CN202011252600.9A CN202011252600A CN112329870A CN 112329870 A CN112329870 A CN 112329870A CN 202011252600 A CN202011252600 A CN 202011252600A CN 112329870 A CN112329870 A CN 112329870A
- Authority
- CN
- China
- Prior art keywords
- training
- yolo3
- model
- image
- images
- 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
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method for identifying the state of a transformer substation pressure plate based on a YOLO3 algorithm, which comprises the following steps: collecting image information, labeling a training set, model training, model prediction and identifying the state of a pressing plate of an image containing the pressing plate by using a training qualified YOLO3 model. According to the transformer substation pressure plate state identification method based on the YOLO3 algorithm, the YOLO3 algorithm in the current deep learning is utilized, and the transformer substation pressure plate state in the collected video and photo can be accurately identified.
Description
Technical Field
The invention relates to the technical field of power grids, in particular to a transformer substation pressure plate state identification method based on a YOLO3 algorithm.
Background
With the rapid development of computer technology and communication technology, various devices in a transformer substation tend to be intelligentized, and as the automation level of the transformer substation is gradually improved and gradually develops to an unattended mode, higher requirements are provided for the reliability and stability of the intelligentization of primary devices and secondary devices. Along with the rapid development of various levels of power grids of national grid companies, the number of transformer substations is gradually increased, and higher requirements are put forward for fixed operation and maintenance personnel.
Operation and maintenance personnel need to execute a large amount of switching operations for each transformer substation during autumn inspection and spring inspection, and multiple operations are needed for switching on and off of pressing plates on a protection screen of the transformer substation. Once a certain pressure plate is dropped or dropped by mistake, huge hidden dangers are buried in the subsequent operation of the corresponding protection device, so that the protection device is rejected or mistakenly operated at a critical moment, and the personnel, the power grid and the equipment are greatly influenced. The current transformer substation pressure plate is not only manually checked, but also checked by a transformer substation inspection robot. The transformer substation inspection robot mainly carries out pressing plate inspection through mode identification, the mode identification needs fixed-point shooting, the identification method is completely compared through a fixed mode manually input in advance, the requirement on identification positions is high, and the universality is not realized on the pressing plate identification of different types of various manufacturers. Therefore, how to further reduce the possible misjudgment of the traditional manual detection and identification and the harsh conditions of the pattern identification becomes a problem to be solved.
In recent years, deep learning and machine learning become increasingly hot, and computer vision recognition algorithms (such as YOLO 3) are more and more diversified. The YOLO3 algorithm directly predicts the object boundary and the class probability by adopting a neural network to realize end-to-end object detection, so that the identification of the transformer substation pressure plate state by utilizing the YOLO3 algorithm in deep learning becomes possible.
Disclosure of Invention
In order to solve the problems in the prior art, the method for identifying the state of the transformer substation pressure plate based on the YOLO3 algorithm is provided, and the method can accurately identify the state of the transformer substation pressure plate in the collected video and photo by using the YOLO3 algorithm in the current deep learning.
In order to achieve the above purpose, the present application provides a transformer substation pressure plate state identification method based on the YOLO3 algorithm, including the following steps:
step 1, collecting image information: collecting videos and photos related to a pressing plate in a transformer substation, wherein each video corresponds to a plurality of frames of images, and adjusting the resolution of each image corresponding to the collected videos and photos to enable the resolution of each image to be uniform;
step 2, marking a training set: dividing each image with unified resolution in the step 1 into two parts according to a mutual exclusion principle, manually marking the pressing plate state of one part of the image, manually marking the pressing plate state as 'input', 'stop using' or 'release' according to the actual state of the pressing plate in the image, and using the part of the image as a training set; the other part of the image is not marked and is used as a test set;
step 3, model training: dividing the images in the training set into n batches randomly, wherein n is more than or equal to 2 and less than or equal to 5, and n is an integer, inputting the first batch of images into a YOLO3 model to train a YOLO3 model, repeating the training for m times, wherein m is more than or equal to 5 and less than or equal to 8, and m is an integer, so that the first batch of images are used for m-round training of the YOLO3 model, then inputting the second batch of images into a YOLO3 model to train the YOLO3 model, repeating the training for m times, so as to perform m-round training of the YOLO3 model through the second batch of images, and repeating the training for m times until the Nth batch of images are used for m-round training of the YOLO3 model;
step 4, model prediction: after the training of the YOLO3 model is completed, all images in the test set are used as input, and the training effect of the YOLO3 model is detected; if the state recognition accuracy of the pressing plate in each test set image is greater than 90%, the training of the YOLO3 model is qualified; otherwise, repeating the step 3, randomly dividing the images in the training set into n batches again, and continuing training the YOLO3 model until the training of the YOLO3 model is qualified;
and 5, identifying the state of the pressure plate of the image containing the pressure plate by using the YOLO3 model qualified in training.
