CN113592822A - Insulator defect positioning method for power inspection image - Google Patents
Insulator defect positioning method for power inspection image Download PDFInfo
- Publication number
- CN113592822A CN113592822A CN202110880016.6A CN202110880016A CN113592822A CN 113592822 A CN113592822 A CN 113592822A CN 202110880016 A CN202110880016 A CN 202110880016A CN 113592822 A CN113592822 A CN 113592822A
- Authority
- CN
- China
- Prior art keywords
- insulator
- inspection image
- insulators
- segmentation
- pixel points
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000012212 insulator Substances 0.000 title claims abstract description 249
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000007689 inspection Methods 0.000 title claims abstract description 69
- 230000007547 defect Effects 0.000 title claims abstract description 41
- 230000011218 segmentation Effects 0.000 claims abstract description 61
- 238000001514 detection method Methods 0.000 claims abstract description 52
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000013135 deep learning Methods 0.000 claims abstract description 22
- 238000010586 diagram Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 14
- 230000006835 compression Effects 0.000 claims description 12
- 238000007906 compression Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 9
- 238000005520 cutting process Methods 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 230000003993 interaction Effects 0.000 claims description 5
- 230000003321 amplification Effects 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 4
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 238000013508 migration Methods 0.000 claims description 2
- 230000005012 migration Effects 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims 2
- 230000005540 biological transmission Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 5
- 230000002950 deficient Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000010248 power generation Methods 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000009432 framing Methods 0.000 description 3
- 238000002372 labelling Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000003989 dielectric material Substances 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000000149 penetrating effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 101100272670 Aromatoleum evansii boxB gene Proteins 0.000 description 1
- 238000000137 annealing Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007664 blowing Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013434 data augmentation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000010292 electrical insulation Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material 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
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
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
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- 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
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
-
- 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
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an insulator defect positioning method of a power inspection image, which comprises the following steps: training a target detection algorithm by using an insulator detection data set to obtain an insulator identification model; training the improved deep learning network U-Net by utilizing an edge segmentation data set to obtain an insulator segmentation model; inputting the inspection image to be processed into an insulator identification model for detection to obtain the number and the position of insulators of the inspection image to be processed, and marking the insulators by a square frame; inputting the marked image into the insulator segmentation model in the second step for segmentation to obtain a binary insulator image; and performing mathematical modeling on the insulator strings in the binarized insulator images by a least square method, detecting the defects of the insulators by using the mathematical modeling, and judging the positions and the number of the defects of the insulator strings. The invention improves the detection precision and the segmentation precision of the insulator, has certain robustness and is not influenced by the size of the insulator in the inspection image.
Description
Technical Field
The invention relates to the technical field of power transmission line inspection, in particular to a method for positioning insulator defects of power inspection images.
Background
With the development of society and the continuous improvement of productivity, electric energy plays an indispensable role in daily life of people, and the operation of large-scale equipment and the smooth progress of scientific research in the aspects of country, society, military, life and the like can not play a role in the electric energy. In addition to traditional thermal power generation, the nation strongly supports novel energy sources such as wind power generation, solar power generation and the like to balance increasing electricity consumption of the nation. According to related data, the generated energy of China stably stays in the world first in recent years, and is a big country for power generation and utilization.
The power system plays a very important role in the production and life of the nation, so that the maintenance of the stable and safe operation of the power system is very important. An overhead transmission line is an important part of an electric power system and is composed of a line tower, a lead, an insulator, grounding equipment and the like. The insulator is a special insulating control, and the main function of the insulator is to realize electrical insulation and mechanical fixation. Therefore, the state of the insulator directly affects the normal operation of the transmission line. Because the insulator is exposed in the air for a long time, the insulator is easy to be damaged and fall in strings after long-term wind blowing oxidation, and the insulator is easy to be defective in extreme weather such as rainstorm, snowstorm, thunder and lightning and the like. In these cases, the insulator may lose its insulating and mechanical fixing functions, seriously threatening the safe operation of the grid. Therefore, the power grid staff can regularly patrol the insulators on the power transmission line, and the safe and stable operation of the power grid is ensured.
The electric power inspection mainly comprises two modes of manual inspection and unmanned aerial vehicle inspection. The manual inspection requires a worker to work on a power transmission tower pole, and has the following disadvantages: the difficulty of high-altitude operation is too high, workers need to climb to an electric tower and a power transmission line of dozens of meters or even dozens of meters for inspection, and particularly in areas with complex geographic environments, the difficulty of manual inspection is increased; secondly, certain potential safety hazards exist in high-altitude operation, and accidents can occur if workers are improperly protected when working at high altitude; finally, the staff cannot move quickly due to the influence of the high-altitude work site, so that the manual inspection efficiency is low. At present, electric power equipment is mainly overhauled in a manual inspection mode in China, but with the development of society and the progress of science and technology, a national power grid tries to find a new method to replace the traditional manual inspection mode. In recent years, the quad-rotor unmanned aerial vehicle has entered the lives of people and is widely applied to the field of aerial photography. In the electric power industry, people are also making an effort to research and utilize quad-rotor unmanned aerial vehicles to patrol electric power. Utilize unmanned aerial vehicle to carry out electric power and patrol and examine some shortcomings that have compensatied artifical the patrolling and examining. At first unmanned aerial vehicle can not be subject to the influence of geographical environment on every side, still can normally work under the geomorphic environment of complicacy, and the nimble free characteristics of unmanned aerial vehicle make it can patrol and examine the artifical place that is difficult to reach, has not only reduced the artifical risk of patrolling and examining probably leading to the fact the staff, also greatly reduced the degree of difficulty of patrolling and examining, also greatly improved the efficiency of patrolling and examining simultaneously.
