CN110033453B - Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 - Google Patents
Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 Download PDFInfo
- Publication number
- CN110033453B CN110033453B CN201910310921.0A CN201910310921A CN110033453B CN 110033453 B CN110033453 B CN 110033453B CN 201910310921 A CN201910310921 A CN 201910310921A CN 110033453 B CN110033453 B CN 110033453B
- Authority
- CN
- China
- Prior art keywords
- insulator
- yolov3
- data set
- fault
- power transmission
- 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
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Insulators (AREA)
- Electric Cable Installation (AREA)
- Testing Of Short-Circuits, Discontinuities, Leakage, Or Incorrect Line Connections (AREA)
Abstract
The invention discloses a power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3, which comprises the following steps: s1: establishing an insulation data set and an insulator fault data set; s2: data enhancement is carried out by using modes of rotation, turnover, contrast enhancement and the like; s3: carrying out model modification and pruning operation by taking YOLOv3 as a basic frame to construct an insulator detection model; s4: building a standard YOLOv3-tiny network; s5: respectively carrying out model training on the modified YOLOv3 network and the modified YOLOv3-tiny by using an insulator data set and an insulator fault data set; s6: and connecting the two trained models to realize the end-to-end rapid detection of the insulator fault. The detection method can quickly and accurately identify the insulator in the complex background image and position and detect the fault area, improves the inspection efficiency, reduces the working intensity of inspection personnel, and protects the normal operation of the power system.
Description
Technical Field
The invention belongs to the field of deep learning computer vision and the field of electric power defect identification, and particularly relates to a fault detection method for an aerial image of an insulator of a power transmission and transformation line based on improved YOLOv 3.
Background
The insulators are numerous in the power transmission and transformation lines and play an important role in electrical insulation and mechanical connection. Because the power transmission and transformation line spans various complex natural geographic environments, various faults such as self-explosion, flashover and the like easily occur after being exposed to wind, rain and sun for a long time. Once the insulator breaks down, the normal and safe transmission of electric power can be seriously influenced, so that the normal operation of an electric power system is threatened, and therefore, the rapid and efficient fault detection of the insulator is one of important programs of the inspection work of a power transmission and transformation line. The current inspection mode commonly used at the stage is that an unmanned aerial vehicle is adopted to shoot inspection images at a specific machine position, and then ground workers manually carry out fault discrimination and analysis on the aerial inspection images, so that the mode not only causes the delay of fault detection, but also brings a large amount of workload for power inspection personnel.
In order to realize automation and real-time inspection of the power transmission and transformation line, an image recognition technology can be adopted to replace manual screening and aerial photography for inspecting the images. However, the traditional target detection algorithm mainly has three defects, and the characteristics designed firstly are low-level characteristics, so that the expressive property of the target is insufficient; secondly, the designed characteristics are single in pertinence, and the robustness of the diversity of the target to be detected of the power transmission and transformation line in a complex environment is not good; and finally, for the current massive polling pictures, the detection speed and the detection effect can not meet the requirements of actual industrial application.
Deep learning develops rapidly since 2012, and a better result is obtained in the field of computer vision, but at present, application research of deep learning for power routing inspection is less, and requirements of instantaneity and accuracy of power transmission and transformation line routing inspection cannot be met, so that the method mainly researches and applies the YOLOv3 model to fault detection of insulator aerial photographing routing inspection images to realize efficient and accurate detection from end to end.
