CN116596920A - Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle - Google Patents
Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle Download PDFInfo
- Publication number
- CN116596920A CN116596920A CN202310850373.7A CN202310850373A CN116596920A CN 116596920 A CN116596920 A CN 116596920A CN 202310850373 A CN202310850373 A CN 202310850373A CN 116596920 A CN116596920 A CN 116596920A
- Authority
- CN
- China
- Prior art keywords
- insulator
- porcelain
- network
- porcelain insulator
- aerial vehicle
- 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 178
- 229910052573 porcelain Inorganic materials 0.000 title claims abstract description 136
- 238000000691 measurement method Methods 0.000 title claims description 5
- 238000000034 method Methods 0.000 claims abstract description 44
- 238000001514 detection method Methods 0.000 claims abstract description 43
- 230000011218 segmentation Effects 0.000 claims abstract description 35
- 238000005457 optimization Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 11
- 230000004927 fusion Effects 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 9
- 238000010586 diagram Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 claims description 5
- 238000003709 image segmentation Methods 0.000 claims description 5
- 238000011176 pooling Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 14
- 238000001931 thermography Methods 0.000 description 3
- 230000015556 catabolic process Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 238000002329 infrared spectrum Methods 0.000 description 2
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000012466 permeate Substances 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1218—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing using optical methods; using charged particle, e.g. electron, beams or X-rays
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- 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
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- 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
- G06T2207/10048—Infrared 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
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/06—Recognition of objects for industrial automation
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Remote Sensing (AREA)
- Databases & Information Systems (AREA)
- Insulators (AREA)
Abstract
The invention discloses a method and a system for measuring zero of a long string porcelain insulator unmanned aerial vehicle in real time, wherein the method comprises the following steps: acquiring an infrared image of a porcelain insulator by using an unmanned aerial vehicle-mounted thermal infrared imager, selecting the porcelain insulator segment infrared image acquired in a short distance as a data set, and dividing the data set; the improved deep LabV3+ semantic segmentation network is constructed, the processed data set is input into the improved deep LabV3+ semantic segmentation network for training, the phantom module in the GhostNet network is adopted to replace the original backbone network in the deep LabV3+ semantic segmentation network, the calculation time and the parameter quantity are reduced, the real-time performance of detection is improved, the weighting of channel-level facing characteristics is realized by introducing the SE channel attention module, the edge optimization module is added, the segmentation error of the edge of the porcelain insulator is reduced, and the detection precision is improved.
Description
Technical Field
The invention relates to the technical field of transmission lines, in particular to a method and a system for measuring zero of a long string porcelain insulator unmanned plane in real time.
Background
The porcelain insulator in operation is subjected to strong electric field force and mechanical stress for a long time, and is influenced by external complex environment, the degradation of the porcelain insulator is related to the structure of the porcelain insulator, and the porcelain insulator has an undensified structure and coexists with polycrystal, so that fine gaps are unavoidable in the porcelain insulator. Under the long-term irregular effect of external factors, micropores in the porcelain piece gradually permeate to be expanded into small cracks, and then the cracks are expanded. Under the action of a strong electric field, the deteriorated porcelain insulator is extremely prone to electrical breakdown, so that a zero-value porcelain insulator is formed. The deteriorated porcelain insulator is easy to flashover in rainy and snowy days, even causes serious broken string and disconnection accidents after being struck by lightning, causes regional power failure in a large area, and seriously threatens the safe and reliable operation of a power system. The degraded porcelain insulator in the power grid is removed in time by an effective method, and the safety operation of the power grid is ensured.
The high-voltage porcelain insulator zero value detection method mainly comprises a spark gap method, an insulation resistance method, a voltage distribution method, an infrared thermal imaging method, an ultraviolet imaging method and the like. The defect detection of the porcelain insulator by using the infrared detection technology is a mature electrified detection technology, and in order to detect the deteriorated porcelain insulator, the porcelain insulator needs to be separated from a complex infrared background. At present, the basis for carrying out zero-value infrared detection of the disk-shaped suspended porcelain insulator by using an infrared thermal imaging method is national electric power industry standard DL/T2390-2021, namely a disk-shaped suspended porcelain insulator zero-value infrared detection method, wherein the judgment criterion of a low-zero-value porcelain insulator is as follows: relative temperature differenceIs a porcelain insulator with zero value and relative temperature difference +.>Is a low-value porcelain insulator. Because the pixel value of the thermal infrared imager carried by the unmanned aerial vehicle applied at present is not high, if the whole string of porcelain insulators is brought into view, the porcelain insulators occupy fewer pixels and the edges are affected by the environment, the segmented image can contain a large amount of background information of the edges of the porcelain insulators, the accuracy of infrared detection of the porcelain insulators is greatly reduced, and the missed judgment and the erroneous judgment of the infrared detection of the unmanned aerial vehicle of the porcelain insulators are caused. In order to reduce interference of background information on infrared detection, the invention provides a real-time zero-measuring method and a real-time zero-measuring system for a long-string porcelain insulator unmanned aerial vehicle.