A large number of inspection images can be shot in the process of unmanned aerial vehicle power inspection, and how to process the inspection images and accurately detect the defects of the insulators is important content needing to be researched. In recent years, a method based on deep learning is rapidly developed in the field of image processing, and a patrol image is processed through a deep learning algorithm, so that a model can learn the relevant characteristics of an insulator, the insulator can be quickly and efficiently identified and segmented, and defect detection is successfully completed. The process does not need human participation, reduces errors caused by human errors, and has important significance for the intelligent development of the power grid.
Disclosure of Invention
Aiming at the technical problem that the existing image processing method is not suitable for insulator defect detection, the invention provides the insulator defect positioning method for the power inspection image, which realizes the defect detection of the insulator based on a deep learning method and has certain application value for maintaining the safe operation of a power grid and improving the intelligent degree of the power grid.
In order to achieve the purpose, the technical scheme of the invention is realized as follows: an insulator defect positioning method of a power inspection image comprises the following steps:
the method comprises the following steps: marking insulators in the collected inspection image, establishing an insulator detection data set, and training a target detection algorithm by using the insulator detection data set to obtain an insulator identification model;
step two: marking insulators in the collected inspection image, establishing an edge segmentation data set, and training an improved deep learning network U-Net by using the edge segmentation data set to obtain an insulator segmentation model;
step three: inputting the inspection image to be processed into an insulator identification model for detection to obtain the number and the position of insulators of the inspection image to be processed, and marking the insulators by a square frame; inputting the marked image into the insulator segmentation model in the second step for segmentation to obtain a binary insulator image;
step four: and performing mathematical modeling on the insulator strings in the binarized insulator images by a least square method, detecting the defects of the insulators by using the mathematical modeling, and judging the positions and the number of the defects of the insulator strings.
Marking insulators in the inspection image by using LabelImg software in the first step, framing all the insulators in the inspection image by using a square frame, selecting a category to which an object belongs, and establishing an insulator detection data set; the data volume of the insulator detection data set is increased by a data amplification method of horizontal turning, random cutting and scaling transformation;
marking insulators in the inspection image by using Labelme software, extracting edges by adopting a Canny operator to serve as an edge segmentation data set, and increasing the data volume of the edge segmentation data set by adopting a turning, random cutting and scaling transformation mode;
after marking of all insulators on one picture, generating a json file by using Labelme software, wherein the json file records information of each marked insulator; and converting json files of all the inspection images into mask label files in batches, and performing binarization processing on the mask label files to obtain an edge segmentation data set.
The target detection algorithm adopts a Yolov3 algorithm based on a GIoU strategy, and a migration learning and small batch gradient descent method is adopted to train the insulator recognition model.
The implementation method of the GIoU strategy comprises the following steps:
1) calculate actual Box BgArea of (d):wherein, the actual frame BpHas the coordinates of the bounding box of
2) Calculate prediction Block BpArea of (d):wherein, the prediction frame BpHas the coordinates of the bounding box ofAnd is
3) Calculate prediction Block BpAnd actual frame BgThe intersection of (I):
4) find the minimum closed frame BcThe coordinates of (a):
8) Calculating a loss function LIoU=1-IoU,LGIOU=1-GIoU。
The idea of multi-task learning is added in the improved deep learning network U-Net, and an edge extraction task is introduced as a network branch on the basis of task segmentation; an ECA attention mechanism module is added into the improved deep learning network U-Net.
The edge extraction task is equivalent to deep supervision on a network of the segmentation task, and a feature diagram finally output by a decoder is extracted for auxiliary edge extraction network branches; the idea of the multi-task learning is to add a branch to divide the edge on the basis of a backbone network, wherein the backbone network adopts a classical coding-decoding structure and consists of a coder and two decoders, the coder is used for extracting a characteristic diagram, and the two decoders are respectively used for extracting edge information and extracting the outline of the whole insulator.
The improved deep learning network U-Net is divided into a compression path and an expansion path, wherein the compression path consists of 4 blocks, each block uses 3 effective convolution volumes and 1 maximum pooling downsampling, then the number of feature maps obtained after downsampling is multiplied by 2, and a feature map with the size of 32 x 32 is obtained through a series of convolution pooling operations; the expansion path is also composed of 4 blocks, the size of the characteristic diagram is multiplied by 2 by deconvolution before each block starts, meanwhile, the number of the characteristic diagrams is halved, and then the characteristic diagrams are merged with the characteristic diagram of the compression path; clipping the feature graph of the compression path to obtain the feature graph with the same size as that of the expansion path, then normalizing, and outputting two feature graphs with the same size;
the ECA attention module performs per-channel global average pooling without dimensionality reduction, capturing local cross-channel interactions by considering each channel and its k neighbors; the size of k is determined by an adaptive determination method, and the size of the kernel k is proportional to the channel dimension.
The improved deep learning network U-Net is optimized by adopting a small-batch gradient descent method; in the improved deep learning network U-Net training process, an SGD optimizer is adopted, and a Cosine optimizing learning rate attenuation strategy is adopted, so that when the loss function value is closer to the global minimum value, the learning rate becomes smaller and smaller.
The method for detecting the insulator defects by using mathematical modeling comprises the following steps:
s1: establishing a rectangular coordinate system by taking the lower left corner pixel point of the binarized insulator image as an origin, the horizontal edge as an X axis and the vertical edge as a Y axis; fitting the insulator string in the insulator image by a least square method to obtain a straight line L passing through the center of the insulator string;
s2: obtaining a straight line D which is perpendicular to the straight line L by using a formula that the two straight lines are perpendicular to each other;
s3: scanning on the insulator string by using the straight line D to horizontally move, and recording the number of pixel points obtained by each scanning; comparing the sum of the pixel points on each insulator sheet on the insulator string with a discrimination threshold, and judging that the insulator sheets are normal when the sum of the pixel points is greater than the discrimination threshold; when the sum of the pixel points is smaller than a judgment threshold value, judging that the insulator sheet is damaged or lost;
s4: and after all the insulator sheets on the insulator string are scanned, displaying the positions and the number of the damaged insulators.