Disclosure of Invention
The invention aims to provide a fault detection method of an aerial image of a power transmission and transformation line insulator based on improved YOLOv3, so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme: the power transmission and transformation line insulator aerial image fault detection method based on the improved YOLOv3 comprises the following steps:
s1: establishing an insulation data set and an insulator fault data set;
s2: data enhancement is carried out by using modes such as rotation, turnover, contrast enhancement and the like; marking the image by using a LabelImg tool to obtain a corresponding XML file, and organizing and storing the data set image and the corresponding XML file according to a PASCAL VOC data set format;
s3: carrying out model modification and pruning operation by taking YOLOv3 as a basic framework to construct an insulator detection model, and comprising the following steps of:
s301: because the fixed machine position shoots the patrol picture, the insulators can be roughly divided into a larger insulator shot in a close shot and a smaller insulator shot in a distant shot, so that three-scale output in YOLOv3 is changed into two-scale output, only the output scales of 13 × 13 and 26 × 26 are reserved, the output scale of 52 × 52 is abandoned, the size of the model is reduced, and meanwhile, the calculated amount in the model detection process is also reduced;
s302: because only one target object of the insulator is detected, the integral network of the YOLOv3 is too large and has excessive functions, the model can be reduced through pruning operation, and the detection speed is increased appropriately;
s303: when model pruning is carried out, in order to ensure that the detection effect of the model is not reduced, 3 anchor boxes are distributed to each original scale and are changed into 4 anchor boxes in each scale, the initial 8 anchor boxes are generated by using a K-means clustering algorithm of which the intersection ratio IOU value replaces the standard Euclidean distance, and the detection effect can be properly improved by using more anchor boxes in each scale;
s304: YOLOv3 considers that the multiple classes are divided by using softmax, and the mutual exclusion among the classes must be satisfied, so that the softmax function is abandoned, and the sigmoid function is adopted for the class division;
s4: building a standard YOLOv3-tiny network;
s5: respectively carrying out model training on the modified YOLOv3 network and the modified YOLOv3-tiny by using an insulator data set and an insulator fault data set;
s6: connecting the two trained models to realize the end-to-end rapid detection of the insulator fault, and the method comprises the following steps:
s601: detecting various insulator regions in the image by using the modified YOLOv3 model;
s602: sending the detected insulator region image into tiny-YOLOv3 for fault detection;
s603: and restoring the fault detection result to the original image, and identifying by using a rectangular frame, a fault type and confidence character.
Preferably, the insulator data set image in the S1 is an aerial image of multiple types of insulators of the power transmission and transformation circuit, which is shot and collected at a specific machine position through an unmanned aerial vehicle according to the inspection standard.
Preferably, the insulator fault data set in S1 includes images of multiple insulator faults such as spontaneous explosion and leakage, where the images are images of a fault-containing insulator region clipped from the initial aerial image and keep the original size.
Preferably, the LabelImg tool in S2 is a visual image labeling tool, the image labeling tool is not only labelImg, but also other tools, different file forms are finally obtained, and then can be converted into data formats required by model training, an environment python + lxml needs to be configured before the tool is used, and the labeling tool can label multiple classes and can directly generate xml files.
Preferably, the prior box sizes of 13 × 13 and 26 × 26 in S3 are the prior box sizes obtained by the kmeans algorithm on the coco data set.
Preferably, the softmax function in S304 can "compress" a K-dimensional vector Z containing arbitrary real numbers into another K-dimensional real vector σ, such that each element ranges from 0 to 1, and the sum of all elements is 1.
Preferably, the Sigmoid function in 304 is often used as a threshold function of a neural network, mapping variables between 0 and 1.
Preferably, in the present invention, the softmax function is adopted again because the insulator detection model only needs to detect the insulator and does not specifically divide the insulator category.
Compared with the prior art, the invention has the beneficial effects that: the invention relates to a method for detecting faults of a power transmission and transformation line, which comprises the steps of using a deep neural network model to detect the faults, greatly improving the detection effect and the detection speed, improving the automation level of power transmission and transformation line inspection, and reducing the labor intensity of inspection personnel.