Therefore, how to design a method and a system for real-time zero measurement of a long string porcelain insulator unmanned aerial vehicle becomes a problem which needs to be solved currently.
Disclosure of Invention
The invention aims to more efficiently identify a deteriorated porcelain insulator by an infrared thermal imaging method, and provides a long-string porcelain insulator unmanned aerial vehicle real-time zero-measuring method and a long-string porcelain insulator unmanned aerial vehicle real-time zero-measuring system, so as to solve the problem that the existing porcelain insulator is low in infrared zero-measuring accuracy.
In order to achieve the above purpose, the present invention provides the following technical solutions: a real-time zero measurement method of a long string porcelain insulator unmanned aerial vehicle comprises the following steps:
s1: acquiring an infrared image of a porcelain insulator by using an unmanned aerial vehicle-mounted thermal infrared imager, selecting the porcelain insulator segment infrared image acquired within a preset distance as a data set, and dividing the data set;
s2: constructing an improved deep LabV3+ semantic segmentation network, adopting a phantom module (Ghost module) in the GhostNet network to replace an original backbone network (Xreception) in the deep LabV3+ semantic segmentation network, introducing an SE channel attention module, adding an edge optimization module (PointRend), inputting a processed data set into the improved deep LabV3+ semantic segmentation network to train the processed data set, and obtaining an improved deep LabV3+ network zero-value insulator detection system;
s3: calculating the temperature curve of the whole string of porcelain insulators by utilizing an improved deep LabV3+ network zero-value insulator detection system, and judging whether the whole string of porcelain insulators contain zero-value insulators or not according to the characteristics of the temperature curve of the whole string of porcelain insulators;
s4: the improved deep LabV3+ network zero-value insulator detection system is embedded into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can accurately detect and judge the zero-value insulator in the whole string of porcelain insulators in real time.
Further, in the step S1, the process of dividing the data set is: dividing the data set into a training set, a testing set and a verification set according to a specific proportion, guiding the data set into an image characteristic enhancement module and a Gaussian noise reduction module for preprocessing, and manufacturing the processed image into a label.
Further, in the step S2, an edge optimization module (pointrand) is added to the output end of the deep labv3+ semantic segmentation network, the segmentation result is post-processed, the image segmentation of the object edge is optimized, and a linear correction unit function prime with parameters is adopted as an activation function:
，
wherein ,is an updatable coefficient->Is->An activation function input value for each channel;
adopting a focus loss function to replace a cross entropy loss function, wherein the focus loss function is as follows:
，
wherein ,the probability of correct prediction for the classification for the model; />Is super parameter and is used for modulating the weight of the difficult sample;is a focal point loss function.
Weighting coefficients modulating the positive and negative sample weights may also be introducedThe formula is as follows:
，
wherein ,for the weighting factor>Along with->Is suitably decreased by an increase in->The value of (2) determines the weighting between the classifications for controlling the imbalance in the number of samples.
Further, the specific implementation method of the attention module for introducing the SE channel comprises the following steps: firstly, converting a feature map, carrying out global average pooling on the feature map to generate a vector, wherein each element value corresponds to each channel, then processing the vector obtained in the previous step through two full-connection layers to obtain a channel weight value, multiplying the generated feature vector by the channel corresponding to the feature map, and giving the obtained channel weight value to the original feature map.
Furthermore, the image segmentation of the edge of the object is optimized, the idea is that the image segmentation is regarded as a rendering problem, and the method is based on iterative subdivision algorithm segmentation prediction capable of adaptively selecting points, and the algorithm steps are as follows: firstly, a rough prediction mask is generated by a lightweight prediction head, then, the rough prediction mask is up-sampled by bilinear interpolation, a prediction mask with double resolution is obtained, uncertain points, namely points distributed on the edge, are selected, point-by-point characteristics are constructed on the selected points by combining fine-grained characteristics and coarse-prediction characteristics, wherein the fine-grained characteristics are extracted from a characteristic diagram generated by a backbone network, and the coarse-prediction characteristics are derived from lightweight coarse-segmentation prediction; the points are predicted continuously by using a simple multi-layer perceptron, and finally the classification label of each point is output.