The method for fitting the insulator string in the insulator image by the least square method in step S1 includes: traversing all pixel points in the binarized insulator image, storing the abscissa of all white pixel points in an array x, and placing the ordinate of all white pixel points in an array y; fitting all the white pixel points by using a least square method to obtain a central straight line L of the insulator string; let the expression of the linear equation L be: y is AX + B, where a is the slope of the central line Z and B is the intercept of the central line Z;
the fitting principle of the least square method is to solve the partial derivative of the parameters of the straight line L so as to ensure that the actual value y and the observed value yiThe error between the two is minimal:wherein, the coordinates (x) of all white pixelsj,yj) J is 0,1,2, and.
The partial derivatives of the parameters A and B are obtained by respectively calculating:
finishing to obtain:
solving to obtain the values of the parameters A and B as follows:
data (x) of all observation pointsj,yj) And j is substituted into a linear equation of the straight line L to enable the linear equation to reach a minimum value Y, AX + B, and then the optimal estimation values A and B are worked out to obtain a fitted linear equation L.
The method for acquiring the discrimination threshold comprises the following steps: the insulator string comprises a plurality of insulator pieces according to the appearance of the insulator string, the sum of pixel points on each insulator piece is calculated according to the pixel points obtained by scanning, and the average value of the sum is taken as the discrimination threshold value of each binarized insulator image.
The invention has the beneficial effects that: identifying the insulators by using a target detection algorithm, and determining the positions and the number of the insulators; partitioning a pure insulator string by an image partitioning algorithm; after obtaining a pure insulator string, reading pixel points of the insulator string in the image; fitting pixel points of the insulator string by using a least square method to obtain a straight line L passing through the center of the insulator and a straight line D perpendicular to the straight line L; scanning the straight line D along the straight line L according to the regularity of the insulator string, and reading the number of pixel points scanned each time; the number of pixel points on each insulator sheet is calculated, the damaged or missing position of the insulator is obtained by comparing the judging threshold, and the defect of the insulator string can be effectively detected and positioned.
In the process of insulator detection, the loss function in the YOLO-v3 algorithm is improved, and the GIoU strategy is used for replacing the IoU strategy in the original loss function, so that the precision of insulator detection is improved; in the process of segmenting the insulator, the U-Net network is improved by combining the multi-task learning idea and the ECA attention mechanism, and the precision of insulator segmentation is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a block diagram of the YOLOv3 algorithm in the present invention.
Fig. 2 shows three cases in which the IoU policy is the same and the GIoU policy is different in the present invention, where (a) is a case where IoU is 0.33 and the GIoU is 0.33, (b) is a case where IoU is 0.33 and the GIoU is 0.24, and (c) is a case where IoU is 0.33 and the GIoU is-0.1.
Fig. 3 is a comparison graph of the results of insulator detection in the present invention, in which (a) is example 1 and (b) is example 2.
FIG. 4 is a structural diagram of U-Net in the present invention.
FIG. 5 is a diagram illustrating multi-task learning in the present invention.
FIG. 6 is a block diagram of an ECA attention module of the present invention.
Fig. 7 is a structural diagram of an improved U-Net network in the present invention.
Fig. 8 shows an example of the segmentation result I of the insulator according to the present invention, wherein (a) is the direct segmentation result of the U-Net algorithm, and (b) is the segmentation result of the present invention.
Fig. 9 shows an example of the result II of the insulator segmentation according to the present invention, wherein (a) is the result of the direct segmentation using the U-Net algorithm, and (b) is the result of the segmentation according to the present invention.
Fig. 10 is a flowchart of insulator defect detection in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
As shown in fig. 1, a method for locating an insulator defect of a power inspection image includes the following steps:
the method comprises the following steps: marking the insulators in the collected inspection image, establishing an insulator detection data set, and training a target detection algorithm by using the insulator detection data set to obtain an insulator identification model.
In terms of data, labeling insulators in the inspection image by using LabelImg software: importing the inspection image into LabelImg software, framing all insulators in the inspection image by using a square frame, selecting the category of an object, establishing an insulator detection data set, and increasing the data volume of the insulator detection data set by a data amplification method of horizontal overturning, random cutting and scaling transformation.
The insulator detection data set is used as a training data set, insulator pictures shot by a Dajiang M200 electric inspection unmanned aerial vehicle are added with the insulator pictures collected on the network, and then the data set with 1000 electric insulator images is manufactured through manual screening. In deep learning, the deeper the network, the more data is needed to support, and only enough data volume can train a better model. However, in practical consideration, the data sets of the power equipment are not easy to obtain, and the manufacturing process also needs to consume a lot of manpower and material resources, so that the data sets are manufactured in the best effort, and meanwhile, the data volume is increased by using some other methods. Data augmentation, also called data enhancement, means that limited data yields equivalent value to more data without substantially increasing the data. In the invention, three modes of horizontal turning, random cutting and scaling transformation are adopted to amplify the data.
The insulator recognition model can realize the insulator detection of the inspection image, and the target detection algorithm adopts a YOLOv3 algorithm based on a GIoU strategy. Aiming at the defects of the IoU strategy in the YOLOv3 algorithm, the GIoU strategy is used for replacing the defect, the insulator recognition model is trained by methods such as transfer learning and small-batch gradient descent in the experimental process, the improved algorithm detects insulators which are not detected by the original algorithm in partial images, and the recall rate and the detection accuracy are improved to a certain extent.
The entire YOLO-v3 network in the YOLOv3 algorithm has 252 layers, the whole network structure is shown in FIG. 1, a DBL module is a combination of convolution + BN + Leaky relu, and the DBL module is a basic component of the YOLO-v3 network. In resn, n represents a number indicating the number of the defective units res-unit contained in the defective block res-block. Concat denotes tensor splicing, splicing the upsampling of the middle layer of Darknet-53 and the later layer.