In some embodiments, in step 1, the resolution of each image is uniformly adjusted to 1024 × 768.
In some embodiments, in step 2, the images with uniform resolution in step 1 are divided according to a mutual exclusion principle at a preset ratio, where the preset ratio is divided according to a ratio of 70% and 30%.
The transformer substation pressure plate state identification method based on the YOLO3 algorithm has the advantages that the YOLO3 algorithm in current deep learning is utilized, and the transformer substation pressure plate state in the collected videos and pictures can be accurately identified.
Detailed Description
The following further describes embodiments of the present application.
The method for identifying the state of the transformer substation pressure plate based on the YOLO3 algorithm comprises the following steps:
step 1, collecting image information: collecting videos and photos related to a pressure plate in a transformer substation, wherein each video corresponds to a plurality of frames of images, and adjusting the resolution of each image corresponding to the collected videos and photos to make the resolution of each image uniform, for example, uniformly adjusting the resolution to 1024 × 768, so as to facilitate the processing of subsequent steps.
Step 2, marking a training set: dividing each image with unified resolution in the step 1 into two parts according to a mutual exclusion principle, for example, dividing the image according to a ratio of 70% to 30%; marking a part of images of the images as the pressing plate state manually, marking the part of images as 'input', 'stop' or 'release' manually according to the actual state of the pressing plate in the images, and using the part of images as a training set; and (4) using the other part of image as a test set without any mark.
Step 3, model training: the images in the training set are divided into n batches randomly, wherein n is more than or equal to 2 and less than or equal to 5, and n is an integer, the first batch of images are input into a YOLO3 model to train a YOLO3 model, the training is repeated for m times, m is more than or equal to 5 and less than or equal to 8, and m is an integer, so that the first batch of images are used for performing m-round training on the YOLO3 model, then the second batch of images are input into a YOLO3 model to train the YOLO3 model, the training is repeated for m times, so that the second batch of images are used for performing m-round training on the YOLO3 model, and the rest is done until the n-th batch of images are used for performing m-round training on the YOLO3 model.
Step 4, model prediction: after the training of the YOLO3 model is completed, all images in the test set are used as input, and the training effect of the YOLO3 model is detected; if the state recognition accuracy of the pressing plate in each test set image is greater than 90%, the training of the YOLO3 model is qualified; otherwise, repeating the step 3, dividing the images in the training set into n batches randomly again, and continuing training the YoLO3 model until the training of the YoLO3 model is qualified.
And 5, identifying the state of the pressure plate of the image containing the pressure plate by using the YOLO3 model qualified in training.
According to the transformer substation pressure plate state identification method based on the YOLO3 algorithm, the pressure plate state of the transformer substation in the collected video and photo can be accurately identified by using the YOLO3 algorithm in the current deep learning. The method has the following advantages: (1) the method innovatively uses a deep learning technology, and realizes target detection through a YOLO3 algorithm; (2) the identification method can be used for detecting the following video in real time, and the detection rate per second can reach 60 frames; (3) the running program has no harsh requirements on hardware, and a common PC can run easily.
Claims (3)
1. A transformer substation pressure plate state identification method based on a YOLO3 algorithm is characterized in that: the method comprises the following steps:
step 1, collecting image information: collecting videos and photos related to a pressing plate in a transformer substation, wherein each video corresponds to a plurality of frames of images, and adjusting the resolution of each image corresponding to the collected videos and photos to enable the resolution of each image to be uniform;
step 2, marking a training set: dividing each image with unified resolution in the step 1 into two parts according to a mutual exclusion principle, manually marking the pressing plate state of one part of the image, manually marking the pressing plate state as 'input', 'stop using' or 'release' according to the actual state of the pressing plate in the image, and using the part of the image as a training set; the other part of the image is not marked and is used as a test set;
step 3, model training: dividing the images in the training set into n batches randomly, wherein n is more than or equal to 2 and less than or equal to 5, and n is an integer, inputting the first batch of images into a YOLO3 model to train a YOLO3 model, repeating the training for m times, wherein m is more than or equal to 5 and less than or equal to 8, and m is an integer, so that the first batch of images are used for m-round training of the YOLO3 model, then inputting the second batch of images into a YOLO3 model to train the YOLO3 model, repeating the training for m times, so as to perform m-round training of the YOLO3 model through the second batch of images, and repeating the training for m times until the Nth batch of images are used for m-round training of the YOLO3 model;
step 4, model prediction: after the training of the YOLO3 model is completed, all images in the test set are used as input, and the training effect of the YOLO3 model is detected; if the state recognition accuracy of the pressing plate in each test set image is greater than 90%, the training of the YOLO3 model is qualified; otherwise, repeating the step 3, randomly dividing the images in the training set into n batches again, and continuing training the YOLO3 model until the training of the YOLO3 model is qualified;
and 5, identifying the state of the pressure plate of the image containing the pressure plate by using the YOLO3 model qualified in training.