Further, the specific implementation method for calculating the temperature curve of the whole string of porcelain insulators by using the improved deep LabV3+ network zero-value insulator detection system in the step S3 is as follows: the long string porcelain insulator refers to porcelain insulator strings with more than 25 porcelain insulators, an onboard lens of a common industrial thermal imager is difficult to clearly shoot the whole infrared spectrum of the long string porcelain insulator, in order to reduce the interference of edge environment information, one part of the porcelain insulator string needs to be shot within a preset distance, a connecting fitting at one end of the porcelain insulator string is used as a starting point of detection, then the porcelain insulator string segments are continuously shot, the temperature of the porcelain insulator in the shooting segments is sequentially detected, an improved deep LabV3+ network zero value insulator detection system is closed when the connecting fitting at the other end of the porcelain insulator string is detected, and then a temperature curve splicing method of weighted data fusion is used to finally obtain the temperature curve of the whole string of porcelain insulator.
Further, the specific steps of the temperature curve splicing method for the weighted data fusion are as follows: firstly, continuously shooting a group of images of the same string of porcelain insulators in a segmented mode, then respectively generating temperature curves of partial strings according to the images, and synthesizing the temperature curves of the porcelain insulators by utilizing a curve splicing technology, wherein the operation formula is as follows:
，
，
wherein ,for the temperature value after the c-th porcelain insulator is fused,/->The temperature value before the fusion of the ith temperature curve of the c-th porcelain insulator is +.>The weighting coefficient of the temperature value before the fusion of the ith curve of the c-th porcelain insulator can be adjusted according to the definition of the pixel>The total number of the temperature curves.
A long string porcelain insulator unmanned aerial vehicle real-time zero-measuring system comprises:
the acquisition module is used for acquiring infrared images of the porcelain insulators by using the unmanned aerial vehicle-mounted thermal infrared imager, selecting the infrared images of the porcelain insulator segments acquired within a preset distance as a data set, and dividing the data set;
the construction module is used for constructing an improved deep LabV3+ semantic segmentation network, a phantom module in the GhostNet network is used for replacing an original backbone network in the deep LabV3+ semantic segmentation network, a SE channel attention module is introduced, an edge optimization module is added, a processed data set is input into the improved deep LabV3+ semantic segmentation network for training, and an improved deep LabV3+ network zero-value insulator detection system is obtained;
the detection module is used for calculating the temperature curve of the whole string of porcelain insulators by utilizing an improved deep LabV3+ network zero-value insulator detection system, and judging whether the whole string of porcelain insulators contains the zero-value insulator or not according to the temperature curve characteristics of the whole string of porcelain insulators;
and the embedded module is used for embedding the improved deep LabV3+ network zero-value insulator detection system into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can accurately detect and judge the zero-value insulator in the whole string of porcelain insulators in real time.
Compared with the prior art, the invention has the following beneficial effects: according to the invention, a phantom module (Ghost module) in the GhostNet network is adopted to replace an original backbone network (Xpercent) in the deep LabV3+ semantic segmentation network, so that the calculation time and the parameter quantity are reduced, the real-time performance of detection is improved, the weighting of the channel-level facing characteristics is realized by introducing an SE channel attention module, and an edge optimization module (PointRend) is added, so that the segmentation error of the edge of the porcelain insulator is reduced, the detection precision is improved, and the method can be carried on an unmanned plane, and the working efficiency and the accuracy of the porcelain insulator zero measurement are improved, so that the method has strong practicability and wide application prospect.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a bottleneck layer structure diagram of the present invention.
FIG. 3 is a schematic diagram of a SE channel attention module of the present invention.
FIG. 4 is a flow chart of an implementation of the join edge optimization module of the present invention.
FIG. 5 is a flow chart of a multi-segment temperature curve splicing process according to the present invention.
Fig. 6 is a schematic diagram of infrared zeroing of the porcelain insulator of the present invention.
Fig. 7 is a schematic diagram of the system configuration of the present invention.
Fig. 8 is a deep labv3+ semantic segmentation network structure diagram of the present invention.
Detailed Description
The invention will be described in further detail with reference to the drawings and the detailed description.
As shown in fig. 1, the present invention provides the following technical solutions: a real-time zero measurement method of a long string porcelain insulator unmanned aerial vehicle comprises the following steps:
s1: acquiring an infrared image of a porcelain insulator by using an unmanned aerial vehicle-mounted thermal infrared imager, selecting the porcelain insulator segment infrared image acquired at a short distance as a data set, dividing the data set into a training set, a test set and a verification set according to a specific ratio of 8:1:1, introducing the data set into an image characteristic enhancement module and a Gaussian noise reduction module for preprocessing, and manufacturing the processed image into a label;
s2: constructing an improved deep LabV3+ semantic segmentation network, replacing an original backbone network (Xattention) in the deep LabV3+ semantic segmentation network by a phantom module (Ghost module) in the GhostNet network, and introducing SE (channel attention module), wherein the implementation mode is as follows: the feature map is subjected to conversion operation, then the feature map is subjected to global average pooling to generate a vector, wherein each element value corresponds to each channel, then the vector obtained in the previous step is processed through two full-connection layers to obtain a channel weight value, finally the generated feature vector is multiplied by the channel corresponding to the feature map, the obtained channel weight value is given to the original feature map, an edge optimization module (PointRend) is added at the output end of the deep LabV & lt3+ & gt semantic segmentation network to improve segmentation precision, and the segmentation result is subjected to post-processing.