The insulator position is accurately positioned in the unmanned aerial vehicle inspection image, and the positioning method is the primary task of power inspection. And only if the insulator exists in the inspection image, subsequent segmentation and defect detection can be carried out. Therefore, on the basis of deep learning, the insulator is positioned by using a target detection algorithm. With the development of deep learning, a target detection method based on a Convolutional Neural Network (CNN) makes a significant breakthrough. The two-stage detection algorithm divides the detection problem into two stages: candidate regions are generated and classified and regressed. Such an algorithm requires a large number of candidate regions to be generated and is therefore time consuming but highly accurate. The single-stage detection algorithm does not need to generate a candidate region, and a detection result can be directly obtained through single detection, so that the detection speed is higher, but the accuracy is not as good as that of the two-stage detection algorithm. Considering timeliness in the inspection process of the unmanned aerial vehicle, mainly aiming at solving the problems of related work of insulator detection and positioning, a YOLOv3 algorithm with high detection speed is adopted, and considering the condition that the accuracy of the YOLO-v3 algorithm is insufficient, a GIoU strategy is used for replacing a IoU strategy in an original algorithm loss function.
IoU is called the intersection-union ratio, which calculates the ratio of the intersection and union of the "predicted bounding box" and the "true bounding box". In the original YOLO-v3 algorithm, the mean square error is used as a loss function, but the mean square error function is sensitive to scale and cannot reflect the prediction results of different qualities. Therefore, in a related study of a large number of YOLO-v3 algorithms, IoU between a prediction box and a real target box is often adopted as a measure for similarity between the two. IoU is a commonly used index in the target detection task, and the calculation formula of IoU is:
IoU as a loss function, the following two problems arise: 1) if the two boxes do not intersect, the distance between the two boxes (coincidence) cannot be reflected by definition. Meanwhile, as loss is equal to 0, no gradient feedback exists, and learning training cannot be carried out. 2) IoU do not accurately reflect the degree of overlap between the two. As shown in fig. 2, all three IoU are equal, but their degrees of overlap are not the same, with the better regression in fig. 2 (a) and the worst regression in fig. 2 (b). Aiming at the defects of IoU, a new index GIoU is adopted to replace IoU.
The implementation method of the GIoU strategy comprises the following steps:
and (3) outputting: l isIoU,LGIOU
3) computing a prediction boxBpAnd actual frame BgThe intersection of (I):
4) find the minimum closed frame BcThe coordinates of (a):
8)LIoU=1-IoU,LGIOU=1-GIoU。
The optimization is carried out by using a small batch gradient descent method, which has the following two advantages: on one hand, data is batched, and only one batch of data needs to be loaded when the computer updates the gradient every time, so that a large amount of memory is not occupied, and the running speed can be increased; on the other hand, the parameters can be updated in batches, the direction of the gradient is determined by data of each batch, the gradient is not easy to deviate when the gradient descends, and the randomness is reduced. In order to enable the gradient descent method to have better performance, the learning rate needs to be set in a reasonable range, learning instability can be caused by too high learning rate, and training time can be prolonged by too low learning rate. Setting a reasonable learning rate not only ensures stable training but also reduces training time. Thus, the learning rate is set to 0.01 just at the beginning of the first stage training and to 0.0005 near the end of the second stage training. And after the training is finished, identifying and detecting the insulators in the aerial images, and determining the positions and the number of the insulators.
Step two: marking the insulators in the collected inspection image, establishing an edge segmentation data set, and training the improved deep learning network U-Net by using the edge segmentation data set to obtain an insulator segmentation model.
And labeling the inspection image by using Labelme software, extracting edges by adopting a Canny operator to serve as an edge segmentation data set, and increasing the data volume of training the edge segmentation data set by adopting data amplification modes such as overturning, random cutting and the like. And importing the inspection image into Labelme software, framing the outline of the object to be labeled, and then selecting the category of the labeled object. After the labeling of all insulators on one picture is finished, the Labelme software can generate a json file, and the content in the json file records the information of each labeled insulator. And obtaining a json file which is correspondingly generated by each picture after all the pictures are marked. And the image segmentation task needs a corresponding label of the png/. bmp file, so the json file is converted into a mask label file in batch by writing a program. And after all json files are converted, performing binarization processing on the mask label file, and finishing the manufacture of the edge segmentation data set of the image.
According to the characteristics of the segmentation task and the idea of multi-task learning, the improved deep learning network U-Net adds a branch to segment the edge on the basis of the original main network, and adds an ECA attention mechanism module to enable the segmentation result to achieve a better effect.
The segmentation of the insulators in the inspection image is a key step in the defect detection scheme, and the effect of the segmented insulator binary image directly influences the result of the next defect detection. Insulators are generally present on high-voltage transmission lines, and the complex geographical environment background of the insulators makes the segmentation of the insulators difficult to achieve. If the traditional method is adopted, an ideal segmentation effect is difficult to achieve, so a deep learning segmentation method is adopted to process the insulator inspection image. The U-Net network is often used for the segmentation of medical images, and can achieve good segmentation effect under the condition of less data sets. The characteristics of the inspection image of the insulator and the medical image have certain similarity, the data volume is small due to the fact that the data set of the electric insulator is difficult to collect, and the factors are comprehensively considered.
The U-Net network has no fully connected operation and is a classic full convolution network. As shown in fig. 4, the U-Net network can be divided into two parts: a compression path and an expansion path. The compression path is composed of 4 blocks, each block uses 3 effective convolution volumes and 1 maximum pooling downsampling, then the number of the feature maps obtained after downsampling is multiplied by 2, and a series of convolution pooling operations are carried out to finally obtain the feature map with the size of 32 x 32. The extended path is also composed of 4 blocks, and each block multiplies the size of the feature map by 2 by deconvolution before starting, reduces the number of the feature maps by half (except for the last layer), and then merges with the feature map of the left symmetric compressed path. However, the feature maps of the left compression path and the right expansion path are different in size and cannot be directly merged. Therefore, the U-Net network cuts the characteristic diagram of the compression path to obtain the characteristic diagram with the same size as that of the expansion path and then performs normalization. Through the above operations, two characteristic maps with the size of 388 × 388 are finally output.