Drawings
FIG. 1 is a flow chart of the detection method of the present invention;
FIG. 2 is a diagram of the YOLOv3 model architecture of the present invention;
FIG. 3 is a flow chart of the detection process of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
Referring to fig. 1-3, the present invention provides a technical solution: the power transmission and transformation line insulator aerial image fault detection method based on the improved YOLOv3 comprises the following steps:
s1: establishing an insulator data set and an insulator fault data set, wherein the insulator data set image is an aerial image of various types of insulators of a power transmission and transformation line shot and collected by an unmanned aerial vehicle at a specific machine position according to a routing inspection standard, the insulator fault data set comprises images of various insulator faults such as self-explosion, electric leakage and the like, and the images are cut-out insulator region images containing faults in an initial aerial image and keep the original size;
s2: data enhancement is carried out by using modes of rotation, turnover, contrast enhancement and the like; the method comprises the steps of marking an image by using a LabelImg tool to obtain a corresponding XML file, organizing and storing a data set image and the corresponding XML file according to a PASCAL VOC data set format, wherein the LabelImg tool is a visual image marking tool, the image marking tool is not only labelImg, but also other tools, and finally obtaining different file forms, and then converting the file forms into data formats required by model training;
s3: carrying out model modification pruning operation by taking YOLOv3 as a basic framework to construct an insulator detection model, comprising the following steps of:
s301: because the fixed machine position shoots the inspection picture, the insulator can be roughly divided into a larger insulator shot in a close shot and a smaller insulator shot in a far shot, so that three-dimensional output in the YOLOv3 is changed into two-dimensional output, only the output dimensions of 13X 13 and 26X 26 are reserved, the output dimension of 52X 52 is abandoned, the calculated amount in the model detection process is reduced while the size of the model is reduced, and the size of the prior frame of 13X 13 and 26X 26 is the size of the prior frame obtained by the kmeans algorithm on the coco data set;
s302: because only one target object of the insulator is detected, the integral network of the YOLOv3 is too large and has excessive functions, the model can be reduced through pruning operation, and the detection speed is increased appropriately;
s303: when model pruning is carried out, in order to ensure that the detection effect of the model is not reduced, 3 anchor boxes are distributed to each original scale and are changed into 4 anchor boxes in each scale, the initial 8 anchor boxes are generated by using a K-means clustering algorithm of which the intersection ratio IOU value replaces the standard Euclidean distance, and more anchor boxes in each scale can properly improve the detection effect;
s304: YOLOv3 considers that the multiple classes are divided by using softmax, and the mutual exclusion among the classes is required to be satisfied, so that a softmax function is abandoned and a sigmoid function is adopted for the class division, in the invention, as an insulator detection model only needs to detect insulators and does not particularly divide insulator classes, the softmax function is adopted again, the softmax function can compress a K-dimensional vector Z containing any real number into another K-dimensional real vector sigma, so that the range of each element is between 0 and 1, the sum of all the elements is 1, the sigmoid function is often used as a threshold function of a neural network, and variables are mapped between 0 and 1;
s4: building a standard YOLOv3-tiny network;
s5: respectively carrying out model training on the modified YOLOv3 network and the modified YOLOv3-tiny by using an insulator data set and an insulator fault data set;
s6: connecting the two trained models to realize the end-to-end rapid detection of the insulator fault, and the method comprises the following steps:
s601: detecting various insulator regions in the image by using the modified YOLOv3 model;
s602: sending the detected insulator region image into tiny-YOLOv3 for fault detection;
s603: and restoring the fault detection result to the original image, and identifying by using a rectangular frame, a fault type and confidence character.
Example two
Referring to fig. 1-3, the present invention provides a technical solution: the fault detection method of the aerial image of the insulator of the power transmission and transformation line based on the improved YOLOv3 comprises the following steps:
s1: establishing an insulator data set and an insulator fault data set, wherein the insulator data set image is an aerial image of various types of insulators of a power transmission and transformation line shot and collected by an unmanned aerial vehicle at a specific machine position according to a routing inspection standard, the insulator fault data set comprises images of various insulator faults such as self-explosion, electric leakage and the like, and the images are cut-out insulator region images containing faults in an initial aerial image and keep the original size;
s2: data enhancement is carried out by using modes of rotation, turnover, contrast enhancement and the like; the method comprises the steps that a LabelImg tool is used for marking images to obtain corresponding XML files, then the data set images and the corresponding XML files are organized and stored according to the format of a PASCAL VOC data set, the LabelImg tool is a visual image marking tool, the image marking tool is not only labelImg but also other tools, finally different file forms are obtained, then the file forms can be converted into data formats required by model training, an environment python + lxml is required to be configured before the tool is used, and a marking tool capable of marking multiple classes and directly generating XML files is provided;