a. Firstly, a rough prediction mask is generated by a lightweight prediction head;
b. then, up-sampling the coarse prediction mask by bilinear interpolation to obtain a prediction mask with double resolution, and selecting uncertain points, namely points distributed on the edge;
c. constructing point-by-point features at selected points by combining fine-grained features extracted from feature maps generated from the backbone network with coarse-predicted features derived from lightweight, coarse-partitioned predictions;
d. the points are continuously predicted by using a simple multi-layer perceptron, and finally, the classification label of each point is output;
repeating steps a to b until the mask with the required resolution is up-sampled;
optimizing for image separation of object edges, and adopting a linear correction unit function PReLU with parameters as an activation function:
，
wherein ,is an updatable coefficient->Is->An activation function input value for each channel;
the focus loss function is used for replacing the cross entropy loss function to solve the problem that the positive and negative sample types are unbalanced during training, and the focus loss function is as follows:
，
wherein ,the probability of correct prediction for the classification for the model; />Is super parameter and is used for modulating the weight of the difficult sample;is a focal point loss function;
weighting coefficients modulating the positive and negative sample weights may also be introducedThe formula is as follows:
，
wherein ,for the weighting factor>Along with->Is suitably decreased by an increase in->The value of (2) determines the weight between the classifications for controlling the unbalance of the sample number;
inputting the processed data set into an improved deep LabV3+ semantic segmentation network for training, and carrying out semantic segmentation on the porcelain insulator infrared image by utilizing an improved deep LabV3+ network zero-value insulator detection system obtained by training.
As shown in fig. 8, the specific operation of each module in the modified deep labv3+ network null insulator detection system is as follows:
s2.1: the method is characterized in that a Ghost module (Ghost module) in a GhostNet network is adopted to replace an original backbone network (Xpercent) of a deep LabV3+ semantic segmentation network, the structure of the GhostNet network consists of a 3×3 convolution layer, a plurality of bottleneck layers (Ghost), a 1×1 convolution layer, an average pooling layer and a full-connection layer, wherein the bottleneck layers (Ghost) are similar to residual blocks of ResNet in structure, the input features and the output features are subjected to feature superposition through residual errors to achieve the effect of feature learning, but the bottleneck layers (Ghost) are different from the basic structure of ResNet in that the Ghost modules are utilized to replace the convolution layers in residual errors, the channel number of the feature images is expanded through a first Ghost module (Ghost module), the channel number of the feature images is reduced through a second Ghost module (Ghost module), the structure is used for reducing the channel number of the feature images between the first Ghost module (Ghost module) and the second Ghost module (ghut module) and the input features are connected with the input features (tcut module) in order to reduce the depth of the input features, and the depth of the input features is reduced.
S3: as shown in fig. 5-6: the long string porcelain insulator refers to porcelain insulator strings with more than 25 porcelain insulators and more than 25 porcelain insulators, an onboard lens of a common industrial thermal imager is difficult to clearly shoot the whole infrared spectrum of the long string porcelain insulator, in order to reduce the interference of edge environment information, a part of the porcelain insulator string needs to be shot at a short distance, a field shot infrared image comprises 5-8 porcelain insulator segments, a first connecting fitting S at one end of the porcelain insulator string is used as a starting point of detection, aiming at the segmented continuous porcelain insulator segment infrared image, a 1 st picture can obtain a temperature curve segment of an I1-I5 porcelain insulator, a 2 nd picture can obtain a temperature curve segment of the I2-I7 porcelain insulator (wherein I1 is the first piece of insulator, i2 is a second piece of insulator, I3 is a third piece of insulator, I4 is a fourth piece of insulator, I5 is a fifth piece of insulator, I6 is a sixth piece of insulator, I7 is a seventh piece of insulator, I8 is an eighth piece of insulator … I25 and is a second fifteen pieces of insulator), temperature curves of all the pieces of porcelain insulators are sequentially detected according to the method, a modified deep labv3+ network zero value insulator detection system is closed when a second connecting fitting E at the other end of a porcelain insulator string is detected, a temperature curve of the whole string of porcelain insulators is finally obtained through a temperature curve splicing method of weighted data fusion, and the specific steps for obtaining the temperature curve of the whole string of porcelain insulators are as follows: firstly, continuously shooting a group of images of the same string of porcelain insulators in a segmented mode, then respectively generating temperature curves of partial strings according to the images, and synthesizing the temperature curves of the porcelain insulators by utilizing a curve splicing technology, wherein the operation formula is as follows:
，
，
wherein ,for the temperature value after the c-th porcelain insulator is fused,/->The temperature value before the fusion of the ith temperature curve of the c-th porcelain insulator is +.>The weighting coefficient of the temperature value before the fusion of the ith curve of the c-th porcelain insulator can be adjusted according to the definition of the pixel>The total number of the temperature curves is;
judging whether the porcelain insulator string contains a zero-value insulator or not according to the temperature curve characteristics of the porcelain insulator string;
s4: the improved deep LabV3+ network zero-value insulator detection system is embedded into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can accurately detect the temperature curve of the porcelain insulator string in real time, and the unmanned aerial vehicle can detect the porcelain insulator string with higher accuracy in infrared zero detection.