2. The method for identifying the state of a substation pressure plate based on the YOLO3 algorithm according to claim 1, wherein the method comprises the following steps: in step 1, the resolution of each image is uniformly adjusted to 1024 × 768.
3. The method for identifying the state of a substation pressure plate based on the YOLO3 algorithm according to claim 1, wherein the method comprises the following steps: in step 2, dividing each image with unified resolution in step 1 according to a mutual exclusion principle at a preset ratio, wherein the preset ratio is divided according to a ratio of 70% and 30%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011252600.9A CN112329870A (en) | 2020-11-11 | 2020-11-11 | Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011252600.9A CN112329870A (en) | 2020-11-11 | 2020-11-11 | Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112329870A true CN112329870A (en) | 2021-02-05 |
Family
ID=74318895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011252600.9A Pending CN112329870A (en) | 2020-11-11 | 2020-11-11 | Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112329870A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108573256A (en) * | 2017-03-14 | 2018-09-25 | 山东鲁能智能技术有限公司 | A kind of substation's plate pressing equipment state identification method and device |
| CN109614969A (en) * | 2018-11-22 | 2019-04-12 | 华南理工大学 | Extensive distribution line the condition of a disaster based on deep learning repairs image-recognizing method |
| CN109711437A (en) * | 2018-12-06 | 2019-05-03 | 武汉三江中电科技有限责任公司 | A kind of transformer part recognition methods based on YOLO network model |
| CN109740696A (en) * | 2019-01-30 | 2019-05-10 | 亮风台（上海）信息科技有限公司 | A kind of method and apparatus for pressing plate identification |
| CN110059556A (en) * | 2019-03-14 | 2019-07-26 | 天津大学 | A kind of transformer substation switch division condition detection method based on deep learning |
| CN110277833A (en) * | 2019-06-20 | 2019-09-24 | 国家电网有限公司 | The method and terminal device of substation's pressing plate state recognition |
| CN110348349A (en) * | 2019-07-01 | 2019-10-18 | 河南牧业经济学院 | A kind of method and system collected, analyze pig behavior video data |
| CN110648307A (en) * | 2019-07-24 | 2020-01-03 | 国网浙江省电力有限公司湖州供电公司 | Method for checking state of transformer substation pressure plate by using image comparison technology |
| CN110889841A (en) * | 2019-11-28 | 2020-03-17 | 江苏电力信息技术有限公司 | YOLOv 3-based bird detection algorithm for power transmission line |
| CN111062282A (en) * | 2019-12-05 | 2020-04-24 | 武汉科技大学 | Transformer substation pointer type instrument identification method based on improved YOLOV3 model |
| CN111814742A (en) * | 2020-07-29 | 2020-10-23 | 南方电网数字电网研究院有限公司 | Knife switch state identification method based on deep learning |
-
2020
- 2020-11-11 CN CN202011252600.9A patent/CN112329870A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108573256A (en) * | 2017-03-14 | 2018-09-25 | 山东鲁能智能技术有限公司 | A kind of substation's plate pressing equipment state identification method and device |
| CN109614969A (en) * | 2018-11-22 | 2019-04-12 | 华南理工大学 | Extensive distribution line the condition of a disaster based on deep learning repairs image-recognizing method |
| CN109711437A (en) * | 2018-12-06 | 2019-05-03 | 武汉三江中电科技有限责任公司 | A kind of transformer part recognition methods based on YOLO network model |
| CN109740696A (en) * | 2019-01-30 | 2019-05-10 | 亮风台（上海）信息科技有限公司 | A kind of method and apparatus for pressing plate identification |
| CN110059556A (en) * | 2019-03-14 | 2019-07-26 | 天津大学 | A kind of transformer substation switch division condition detection method based on deep learning |
| CN110277833A (en) * | 2019-06-20 | 2019-09-24 | 国家电网有限公司 | The method and terminal device of substation's pressing plate state recognition |
| CN110348349A (en) * | 2019-07-01 | 2019-10-18 | 河南牧业经济学院 | A kind of method and system collected, analyze pig behavior video data |
| CN110648307A (en) * | 2019-07-24 | 2020-01-03 | 国网浙江省电力有限公司湖州供电公司 | Method for checking state of transformer substation pressure plate by using image comparison technology |