s3: carrying out model modification and pruning operation by taking YOLOv3 as a basic framework to construct an insulator detection model, and comprising the following steps of:
s301: because the fixed machine position shoots the inspection picture, the insulator can be roughly divided into a larger insulator shot in a close shot and a smaller insulator shot in a far shot, so that three-dimensional output in the YOLOv3 is changed into two-dimensional output, only the output dimensions of 13X 13 and 26X 26 are reserved, the output dimension of 52X 52 is abandoned, the calculated amount in the model detection process is reduced while the size of the model is reduced, and the size of the prior frame of 13X 13 and 26X 26 is the size of the prior frame obtained by the kmeans algorithm on the coco data set;
s302: because only one target object of the insulator is detected, the integral network of the YOLOv3 is too large and has excessive functions, the model can be reduced through pruning operation, and the detection speed is increased appropriately;
s303: when model pruning is carried out, in order to ensure that the detection effect of the model is not reduced, 3 anchor boxes are distributed to each original scale and are changed into 6 anchor boxes in each scale, the initial 12 anchor boxes are generated by using a K-means clustering algorithm of which the intersection ratio IOU value replaces the standard Euclidean distance, and more anchor boxes in each scale can properly improve the detection effect;
s304: YOLOv3 considers that the multiple classes are divided by using softmax, and the mutual exclusion among the classes is required to be satisfied, so that a softmax function is abandoned and a sigmoid function is adopted for the class division, in the invention, as an insulator detection model only needs to detect insulators and does not particularly divide insulator classes, the softmax function is adopted again, the softmax function can compress a K-dimensional vector Z containing any real number into another K-dimensional real vector sigma, so that the range of each element is between 0 and 1, the sum of all the elements is 1, the sigmoid function is often used as a threshold function of a neural network, and variables are mapped between 0 and 1;
s4: building a standard YOLOv3-tiny network;
s5: respectively carrying out model training on the modified YOLOv3 network and the modified YOLOv3-tiny by using an insulator data set and an insulator fault data set;
s6: connecting the two trained models to realize the end-to-end rapid detection of the insulator fault, and the method comprises the following steps:
s601: detecting various insulator regions in the image by using the modified YOLOv3 model;
s602: sending the detected insulator region image into tiny-YOLOv3 for fault detection;
s603: and restoring the fault detection result to the original image, and identifying by using a rectangular frame, a fault type and confidence character.
Through comparison experiments of the two groups of embodiments, the first embodiment and the second embodiment can detect faults, the first embodiment and the second embodiment are different in anchor box distributed in each scale, too many anchor boxes distributed in the first embodiment cause slow detection speed, the first embodiment does not have high detection speed, a deep neural network model is used for fault detection, real-time routing inspection of a power transmission and transformation line can be achieved, the detection effect and the detection speed are greatly improved, the automation level of routing inspection of the power transmission and transformation line is improved, and the labor intensity of routing inspection personnel is reduced.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (7)
1. The power transmission and transformation line insulator aerial image fault detection method based on the improved YOLOv3 is characterized by comprising the following steps of:
s1: establishing an insulation data set and an insulator fault data set;
s2: performing data enhancement by using rotation, turnover and contrast enhancement; marking the image by using a LabelImg tool to obtain a corresponding XML file, and organizing and storing the data set image and the corresponding XML file according to a PASCAL VOC data set format;
s3: carrying out model modification pruning operation by taking YOLOv3 as a basic framework to construct an insulator detection model, comprising the following steps of:
s301: because the fixed machine position shoots the inspection picture, the insulator is divided into the insulator shot in a close view and the insulator shot in a distant view, so that three-scale output in YOLOv3 is changed into two-scale output, only the output scales of 13 × 13 and 26 × 26 are reserved, and the output scale of 52 × 52 is abandoned;
s302: carrying out model reduction through pruning operation;
s303: while carrying out model pruning, changing 3 anchor boxes allocated to each original scale into 4 anchor boxes at each scale, and generating initial 8 anchor boxes by using a K-means clustering algorithm of which the intersection-to-average ratio IOU value replaces the standard Euclidean distance;
s304: classifying by adopting a sigmoid function;
s4: building a standard YOLOv3-tiny network;
s5: respectively performing model training on the modified YOLOv3 network and the modified YOLOv3-tiny by using an insulator data set and an insulator fault data set;
s6: connecting the two trained models to realize the end-to-end rapid detection of the insulator fault, and the method comprises the following steps:
s601: detecting various insulator regions in the image using the modified YOLOv3 model;
s602: sending the detected insulator region image into YOLOv3-tiny for fault detection;
s603: and restoring the fault detection result to the original image, and identifying by using a rectangular frame, a fault type and confidence character.
2. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: and the insulator data set image in the S1 is an aerial image of various types of insulators of the power transmission and transformation line, which is shot and collected at a specific machine position through an unmanned aerial vehicle according to the inspection standard.
3. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: and the insulator fault data set in the S1 comprises self-explosion and leakage insulator fault images, and the images are the images of the insulator region containing the fault cut from the initial aerial image and keep the original size.
4. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: the LabelImg tool in the S2 is a visual image marking tool, a file form is obtained and then converted into a data format required by model training, an environment python + lxml is configured before the tool is used, and a labeling tool capable of labeling multiple classes and directly generating an xml file is provided.
5. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: the prior box sizes of 13 × 13 and 26 × 26 in S3 are the prior box sizes obtained by the K-means algorithm on the coco data set.
6. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: the softmax function in S304 can "compress" a K-dimensional vector Z containing arbitrary real numbers into a K-dimensional real vector σ on the other Z-axis, such that each element ranges from 0 to 1, and the sum of all elements is 1.
7. The improved YOLOv 3-based power transmission and transformation line insulator aerial image fault detection method as claimed in claim 1, wherein: the Sigmoid function in 304 is often used as a threshold function for neural networks, mapping variables between 0-1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910310921.0A CN110033453B (en) | 2019-04-18 | 2019-04-18 | Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910310921.0A CN110033453B (en) | 2019-04-18 | 2019-04-18 | Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110033453A CN110033453A (en) | 2019-07-19 |
| CN110033453B true CN110033453B (en) | 2023-02-24 |
Family
ID=67238836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910310921.0A Active CN110033453B (en) | 2019-04-18 | 2019-04-18 | Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110033453B (en) |
Families Citing this family (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110517228A (en) * | 2019-07-30 | 2019-11-29 | 浙江农林大学 | Trunk image rapid detection method based on convolutional neural networks and transfer learning |
| CN110543986A (en) * | 2019-08-27 | 2019-12-06 | 广东电网有限责任公司 | Intelligent monitoring system and monitoring method for external hidden danger of power transmission line |
| CN110705397A (en) * | 2019-09-19 | 2020-01-17 | 东南大学 | YOLOv3 pruning identification insulator defect method suitable for small field sample amount |
| CN111337789A (en) * | 2019-10-23 | 2020-06-26 | 西安科技大学 | Method and system for detecting fault electrical element in high-voltage transmission line |
| CN110826473A (en) * | 2019-11-01 | 2020-02-21 | 广东电网有限责任公司 | Neural network-based automatic insulator image identification method |
| CN110992307A (en) * | 2019-11-04 | 2020-04-10 | 华北电力大学（保定） | Insulator positioning and identifying method and device based on YOLO |
| CN110826520B (en) * | 2019-11-14 | 2023-05-26 | 燕山大学 | Port grab bucket detection method based on improved YOLOv3-tiny algorithm |
| CN110956176B (en) * | 2019-11-14 | 2023-06-16 | 国网山西省电力公司大同供电公司 | Method and detection device for fitting transmission line between tower bases |
| CN111161210A (en) * | 2019-12-02 | 2020-05-15 | 国网浙江省电力有限公司 | Method for detecting small parts in electric meter photo |
| CN111062915B (en) * | 2019-12-03 | 2023-10-24 | 浙江工业大学 | Real-time steel pipe defect detection method based on improved YOLOv3 model |
| CN111160440B (en) * | 2019-12-24 | 2023-11-21 | 广东省智能制造研究所 | Deep learning-based safety helmet wearing detection method and device |