As shown in fig. 7, a long string porcelain insulator unmanned aerial vehicle real-time zero measurement system includes:
the acquisition module is used for acquiring an infrared image of the porcelain insulator by using an unmanned aerial vehicle-mounted thermal infrared imager, selecting the porcelain insulator segment infrared image acquired in a short distance as a data set, and dividing the data set;
the construction module is used for constructing an improved deep LabV3+ semantic segmentation network, a phantom module in the GhostNet network is used for replacing an original backbone network in the deep LabV3+ semantic segmentation network, a SE channel attention module is introduced, an edge optimization module is added, a processed data set is input into the improved deep LabV3+ semantic segmentation network for training, and an improved deep LabV3+ network zero-value insulator detection system is obtained;
the detection module is used for calculating the temperature curve of the whole string of porcelain insulators by utilizing an improved deep LabV3+ network zero-value insulator detection system, and judging whether the whole string of porcelain insulators contains the zero-value insulator or not according to the temperature curve characteristics of the whole string of porcelain insulators;
and the embedded module is used for embedding the improved deep LabV3+ network zero-value insulator detection system into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can accurately detect and judge the zero-value insulator in the whole string of porcelain insulators in real time.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. The real-time zero measurement method of the long porcelain insulator unmanned aerial vehicle is characterized by comprising the following steps of:
s1: acquiring an infrared image of a porcelain insulator by using an unmanned aerial vehicle-mounted thermal infrared imager, selecting the porcelain insulator segment infrared image acquired within a preset distance as a data set, and dividing the data set;
s2: constructing an improved deep LabV3+ semantic segmentation network, adopting a phantom module in a GhostNet network to replace an original backbone network in the deep LabV3+ semantic segmentation network, introducing an SE channel attention module, adding an edge optimization module, inputting a processed data set into the improved deep LabV3+ semantic segmentation network for training, and obtaining an improved deep LabV3+ network zero-value insulator detection system;
s3: calculating the temperature curve of the whole string of porcelain insulators by utilizing an improved deep LabV3+ network zero-value insulator detection system, and judging whether the whole string of porcelain insulators contain zero-value insulators or not according to the characteristics of the temperature curve of the whole string of porcelain insulators;
s4: the improved deep LabV3+ network zero-value insulator detection system is embedded into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can detect and judge the zero-value insulator in the whole string of porcelain insulators in real time.
2. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 1, wherein the method comprises the following steps: the process of dividing the data set in the step S1 is as follows: dividing the data set into a training set, a testing set and a verification set according to a specific proportion, guiding the data set into an image characteristic enhancement module and a Gaussian noise reduction module for preprocessing, and manufacturing the processed image into a label.
3. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 1, wherein the method comprises the following steps: in the step S2, an edge optimization module is added at the output end of the deep labv3+ semantic segmentation network, the segmentation result is post-processed, the image segmentation of the object edge is optimized, a linear correction unit function pralu with parameters is adopted as an activation function, and a focus loss function is adopted to replace a cross entropy loss function.
4. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 3, wherein the method comprises the following steps of: the specific implementation method of the module for introducing the SE channel attention comprises the following steps: firstly, converting a feature map, carrying out global average pooling on the feature map to generate a vector, wherein each element value corresponds to each channel, then processing the vector obtained in the previous step through two full-connection layers to obtain a channel weight value, multiplying the generated feature vector by the channel corresponding to the feature map, and giving the obtained channel weight value to the original feature map.
5. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 3, wherein the method comprises the following steps of: the method for optimizing the image segmentation of the object edge comprises the following steps: firstly, a rough prediction mask is generated by a lightweight prediction head, then, the rough prediction mask is up-sampled by bilinear interpolation, a prediction mask with double resolution is obtained, uncertain points, namely points distributed on the edge, are selected, point-by-point characteristics are constructed on the selected points by combining fine-grained characteristics and coarse-prediction characteristics, wherein the fine-grained characteristics are extracted from a characteristic diagram generated by a backbone network, and the coarse-prediction characteristics are derived from lightweight coarse-segmentation prediction; the points are predicted continuously by using a simple multi-layer perceptron, and finally the classification label of each point is output.
6. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 5, wherein the method comprises the following steps: in the step S3, the concrete implementation method for calculating the temperature curve of the whole string of porcelain insulators by using the improved deep LabV3+ network zero-value insulator detection system comprises the following steps: and taking a connecting fitting at one end of the porcelain insulator string as a starting point of detection, continuously shooting the porcelain insulator string segments, sequentially detecting the temperature of the porcelain insulator in the shot segments, closing the improved deep LabV3+ network zero-value insulator detection system when the connecting fitting at the other end of the porcelain insulator string is detected, and obtaining the temperature curve of the whole string of porcelain insulators by using a temperature curve splicing method of weighted data fusion.
7. The method for real-time zero measurement of the long porcelain insulator unmanned aerial vehicle according to claim 6, wherein the method comprises the following steps: the temperature curve splicing method for the weighted data fusion comprises the following specific steps: firstly, continuously shooting a group of images of the same string of porcelain insulators in a segmented mode, then respectively generating temperature curves of partial strings according to the images, and synthesizing the temperature curves of the porcelain insulators by utilizing a curve splicing technology, wherein the operation formula is as follows:
，
，
wherein ,for the temperature value after the c-th porcelain insulator is fused,/->The temperature value before the fusion of the ith temperature curve of the c-th porcelain insulator is +.>Weighting coefficient of temperature value before fusing ith curve of c-th porcelain insulator, +.>The total number of the temperature curves.
8. A long string porcelain insulator unmanned aerial vehicle real-time zero-measuring system which is characterized by comprising:
the acquisition module is used for acquiring infrared images of the porcelain insulators by using the unmanned aerial vehicle-mounted thermal infrared imager, selecting the infrared images of the porcelain insulator segments acquired within a preset distance as a data set, and dividing the data set;
the construction module is used for constructing an improved deep LabV3+ semantic segmentation network, a phantom module in the GhostNet network is used for replacing an original backbone network in the deep LabV3+ semantic segmentation network, a SE channel attention module is introduced, an edge optimization module is added, a processed data set is input into the improved deep LabV3+ semantic segmentation network for training, and an improved deep LabV3+ network zero-value insulator detection system is obtained;
the detection module is used for calculating the temperature curve of the whole string of porcelain insulators by utilizing an improved deep LabV3+ network zero-value insulator detection system, and judging whether the whole string of porcelain insulators contains the zero-value insulator or not according to the temperature curve characteristics of the whole string of porcelain insulators;
and the embedded module is used for embedding the improved deep LabV3+ network zero-value insulator detection system into the unmanned aerial vehicle system, so that the unmanned aerial vehicle can detect and judge the zero-value insulator in the whole string of porcelain insulators in real time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310850373.7A CN116596920B (en) | 2023-07-12 | 2023-07-12 | Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310850373.7A CN116596920B (en) | 2023-07-12 | 2023-07-12 | Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116596920A true CN116596920A (en) | 2023-08-15 |
| CN116596920B CN116596920B (en) | 2023-11-07 |
Family
ID=87604756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310850373.7A Active CN116596920B (en) | 2023-07-12 | 2023-07-12 | Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116596920B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117670882A (en) * | 2024-01-31 | 2024-03-08 | 国网江西省电力有限公司电力科学研究院 | Unmanned aerial vehicle infrared automatic focusing method and system for porcelain insulator string |
Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007098659A1 (en) * | 2006-03-03 | 2007-09-07 | Dongguan Gaoneng Industrial Co., Ltd | A synthetic insulator for an ultrahigh voltage ac transmitting line |
| CN108181556A (en) * | 2017-12-18 | 2018-06-19 | 国网浙江省电力有限公司检修分公司 | Porcelain insulator zero value detection method based on chapeau de fer temperature difference time series analysis |