In the process of insulator segmentation, if the U-Net network is directly adopted for segmentation, the obtained result can segment the whole outline of the insulator, but the edge of the insulator is segmented and the edge is unclear. In order to better solve the problem of incomplete edges of the electric tower insulator segmentation, the invention adopts the idea of multi-task learning, and introduces an edge extraction task as a network branch on the basis of the segmentation task, as shown in fig. 5. Meanwhile, because the two tasks are very similar, the edge extraction task is also equivalent to deep supervision on a network of the segmentation task, and the problems of gradient disappearance and the like can be relieved. The invention applies the idea of multi-task learning to the U-Net segmentation network, extracts the characteristic diagram finally output by the decoder and applies the characteristic diagram to the auxiliary edge extraction network branch, so as to help the network to better extract the characteristics and pay more attention to the acquisition of the edge information of the insulator. The backbone part of the network adopts a classical coding-decoding structure, is similar to the design of U-Net, and consists of an encoder and two decoders, wherein the encoder is used for extracting a characteristic diagram, and the two decoders are respectively used for extracting edge information and extracting the outline of the whole insulator.
As shown in fig. 7, the present invention introduces an ECA attention module (an effective channel attention module for deep CNN), which not only can effectively perform cross-channel interaction, but also avoids dimension reduction. After per-channel global average pooling without dimensionality reduction, the ECA attention module captures local cross-channel interactions by considering each channel and its k neighbors. As shown in fig. 6, ECA may be implemented by fast one-dimensional convolution with size k, where k is the size of the kernel, which represents the coverage of local cross-channel interaction, i.e., how many neighbors are involved in the attention prediction of a channel. The size of k is determined adaptively, where the kernel size k is proportional to the channel dimension.
In the training strategy, the small-batch gradient descent method is still adopted for optimization, the error fluctuation during training is within a reasonable range, the correctness of the final result is ensured, the training process is accelerated, and the computer memory is saved, so that the batch size of each training is selected to be 8 after multiple attempts. The momentum parameter value was empirically set to 0.9 using the SGD optimizer. By adopting a Cosine Annealing learning rate attenuation strategy, when the loss function value is closer to the global minimum value in the gradient descending process, the learning rate becomes smaller and smaller to reach the minimum value, and the initial learning rate is set to be 3 e-3. Finally, through experiments, after iteration is carried out to 150 epochs, the loss function tends to be stable, and the training is completed. Finally, compared with the traditional segmentation method and the original algorithm of the U-Net network, the comparison experiment result shows that the trained insulator segmentation model has good performance, the segmented insulator string has clear outline and the segmentation result is better than the result of the direct segmentation of the original algorithm.
Step three: inputting the inspection image to be processed into an insulator identification model for detection to obtain the number and the position of insulators of the inspection image to be processed, and marking the insulators by a square frame; and inputting the marked image into the insulator segmentation model in the second step for segmentation to obtain a binary insulator image.
As shown in fig. 3, the result of the insulator detection by the insulator recognition model of the present invention can be seen from fig. 8 and 9, the improved algorithm of the present invention is significantly better than the original algorithm in terms of segmentation effect, and the segmentation of the insulator edge is clearer.
Step four: and performing mathematical modeling on the insulator strings in the binarized insulator images by a least square method, detecting the defects of the insulators by using the mathematical modeling, and judging the positions and the number of the defects of the insulator strings.
The distribution of pixel points of the insulator images has certain regularity, and according to the regularity, the insulator string is modeled by using a least square method to fit a straight line passing through the center point of the insulator; and then linearly scanning the insulator string, calculating the sum of the pixel points of each insulator sheet according to the distribution rule of the pixel points, and comparing the sum with a set threshold value to judge whether the insulator sheet has defects or not. The modeling method has the greatest advantage of certain robustness and is not influenced by the size of a target object in the inspection image.
The method comprises the steps that a pure insulator string is arranged in a binarized insulator image, after the pure insulator string is obtained, the position coordinates of pixel points of the insulator string in the image are read, the insulator string is subjected to least square fitting, a straight line L penetrating through the center of the insulator string is obtained, and meanwhile a straight line D perpendicular to the straight line L is obtained by utilizing a formula that the two straight lines are perpendicular to each other; scanning the insulator string by using a straight line D, and recording the number of pixel points obtained by each scanning; according to the shape of the insulator string, certain regularity exists, the sum of pixel points on each insulator sheet can be calculated, and the average value of the sum is taken as a discrimination threshold; comparing the sum of the pixel points on each insulator piece with a discrimination threshold, and if the sum is smaller than the discrimination threshold, judging that the insulator is damaged or defective; if the judgment threshold value is larger than the judgment threshold value, the insulator is judged to be normal. The insulator string is provided with a plurality of insulator sheets, and the insulator sheets are in an ellipsoid shape with thick middle parts and thin two ends.
As shown in fig. 10, the method for detecting the insulator defect by using mathematical modeling includes:
s1: establishing a rectangular coordinate system by taking the pixel point at the lower left corner of the segmented image as an origin, the horizontal edge as an X axis and the vertical edge as a Y axis; fitting the insulator string by a least square method to obtain a straight line L passing through the center of the insulator string.
According to the shape of the insulator string in the inspection image, the insulator string is linear, all white pixels are regularly distributed, and by means of the regularity, the insulator string can be fitted through a least square method to form a straight line L penetrating through the center of the insulator string. The least squares method is a commonly used method of fitting data by finding the best functional match of the data by minimizing the sum of the squares of the errors.