| CN111239550A (en) * | 2020-02-27 | 2020-06-05 | 东南大学 | Unmanned aerial vehicle full-automatic multi-machine cooperative power transmission line fault identification method for inspection |
| CN111582323B (en) * | 2020-04-17 | 2023-06-16 | 山东信通电子股份有限公司 | Transmission line channel detection method, device and medium |
| CN111753666B (en) * | 2020-05-21 | 2024-01-23 | 西安科技大学 | Small target fault detection method, detection system and storage medium for power transmission line |
| CN111784685B (en) * | 2020-07-17 | 2023-08-18 | 国网湖南省电力有限公司 | Power transmission line defect image identification method based on cloud edge cooperative detection |
| CN111784692B (en) * | 2020-08-11 | 2024-07-02 | 国网内蒙古东部电力有限公司 | Method and device for detecting defects of insulators in power system and electronic equipment |
| CN112036464A (en) * | 2020-08-26 | 2020-12-04 | 国家电网有限公司 | Insulator infrared image fault detection method based on YOLOv3-tiny algorithm |
| CN112051270B (en) * | 2020-09-10 | 2021-08-13 | 广州云从人工智能技术有限公司 | Power transmission line defect detection method, system, equipment and medium |
| CN112541389B (en) * | 2020-09-29 | 2023-08-22 | 西安交通大学 | Transmission line fault detection method based on EfficientDet network |
| CN112183667B (en) * | 2020-10-31 | 2022-06-14 | 哈尔滨理工大学 | Insulator fault detection method in cooperation with deep learning |
| CN112598054B (en) * | 2020-12-21 | 2023-09-22 | 福建京力信息科技有限公司 | Power transmission and transformation project quality common disease prevention and detection method based on deep learning |
| CN112614130A (en) * | 2021-01-04 | 2021-04-06 | 东华大学 | Unmanned aerial vehicle power transmission line insulator fault detection method based on 5G transmission and YOLOv3 |
| CN112712516B (en) * | 2021-01-08 | 2021-10-15 | 哈尔滨市科佳通用机电股份有限公司 | High-speed rail bottom rubber strip fault detection method and system based on YOLOv5 |
| CN113112489B (en) * | 2021-04-22 | 2022-11-15 | 池州学院 | Insulator string-dropping fault detection method based on cascade detection model |
| CN113297915A (en) * | 2021-04-28 | 2021-08-24 | 江苏师范大学 | Insulator recognition target detection method based on unmanned aerial vehicle inspection |
| CN113239838B (en) * | 2021-05-24 | 2022-09-20 | 国网湖北省电力有限公司黄石供电公司 | Unmanned aerial vehicle intelligent inspection identification method for power transmission tower |
| US11836968B1 (en) * | 2022-12-08 | 2023-12-05 | Sas Institute, Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
| US12002256B1 (en) | 2022-12-08 | 2024-06-04 | Sas Institute Inc. | Systems and methods for configuring and using a multi-stage object classification and condition pipeline |
| CN116721095B (en) * | 2023-08-04 | 2023-11-03 | 杭州瑞琦信息技术有限公司 | Aerial photographing road illumination fault detection method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BR102016016197A2 (en) * | 2016-07-12 | 2018-02-06 | Instituto Federal De Educação, Ciência E Tecnologia Do Ceará | INSULATOR FAULT SIGNALING SYSTEM IN HIGH-VOLTAGE STRUCTURES USING ULTRASOUND. |
| CN108527382A (en) * | 2018-04-09 | 2018-09-14 | 上海方立数码科技有限公司 | A kind of crusing robot |
| CN108596886A (en) * | 2018-04-17 | 2018-09-28 | 福州大学 | Aerial Images insulator based on deep learning falls piece fault rapid detecting method |
| CN108629288A (en) * | 2018-04-09 | 2018-10-09 | 华中科技大学 | A kind of gesture identification model training method, gesture identification method and system |
| CN108647655A (en) * | 2018-05-16 | 2018-10-12 | 北京工业大学 | Low latitude aerial images power line foreign matter detecting method based on light-duty convolutional neural networks |
| CN109325454A (en) * | 2018-09-28 | 2019-02-12 | 合肥工业大学 | A kind of static gesture real-time identification method based on YOLOv3 |
-
2019
- 2019-04-18 CN CN201910310921.0A patent/CN110033453B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BR102016016197A2 (en) * | 2016-07-12 | 2018-02-06 | Instituto Federal De Educação, Ciência E Tecnologia Do Ceará | INSULATOR FAULT SIGNALING SYSTEM IN HIGH-VOLTAGE STRUCTURES USING ULTRASOUND. |