| CN110889841A (en) * | 2019-11-28 | 2020-03-17 | 江苏电力信息技术有限公司 | YOLOv 3-based bird detection algorithm for power transmission line |
| CN111062282A (en) * | 2019-12-05 | 2020-04-24 | 武汉科技大学 | Transformer substation pointer type instrument identification method based on improved YOLOV3 model |
| CN111814742A (en) * | 2020-07-29 | 2020-10-23 | 南方电网数字电网研究院有限公司 | Knife switch state identification method based on deep learning |
Non-Patent Citations (8)
| Title |
|---|
| CHEN-ZHAO FU等: "Research on a detection and Recognition algorithm for high-voltage switch cabinet based on deep learning with an improved YOLOv2 network", 《2018 11TH INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTATION TECHNOLOGY AND AUTOMATION》 * |
| 冷从林: "基于机器视觉的变电站压板开关状态识别系统研究", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅱ辑》 * |
| 周克等: "一种改进的压板状态识别SSD算法", 《电测与仪表》 * |
| 周自强等: "一种改进的YOLO目标检测方法在电缆设备异常状态识别中的应用", 《电测与仪表》 * |
| 张曼等: "500kV智能变电站压板状态监测及智能校核技术", 《电力自动化》 * |
| 王祎瑞: "基于深度卷积神经网络的变电站设备识别", 《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅱ辑》 * |
| 程华福等: "智慧变电站设备状态识别技术", 《2019电力行业信息化年会论文集》 * |
| 颜宏文等: "基于改进 YOLOv3的绝缘子串定位与状态识别方法", 《高电压技术》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102673106B (en) | Silk screen print positioning equipment and method for photovoltaic solar silicon chip | |
| WO2017084186A1 (en) | System and method for automatic monitoring and intelligent analysis of flexible circuit board manufacturing process | |
| CN111063144A (en) | Abnormal behavior monitoring method, device, equipment and computer readable storage medium | |
| Jiang et al. | Research on infrared image recognition method of power equipment based on deep learning | |
| CN112257500A (en) | Intelligent image recognition system and method for power equipment based on cloud edge cooperation technology | |
| CN112687022A (en) | Intelligent building inspection method and system based on video | |
| CN109919451A (en) | Based on true power network topology analysis and security knowledge base digital electric network system and method | |
| CN111199250A (en) | Transformer substation air switch state checking method and device based on machine learning | |
| Wang et al. | Pointer meter recognition in UAV inspection of overhead transmission lines | |
| CN114581760B (en) | Equipment fault detection method and system for machine room inspection | |
| CN107316131A (en) | A kind of electric power meter mounting process quality detecting system based on image recognition | |
| CN114241413A (en) | Substation multi-target detection method based on attention mechanism and feature balance | |
| CN103093028A (en) | Modeling method and simulation method for monitoring signals in electric power dispatching simulation training system | |
| CN108761263B (en) | Fault diagnosis system based on evidence theory | |
| CN114460519A (en) | On-site and terminal fusion management system and method based on power transformer inspection | |
| CN112329870A (en) | Method for identifying state of transformer substation pressure plate based on YOLO3 algorithm | |
| CN111953657B (en) | Sequence-data joint driven CPS network attack identification method for power distribution network | |
| CN116344378B (en) | Intelligent detection system for photovoltaic panel production and detection method thereof | |
| CN116561638B (en) | Protective pressing plate non-correspondence checking method based on neural network learning and state evaluation | |
| CN111950745A (en) | Switching operation management method and system based on image processing | |
| CN113920535B (en) | Electronic region detection method based on YOLOv5 | |
| CN113780224B (en) | Unmanned inspection method and system for transformer substation | |
| CN115131335A (en) | Power transmission line pin missing detection method | |
| CN111028289B (en) | Method for positioning foreign matters in equipment in transformer substation based on template matching | |
| He et al. | Multi-State Recognition Method of Substation Switchgear Based on Image Enhancement and Deep 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210205 |