The insulator string in the binarized insulator image is a white pixel (with a value of 255), the other background elements are black pixels (with a value of 0), and the insulator string and the background are completely separated. And establishing a rectangular coordinate system by taking the lower left corner pixel point of the image as an original point, the horizontal edge as an X axis and the vertical edge as a Y axis. And traversing all the pixel points, storing the abscissa of all the white pixel points in an x array, and storing the ordinate of all the white pixel points in a y array. And after traversing all the pixel points, fitting all the white pixel points by using a least square method to obtain a central straight line L of the insulator string. First, assume that the expression of the linear equation L is:
Y=AX+B
where A is the slope of the central line L and B is the intercept of the central line L. According to the coordinates (x) of all the traversed white pixel pointsi,yi) I is 0,1,2, …, N is the number of white pixels, and parameters a and B of the center straight line L can be obtained. The fitting principle of the least square method is that the actual value y and the observed value y are enabled to be obtained by solving the partial derivatives of the parameters of the straight lineiThe error between the two is minimal:
the partial derivatives of the parameters A and B in the formula can be obtained:
the formula is arranged to obtain:
by solving the above equations, the values of parameters a and B can be obtained:
the principle of least square method to fit straight line is to take the data (x) of all observation pointsj,yj) N is substituted into the equation of the straight line L to a minimum value, and then the equation is solvedTo obtain the best estimated valueAndand a fitted linear L equation can be obtained. And drawing the straight line L in the binary image according to the equation of the straight line L.
S2: and obtaining a straight line D which is perpendicular to the straight line L by using a mathematical formula that the two straight lines are perpendicular to each other, scanning the insulator string by using the straight line D, and recording the number of pixel points obtained by each scanning.
And finding a straight line D vertical to the straight line L, and finding the starting point and the end point of the insulator string according to the number of pixel points on the straight line D. In the rectangular coordinate system of the previous step, the straight line D is perpendicular to the straight line L, and the slope of the straight line L is known, so that the slope of the straight line D can be obtained.
S3: and in the scanning process of the straight line D on the insulator string, recording the number of pixel points obtained in each scanning process. When an insulator sheet is scanned, counting and summing the number of pixel points on the insulator sheet, then comparing the number with a calculated discrimination threshold value, and when the number of the pixel points is greater than the discrimination threshold value, judging that the insulator sheet is normal; and when the number of the pixel points is smaller than the judgment threshold, judging that the insulator sheet is damaged or lost.
S4: and finally, after the insulator string is scanned, displaying the positions and the number of the damaged insulators.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. The method for positioning the insulator defects of the power inspection image is characterized by comprising the following steps of:
the method comprises the following steps: marking insulators in the collected inspection image, establishing an insulator detection data set, and training a target detection algorithm by using the insulator detection data set to obtain an insulator identification model;
step two: marking insulators in the collected inspection image, establishing an edge segmentation data set, and training an improved deep learning network U-Net by using the edge segmentation data set to obtain an insulator segmentation model;
step three: inputting the inspection image to be processed into an insulator identification model for detection to obtain the number and the position of insulators of the inspection image to be processed, and marking the insulators by a square frame; inputting the marked image into the insulator segmentation model in the second step for segmentation to obtain a binary insulator image;
step four: and performing mathematical modeling on the insulator strings in the binarized insulator images by a least square method, detecting the defects of the insulators by using the mathematical modeling, and judging the positions and the number of the defects of the insulator strings.
2. The insulator defect positioning method for the power inspection image according to claim 1, wherein in the first step, LabelImg software is used for marking insulators in the inspection image, all insulators in the inspection image are framed in a square frame, the category of the object is selected, and an insulator detection data set is established; the data volume of the insulator detection data set is increased by a data amplification method of horizontal turning, random cutting and scaling transformation;
marking insulators in the inspection image by using Labelme software, extracting edges by adopting a Canny operator to serve as an edge segmentation data set, and increasing the data volume of the edge segmentation data set by adopting a turning, random cutting and scaling transformation mode;
after marking of all insulators on one picture, generating a json file by using Labelme software, wherein the json file records information of each marked insulator; and converting json files of all the inspection images into mask label files in batches, and performing binarization processing on the mask label files to obtain an edge segmentation data set.
3. The insulator defect positioning method for the power inspection image according to claim 1 or 2, wherein the target detection algorithm adopts a YOLOv3 algorithm based on a GIoU strategy, and a migration learning and small batch gradient descent method is adopted to train an insulator recognition model.
4. The insulator defect positioning method for the power inspection image according to claim 3, wherein the implementation method of the GIoU strategy is as follows:
1) calculate actual Box BgArea of (d):wherein, the actual frame BpHas the coordinates of the bounding box of
2) Calculate prediction Block BpArea of (d):wherein, the prediction frame BpHas the coordinates of the bounding box ofAnd is
3) Calculate prediction Block BpAnd actual frame BgThe intersection of (I):
4) find the minimum closed frame BcThe coordinates of (a):
8) Calculating a loss function LIoU=1-IoU,LGIOU=1-GIoU。
5. The insulator defect positioning method for the power inspection image according to claim 1, wherein a multi-task learning idea is added to the improved deep learning network U-Net, and an edge extraction task is introduced as a network branch on the basis of a segmentation task; an ECA attention mechanism module is added into the improved deep learning network U-Net.
6. The insulator defect positioning method for the power inspection image according to claim 5, wherein the edge extraction task is equivalent to deep supervision of a network of the segmentation task, and a feature map finally output by the extraction decoder is used for auxiliary edge extraction network branches; the idea of the multi-task learning is to add a branch to divide the edge on the basis of a backbone network, wherein the backbone network adopts a classical coding-decoding structure and consists of a coder and two decoders, the coder is used for extracting a characteristic diagram, and the two decoders are respectively used for extracting edge information and extracting the outline of the whole insulator.