| CN109142991A (en) * | 2018-07-05 | 2019-01-04 | 国网湖南省电力有限公司电力科学研究院 | A kind of infrared survey zero-temperature coefficient threshold determination method of porcelain insulator based on Burr distribution |
| CN112037137A (en) * | 2020-07-21 | 2020-12-04 | 国网湖北省电力有限公司电力科学研究院 | Method and device for eliminating fuzzy region of insulator disc surface edge in infrared image |
| CN112699897A (en) * | 2020-12-04 | 2021-04-23 | 国网湖北省电力有限公司电力科学研究院 | Method and system for splicing temperature curves of string iron caps of ultra-high voltage porcelain insulators |
| WO2021190056A1 (en) * | 2020-03-26 | 2021-09-30 | 国网湖北省电力有限公司电力科学研究院 | Infrared zero value diagnosis method and system for porcelain insulator string |
| CN113887517A (en) * | 2021-10-29 | 2022-01-04 | 桂林电子科技大学 | Crop remote sensing image semantic segmentation method based on parallel attention mechanism |
| CN114549563A (en) * | 2022-02-26 | 2022-05-27 | 福建工程学院 | Real-time composite insulator segmentation method and system based on deep LabV3+ |
| CN114612477A (en) * | 2022-03-03 | 2022-06-10 | 成都信息工程大学 | Lightweight image segmentation method, system, medium, terminal and application |
| CN114663759A (en) * | 2022-03-24 | 2022-06-24 | 东南大学 | Remote sensing image building extraction method based on improved deep LabV3+ |
| CN114724042A (en) * | 2022-06-09 | 2022-07-08 | 国网江西省电力有限公司电力科学研究院 | Automatic detection method for zero-value insulator in power transmission line |
| CN114966332A (en) * | 2022-04-27 | 2022-08-30 | 湖南湖大华龙电气与信息技术有限公司 | Degradation judgment method for disc-shaped suspension type porcelain insulator string based on temperature curve similarity |
| CN115588021A (en) * | 2022-11-04 | 2023-01-10 | 国网湖北省电力有限公司荆州供电公司 | Ceramic insulator and hardware unmanned aerial vehicle infrared inspection image fusion and segmentation method |
| CN115661645A (en) * | 2022-10-21 | 2023-01-31 | 国网江西省电力有限公司电力科学研究院 | Power transmission line icing thickness prediction method based on improved Unet network |
| CN115719475A (en) * | 2022-10-24 | 2023-02-28 | 北京交通大学 | Three-stage trackside equipment fault automatic detection method based on deep learning |
| CN115984226A (en) * | 2023-01-09 | 2023-04-18 | 长沙理工大学 | Insulator defect detection method, device, medium, and program product |
| CN115994892A (en) * | 2022-11-28 | 2023-04-21 | 淮阴工学院 | Lightweight medical image segmentation method and system based on ghostnet |
| CN116342391A (en) * | 2023-03-22 | 2023-06-27 | 国网浙江省电力有限公司超高压分公司 | Infrared image side unfolding method and device for pillar porcelain insulator |
| CN116363358A (en) * | 2023-01-11 | 2023-06-30 | 河南大学 | Road scene image real-time semantic segmentation method based on improved U-Net |
-
2023
- 2023-07-12 CN CN202310850373.7A patent/CN116596920B/en active Active
Patent Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007098659A1 (en) * | 2006-03-03 | 2007-09-07 | Dongguan Gaoneng Industrial Co., Ltd | A synthetic insulator for an ultrahigh voltage ac transmitting line |
| CN108181556A (en) * | 2017-12-18 | 2018-06-19 | 国网浙江省电力有限公司检修分公司 | Porcelain insulator zero value detection method based on chapeau de fer temperature difference time series analysis |
| CN109142991A (en) * | 2018-07-05 | 2019-01-04 | 国网湖南省电力有限公司电力科学研究院 | A kind of infrared survey zero-temperature coefficient threshold determination method of porcelain insulator based on Burr distribution |
| WO2021190056A1 (en) * | 2020-03-26 | 2021-09-30 | 国网湖北省电力有限公司电力科学研究院 | Infrared zero value diagnosis method and system for porcelain insulator string |
| CN112037137A (en) * | 2020-07-21 | 2020-12-04 | 国网湖北省电力有限公司电力科学研究院 | Method and device for eliminating fuzzy region of insulator disc surface edge in infrared image |
| CN112699897A (en) * | 2020-12-04 | 2021-04-23 | 国网湖北省电力有限公司电力科学研究院 | Method and system for splicing temperature curves of string iron caps of ultra-high voltage porcelain insulators |
| CN113887517A (en) * | 2021-10-29 | 2022-01-04 | 桂林电子科技大学 | Crop remote sensing image semantic segmentation method based on parallel attention mechanism |
| CN114549563A (en) * | 2022-02-26 | 2022-05-27 | 福建工程学院 | Real-time composite insulator segmentation method and system based on deep LabV3+ |
| CN114612477A (en) * | 2022-03-03 | 2022-06-10 | 成都信息工程大学 | Lightweight image segmentation method, system, medium, terminal and application |
| CN114663759A (en) * | 2022-03-24 | 2022-06-24 | 东南大学 | Remote sensing image building extraction method based on improved deep LabV3+ |
| CN114966332A (en) * | 2022-04-27 | 2022-08-30 | 湖南湖大华龙电气与信息技术有限公司 | Degradation judgment method for disc-shaped suspension type porcelain insulator string based on temperature curve similarity |
| CN114724042A (en) * | 2022-06-09 | 2022-07-08 | 国网江西省电力有限公司电力科学研究院 | Automatic detection method for zero-value insulator in power transmission line |
| CN115661645A (en) * | 2022-10-21 | 2023-01-31 | 国网江西省电力有限公司电力科学研究院 | Power transmission line icing thickness prediction method based on improved Unet network |