| CN108527382A (en) * | 2018-04-09 | 2018-09-14 | 上海方立数码科技有限公司 | A kind of crusing robot |
| CN108629288A (en) * | 2018-04-09 | 2018-10-09 | 华中科技大学 | A kind of gesture identification model training method, gesture identification method and system |
| CN108596886A (en) * | 2018-04-17 | 2018-09-28 | 福州大学 | Aerial Images insulator based on deep learning falls piece fault rapid detecting method |
| CN108647655A (en) * | 2018-05-16 | 2018-10-12 | 北京工业大学 | Low latitude aerial images power line foreign matter detecting method based on light-duty convolutional neural networks |
| CN109325454A (en) * | 2018-09-28 | 2019-02-12 | 合肥工业大学 | A kind of static gesture real-time identification method based on YOLOv3 |
Non-Patent Citations (7)
| Title |
|---|
| Rail Surface Defect Detection Method Based on YOLOv3 Deep Learning Networks;Song Yanan等;《2018 Chinese Automation Congress(CAC)》;20190124;第1563-1568页 * |
| 基于卷积神经网络的工件缺陷检测研究;于科为;《信息与电脑(理论版)》;20181115(第21期);第7-9、12页 * |
| 基于改进损失函数的YOLOv3网络;吕铄等;《计算机系统应用》;20190228;第28卷(第2期);第1-7页 * |
| 基于深度神经压缩的YOLO加速研究;陈莉君等;《现代计算机》;20190415;第1-7页 * |
| 航拍图像中绝缘子目标检测的研究;高强等;《电测与仪表》;20190107;第1-7页 * |
| 轻量级YOLOV3的绝缘子缺陷检测方法;吴涛等;《计算机工程》;20190307;第1-8页 * |
| 遥感图像中飞机的改进YOLOv3实时检测算法;戴伟聪等;《光电工程》;20181210;第45卷(第12期);第180350-1-180350-9页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110033453A (en) | 2019-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110033453B (en) | Power transmission and transformation line insulator aerial image fault detection method based on improved YOLOv3 | |
| CN111695622B (en) | Identification model training method, identification method and identification device for substation operation scene | |
| CN112379231B (en) | Equipment detection method and device based on multispectral image | |
| CN112184692B (en) | Multi-target detection method for power transmission line | |
| CN112990310A (en) | Artificial intelligence system and method for serving electric power robot | |
| CN107292886B (en) | Target object intrusion detection method and device based on grid division and neural network | |
| CN113515655B (en) | Fault identification method and device based on image classification | |
| CN111985455A (en) | Training and identifying method and device for photovoltaic module visible light fault model | |
| CN113344852A (en) | Target detection method and device for power scene general-purpose article and storage medium | |
| CN111241615A (en) | Highly realistic multi-source fusion three-dimensional modeling method for transformer substation | |
| CN112116594A (en) | Wind floating foreign matter identification method and device based on semantic segmentation | |
| CN112052855A (en) | License plate recognition method and device based on deep learning | |
| CN110751642A (en) | Insulator crack detection method and system | |
| CN110414351A (en) | The intelligent scissor and recognition methods, system and storage medium of substation | |
| CN112446376B (en) | Intelligent segmentation and compression method for industrial image | |
| CN115147591A (en) | Transformer equipment infrared image voltage heating type defect diagnosis method and system | |
| CN111402250A (en) | Machine vision defect detection method and platform based on edge calculation | |
| CN110334818B (en) | Method and system for automatically identifying pipeline | |
| CN210835242U (en) | Defect sample collection system of electric power tower | |
| CN113610838A (en) | Bolt defect data set expansion method | |
| CN114821042A (en) | R-FCN disconnecting link detection method combining local features and global features | |
| CN113869122A (en) | Distribution network engineering reinforced control method | |
| CN113378918A (en) | Insulator binding wire state detection method based on metric learning | |
| Gao et al. | Intelligent appearance quality detection of air conditioner external unit and dataset construction | |
| CN117290766B (en) | Intelligent detection method for high-voltage switch equipment in outdoor environment based on feature selector |
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 |