7. The insulator defect positioning method for the power inspection image according to claim 6, wherein the improved deep learning network U-Net is divided into two parts, namely a compression path and an expansion path, the compression path is composed of 4 blocks, each block uses 3 effective convolution volumes and 1 maximum pooling down-sampling, then the number of feature maps obtained after the down-sampling is multiplied by 2, and a feature map with the size of 32 x 32 is obtained through a series of convolution pooling operations; the expansion path is also composed of 4 blocks, the size of the characteristic diagram is multiplied by 2 by deconvolution before each block starts, meanwhile, the number of the characteristic diagrams is halved, and then the characteristic diagrams are merged with the characteristic diagram of the compression path; clipping the feature graph of the compression path to obtain the feature graph with the same size as that of the expansion path, then normalizing, and outputting two feature graphs with the same size;
the ECA attention module performs per-channel global average pooling without dimensionality reduction, capturing local cross-channel interactions by considering each channel and its k neighbors; the size of k is determined by an adaptive determination method, and the size of the kernel k is proportional to the channel dimension.
The improved deep learning network U-Net is optimized by adopting a small-batch gradient descent method; in the improved deep learning network U-Net training process, an SGD optimizer is adopted, and a Cosine optimizing learning rate attenuation strategy is adopted, so that when the loss function value is closer to the global minimum value, the learning rate becomes smaller and smaller.
8. The method for locating the insulator defect of the power inspection image according to claim 1, wherein the method for detecting the insulator defect by using mathematical modeling comprises the following steps:
s1: establishing a rectangular coordinate system by taking the lower left corner pixel point of the binarized insulator image as an origin, the horizontal edge as an X axis and the vertical edge as a Y axis; fitting the insulator string in the insulator image by a least square method to obtain a straight line L passing through the center of the insulator string;
s2: obtaining a straight line D which is perpendicular to the straight line L by using a formula that the two straight lines are perpendicular to each other;
s3: scanning on the insulator string by using the straight line D to horizontally move, and recording the number of pixel points obtained by each scanning; comparing the sum of the pixel points on each insulator sheet on the insulator string with a discrimination threshold, and judging that the insulator sheets are normal when the sum of the pixel points is greater than the discrimination threshold; when the sum of the pixel points is smaller than a judgment threshold value, judging that the insulator sheet is damaged or lost;
s4: and after all the insulator sheets on the insulator string are scanned, displaying the positions and the number of the damaged insulators.
9. The method for locating the insulator defects in the power inspection image according to claim 8, wherein the method for fitting the insulator strings in the insulator images by a least square method in the step S1 comprises the following steps: traversing all pixel points in the binarized insulator image, storing the abscissa of all white pixel points in an array x, and placing the ordinate of all white pixel points in an array y; fitting all the white pixel points by using a least square method to obtain a central straight line L of the insulator string; let the expression of the linear equation L be: y is AX + B, where a is the slope of the central line Z and B is the intercept of the central line Z;
the fitting principle of the least square method is to solve the partial derivative of the parameters of the straight line L so as to ensure that the actual value y and the observed value yiThe error between the two is minimal:wherein, the coordinates (x) of all white pixelsj,yj) J is 0,1,2, and.
The partial derivatives of the parameters A and B are obtained by respectively calculating:
finishing to obtain:
solving to obtain the values of the parameters A and B as follows:
data (x) of all observation pointsj,yj) And j is substituted into a linear equation of the straight line L to enable the linear equation to reach a minimum value Y, AX + B, and then the optimal estimation values A and B are worked out to obtain a fitted linear equation L.
10. The method for locating the insulator defect in the power inspection image according to claim 8 or 9, wherein the method for obtaining the discrimination threshold is as follows: the insulator string comprises a plurality of insulator pieces according to the appearance of the insulator string, the sum of pixel points on each insulator piece is calculated according to the pixel points obtained by scanning, and the average value of the sum is taken as the discrimination threshold value of each binarized insulator image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110880016.6A CN113592822B (en) | 2021-08-02 | 2021-08-02 | Insulator defect positioning method for electric power inspection image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110880016.6A CN113592822B (en) | 2021-08-02 | 2021-08-02 | Insulator defect positioning method for electric power inspection image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113592822A true CN113592822A (en) | 2021-11-02 |
CN113592822B CN113592822B (en) | 2024-02-09 |
Family
ID=78253620
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110880016.6A Active CN113592822B (en) | 2021-08-02 | 2021-08-02 | Insulator defect positioning method for electric power inspection image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113592822B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114061476A (en) * | 2021-11-17 | 2022-02-18 | 国网宁夏电力有限公司建设分公司 | Deflection detection method for insulator of power transmission line |
CN115082695A (en) * | 2022-05-31 | 2022-09-20 | 中国科学院沈阳自动化研究所 | Transformer substation insulator string modeling and detecting method based on improved Yolov5 |
CN115115590A (en) * | 2022-06-23 | 2022-09-27 | 华南理工大学 | Composite insulator overheating defect detection method based on rotary RetinaNet |
CN115272310A (en) * | 2022-09-26 | 2022-11-01 | 江苏智云天工科技有限公司 | Method and device for detecting defects of workpiece |
CN115870256A (en) * | 2022-12-02 | 2023-03-31 | 广西电网有限责任公司电力科学研究院 | Composite insulator cleans and device of detecting a flaw |