| CN115719475A (en) * | 2022-10-24 | 2023-02-28 | 北京交通大学 | Three-stage trackside equipment fault automatic detection method based on deep learning |
| CN115588021A (en) * | 2022-11-04 | 2023-01-10 | 国网湖北省电力有限公司荆州供电公司 | Ceramic insulator and hardware unmanned aerial vehicle infrared inspection image fusion and segmentation method |
| CN115994892A (en) * | 2022-11-28 | 2023-04-21 | 淮阴工学院 | Lightweight medical image segmentation method and system based on ghostnet |
| CN115984226A (en) * | 2023-01-09 | 2023-04-18 | 长沙理工大学 | Insulator defect detection method, device, medium, and program product |
| CN116363358A (en) * | 2023-01-11 | 2023-06-30 | 河南大学 | Road scene image real-time semantic segmentation method based on improved U-Net |
| CN116342391A (en) * | 2023-03-22 | 2023-06-27 | 国网浙江省电力有限公司超高压分公司 | Infrared image side unfolding method and device for pillar porcelain insulator |
Non-Patent Citations (4)
| Title |
|---|
| LAXMAN SINGH等: ""Design of thermal imaging-based health condition monitoring and early fault detection technique for porcelain insulators using Machine learning"", 《ENVIRONMENTAL TECHNOLOGY & INNOVATION》, vol. 24 * |
| 周学明等: ""超—特高压长串瓷绝缘子温度分布曲线拼接方法"", 《电力科学与技术学报》, vol. 37, no. 03 * |
| 宁奉阁等: ""基于S-ASPP和双注意力机制的磁瓦外观缺陷检测算法"", 《电子测量技术》, vol. 46, no. 02 * |
| 张彦;姚建刚;毛田;邹涛;龚磊;: "WebGIS红外热像零值绝缘子在线检测***", 电力***及其自动化学报, no. 04 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117670882A (en) * | 2024-01-31 | 2024-03-08 | 国网江西省电力有限公司电力科学研究院 | Unmanned aerial vehicle infrared automatic focusing method and system for porcelain insulator string |
| CN117670882B (en) * | 2024-01-31 | 2024-06-04 | 国网江西省电力有限公司电力科学研究院 | Unmanned aerial vehicle infrared automatic focusing method and system for porcelain insulator string |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116596920B (en) | 2023-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108647655B (en) | Low-altitude aerial image power line foreign matter detection method based on light convolutional neural network | |
| Sohn et al. | Automatic powerline scene classification and reconstruction using airborne lidar data | |
| CN112184654A (en) | High-voltage line insulator defect detection method based on generation countermeasure network | |
| CN113486865B (en) | Power transmission line suspended foreign object target detection method based on deep learning | |
| CN116596920B (en) | Real-time zero measurement method and system for long-string porcelain insulator unmanned aerial vehicle | |
| CN114943963B (en) | Remote sensing image cloud and cloud shadow segmentation method based on double-branch fusion network | |
| CN116468730B (en) | Aerial Insulator Image Defect Detection Method Based on YOLOv5 Algorithm | |
| CN111428862B (en) | Polar unbalanced space-time combined convection primary short-term prediction method | |
| CN111553204B (en) | Transmission tower detection method based on remote sensing image | |
| CN110705397A (en) | YOLOv3 pruning identification insulator defect method suitable for small field sample amount | |
| Li et al. | Insulator and damage detection and location based on YOLOv5 | |
| CN115661655B (en) | Southwest mountain area cultivated land extraction method with hyperspectral and hyperspectral image depth feature fusion | |
| CN115588021A (en) | Ceramic insulator and hardware unmanned aerial vehicle infrared inspection image fusion and segmentation method | |
| CN115239710A (en) | Insulator defect detection method based on attention feedback and double-space pyramid | |
| CN116703885A (en) | Swin transducer-based surface defect detection method and system | |
| CN109919936B (en) | Method, device and equipment for analyzing running state of composite insulator | |
| CN114241310A (en) | Improved YOLO model-based intelligent identification method for piping dangerous case of dike | |
| CN113450318A (en) | Primary power distribution system porcelain insulator detection method based on unmanned aerial vehicle autonomous vision | |
| CN114140446A (en) | Insulator defect detection method based on semantic segmentation and target detection | |
| CN114612803A (en) | Transmission line insulator defect detection method for improving CenterNet | |
| CN114359167A (en) | Insulator defect detection method based on lightweight YOLOv4 in complex scene | |
| CN117292277A (en) | Insulator fault detection method based on binocular unmanned aerial vehicle system and deep learning | |
| CN115631405A (en) | SegFormer-based SAR image ocean inner wave stripe segmentation method | |
| CN117949794B (en) | Cable partial discharge fault detection method | |
| CN117557775B (en) | Substation power equipment detection method and system based on infrared and visible light fusion |
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 |