CN116152211A (en) * | 2023-02-28 | 2023-05-23 | 哈尔滨市科佳通用机电股份有限公司 | Identification method for brake shoe abrasion overrun fault |
CN116468730A (en) * | 2023-06-20 | 2023-07-21 | 齐鲁工业大学(山东省科学院) | Aerial insulator image defect detection method based on YOLOv5 algorithm |
CN118097092A (en) * | 2024-04-29 | 2024-05-28 | 西北工业大学 | Intelligent inspection method, device and system for press-connection quality of miniature pins of electric connector |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109118479A (en) * | 2018-07-26 | 2019-01-01 | 中睿能源(北京)有限公司 | Defects of insulator identification positioning device and method based on capsule network |
CN112184746A (en) * | 2020-08-27 | 2021-01-05 | 西北工业大学 | Transmission line insulator defect analysis method |
CN112233092A (en) * | 2020-10-16 | 2021-01-15 | 广东技术师范大学 | Deep learning method for intelligent defect detection of unmanned aerial vehicle power inspection |
CN112508973A (en) * | 2020-10-19 | 2021-03-16 | 杭州电子科技大学 | MRI image segmentation method based on deep learning |
US20210224512A1 (en) * | 2020-01-17 | 2021-07-22 | Wuyi University | Danet-based drone patrol and inspection system for coastline floating garbage |
-
2021
- 2021-08-02 CN CN202110880016.6A patent/CN113592822B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109118479A (en) * | 2018-07-26 | 2019-01-01 | 中睿能源(北京)有限公司 | Defects of insulator identification positioning device and method based on capsule network |
US20210224512A1 (en) * | 2020-01-17 | 2021-07-22 | Wuyi University | Danet-based drone patrol and inspection system for coastline floating garbage |
CN112184746A (en) * | 2020-08-27 | 2021-01-05 | 西北工业大学 | Transmission line insulator defect analysis method |
CN112233092A (en) * | 2020-10-16 | 2021-01-15 | 广东技术师范大学 | Deep learning method for intelligent defect detection of unmanned aerial vehicle power inspection |
CN112508973A (en) * | 2020-10-19 | 2021-03-16 | 杭州电子科技大学 | MRI image segmentation method based on deep learning |
Non-Patent Citations (4)
Title |
---|
HAMID REZATOFIGHI ET AL.: "Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression", pages 658 - 666 * |
LINA YAO,QIN YAOYAO: "Insulator Detection Dased on GIOU-YOLOv3" * |
袁金丽 等: "改进U 型残差网络用于肺结节检测", pages 1 * |
陈文浩 等: "无人机电网巡检中的绝缘子缺陷检测与定位", vol. 39, no. 1, pages 5 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114061476A (en) * | 2021-11-17 | 2022-02-18 | 国网宁夏电力有限公司建设分公司 | Deflection detection method for insulator of power transmission line |
CN115082695A (en) * | 2022-05-31 | 2022-09-20 | 中国科学院沈阳自动化研究所 | Transformer substation insulator string modeling and detecting method based on improved Yolov5 |
CN115115590A (en) * | 2022-06-23 | 2022-09-27 | 华南理工大学 | Composite insulator overheating defect detection method based on rotary RetinaNet |
CN115115590B (en) * | 2022-06-23 | 2024-03-08 | 华南理工大学 | Composite insulator overheat defect detection method based on rotary RetinaNet |
CN115272310A (en) * | 2022-09-26 | 2022-11-01 | 江苏智云天工科技有限公司 | Method and device for detecting defects of workpiece |
CN115870256A (en) * | 2022-12-02 | 2023-03-31 | 广西电网有限责任公司电力科学研究院 | Composite insulator cleans and device of detecting a flaw |
CN115870256B (en) * | 2022-12-02 | 2024-06-04 | 广西电网有限责任公司电力科学研究院 | Composite insulator cleans and flaw detection device |
CN116152211A (en) * | 2023-02-28 | 2023-05-23 | 哈尔滨市科佳通用机电股份有限公司 | Identification method for brake shoe abrasion overrun fault |
CN116468730A (en) * | 2023-06-20 | 2023-07-21 | 齐鲁工业大学(山东省科学院) | Aerial insulator image defect detection method based on YOLOv5 algorithm |
CN116468730B (en) * | 2023-06-20 | 2023-09-05 | 齐鲁工业大学(山东省科学院) | Aerial Insulator Image Defect Detection Method Based on YOLOv5 Algorithm |
CN118097092A (en) * | 2024-04-29 | 2024-05-28 | 西北工业大学 | Intelligent inspection method, device and system for press-connection quality of miniature pins of electric connector |
Also Published As
Publication number | Publication date |
---|---|
CN113592822B (en) | 2024-02-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113592822B (en) | Insulator defect positioning method for electric power inspection image | |
CN108961235B (en) | Defective insulator identification method based on YOLOv3 network and particle filter algorithm | |
CN111062282B (en) | Substation pointer instrument identification method based on improved YOLOV3 model | |
CN108846331B (en) | Video identification method for judging whether screw fastener of motor train unit chassis falls off or not | |
CN111814597A (en) | Urban function partitioning method coupling multi-label classification network and YOLO | |
CN111598942A (en) | Method and system for automatically positioning electric power facility instrument | |
CN116206112A (en) | Remote sensing image semantic segmentation method based on multi-scale feature fusion and SAM | |
CN113902792A (en) | Building height detection method and system based on improved RetinaNet network and electronic equipment | |
CN111680759B (en) | Power grid inspection insulator detection classification method | |
CN115205256A (en) | Power transmission line insulator defect detection method and system based on fusion of transfer learning | |
CN115131747A (en) | Knowledge distillation-based power transmission channel engineering vehicle target detection method and system | |
CN114694130A (en) | Method and device for detecting telegraph poles and pole numbers along railway based on deep learning | |
Wang et al. | Automatic identification and location of tunnel lining cracks | |
CN113902793A (en) | End-to-end building height prediction method and system based on single vision remote sensing image and electronic equipment | |
CN116580285B (en) | Railway insulator night target identification and detection method | |
CN113536944A (en) | Distribution line inspection data identification and analysis method based on image identification | |
CN116452604A (en) | Complex substation scene segmentation method, device and storage medium | |
CN116630989A (en) | Visual fault detection method and system for intelligent ammeter, electronic equipment and storage medium | |
CN116403071A (en) | Method and device for detecting few-sample concrete defects based on feature reconstruction | |
CN114418968A (en) | Power transmission line small target defect detection method based on deep learning | |
CN114187501A (en) | Package detection method, device and system | |
CN115170970B (en) | Method for detecting urban street landscape damage | |
CN117876362B (en) | Deep learning-based natural disaster damage assessment method and device | |
CN113378918B (en) | Insulator binding wire state detection method based on metric learning | |
CN117173385B (en) | Detection method, device, medium and equipment of transformer substation |
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 |