CN110378221A - A kind of power grid wire clamp detects and defect identification method and device automatically - Google Patents
A kind of power grid wire clamp detects and defect identification method and device automatically Download PDFInfo
- Publication number
- CN110378221A CN110378221A CN201910513439.7A CN201910513439A CN110378221A CN 110378221 A CN110378221 A CN 110378221A CN 201910513439 A CN201910513439 A CN 201910513439A CN 110378221 A CN110378221 A CN 110378221A
- Authority
- CN
- China
- Prior art keywords
- wire clamp
- picture
- classification
- trained
- power grid
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000007547 defect Effects 0.000 title claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 37
- 238000001514 detection method Methods 0.000 claims abstract description 34
- 230000005540 biological transmission Effects 0.000 claims abstract description 22
- 230000003416 augmentation Effects 0.000 claims abstract description 20
- 230000005611 electricity Effects 0.000 claims abstract description 13
- 238000007689 inspection Methods 0.000 abstract description 7
- 238000012986 modification Methods 0.000 abstract description 7
- 230000004048 modification Effects 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000013527 convolutional neural network Methods 0.000 description 30
- 238000010586 diagram Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000000605 extraction Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 230000004913 activation Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000032696 parturition Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—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
- G06V2201/07—Target detection
Abstract
It is detected automatically the invention discloses a kind of power grid wire clamp and defect identification method, comprising the following steps: pass through image capture module and obtain wire clamp picture；Wire clamp picture is trained and obtains wire clamp type；Classification is carried out to wire clamp picture and deletes choosing；Failure wire clamp is identified and exported.It is detected automatically invention additionally discloses a kind of power grid wire clamp and defect recognizing device.The present invention carries out Image Acquisition to the wire clamp on transmission line of electricity by image capture module, wire clamp picture is trained by using Faster R-CNN technology, using augmentation training technique, it is long to solve the time caused by needing re -training for new wire clamp feature, model modification period slow defect, fault detection to wire clamp, dramatically saves the cost of overhaul, keeps inspection system highly efficient and intelligent.
Description
Technical field
The present invention relates to overhead transmission line security technology area more particularly to a kind of power grid wire clamp detects automatically and defect
Recognition methods and device.
Background technique
In overhead transmission lines, various power transmission lines are indenting, may mutually wind between power transmission line, pull, shake
Dynamic, these are all easy to cause strand breakage of circuit, seriously threaten the safe and reliable operation of overhead transmission line.Wire clamp is overhead power transmission
Important devices in route can wrap up a large amount of transmission lines of electricity, play the role of protection and auxiliary vibration damping, pole to transmission line of electricity
The earth extends the service life of transmission line of electricity.But overhead transmission line is transmitted for remote distance power, it during which will necessarily be across
More different natural environments.After undergoing long-time exposing to the weather, wire clamp can generate various defects, cause wire clamp that cannot play original
This effect.So the defects detection of wire clamp is the important process of inspection during the inspection of overhead transmission line.It is more important
: wire clamp it is many kinds of, it would be desirable to distinguish a large amount of different types of wire clamps.In the inspection of current overhead transmission line
In the process, manual inspection is very time-consuming, and there are certain risk for the work of patrol officer.
In recent years, as the unmanned plane of the fast development of unmanned air vehicle technique and efficient stable is in the application of other field.
Domestic many Utilities Electric Co.s start to assist inspection to work using unmanned plane successively.The picture shot by unmanned plane obtains line
Picture is pressed from both sides, defect recognition is carried out by the detection to wire clamp picture.In current object detection field, convolutional neural networks are answered
With making the performance of target detection have one greatly to leap, Faster R-CNN belongs to two stages algorithm of target detection, the
One step extracts proposal；Second step, by classifying and returning completion target detection, classification then marks interesting target
Classification, recurrence are then the bounding box(bounding boxes to object) make amendment.The detection that YOLO and SSD belongs to single phase is calculated
Method directly realizes target detection by classification and recurrence；Faster R-CNN belongs to two stages algorithm of target detection, may be implemented
Reach precision more higher than single phase detection algorithm.This method is not only from the color of target and texture, and from the more face of various features
Chrominance channel can filter a part of noise and environmental disturbances to Objective extraction feature, by convolution, keep accurately discrimination,
It is significant that this is still able to maintain higher discrimination in the case where background complexity for wire clamp.
Wire clamp is many kinds of, often will appear new power grid wire clamp after power grid wire clamp detection model on-line running because of early period
There is no a training data and whole missing inspections, goes to train Faster R-CNN model using a large amount of wire clamp data, to enable model
Convergence achievees the purpose that detect terminal clamp and makes classification to it so as to fit the wire clamp in picture to be detected.So
And in practical applications, the wire clamp of a large amount of different characteristic, such as appearance, texture are difficult once to have collected, then
There is the wire clamp of new feature, current solution is that the data being now collected into and the data being collected into before are united
Training is re-started, such solution just has a defect, with the addition for the data newly collected, so that overall data is got over
Come more, model training is slower and slower, and the model modification period is increasingly longer, has seriously affected working efficiency.
Summary of the invention
For above-mentioned prior art Shortcomings, the present invention provides a kind of power grid wire clamp and detects automatically and defect identification method
And device, it is directed to newly-increased classification and augmentation training is done to full articulamentum, to realize the quick of power grid wire clamp target detection model
The automatic identification of update and power grid wire clamp defect.
The technical solution adopted by the present invention are as follows:
A kind of power grid wire clamp detects automatically and defect identification method, comprising the following steps:
Wire clamp picture is obtained by image capture module；
Wire clamp picture is trained and obtains wire clamp type；
Classification is carried out to wire clamp picture and deletes choosing；
Failure wire clamp is identified and exported.
As further technical solution of the present invention are as follows: it is described that wire clamp picture is obtained by image capture module, it is specific to wrap
Include: the style of shooting combined using unmanned plane and photographic device shoots the wire clamp on transmission line of electricity to obtain wire clamp figure
Piece.
As further technical solution of the present invention are as follows: described be trained to wire clamp picture obtains wire clamp classification；Specifically
Include:
A certain number of wire clamp pictures are chosen as training sample；
Terminal clamp feature, which is extracted, using the convolution kernel in Fast R-CNN is trained acquisition wire clamp classification.
As further technical solution of the present invention are as follows: described be trained to wire clamp picture obtains wire clamp classification；Also wrap
It includes: augmentation training being carried out for non-classified wire clamp picture, obtains wire clamp augmentation type.
Further, the wire clamp feature includes wire clamp color, wire clamp profile, wire clamp texture.
As further technical solution of the present invention are as follows: it is described to wire clamp picture carry out classification delete choosing；It specifically includes:
Propose that network generates on wire clamp picture by region and surrounds frame；
Whether judge to surround in frame is target wire clamp；
If surrounding is target wire clamp in frame, point that its classification completes wire clamp picture is identified by the classifier in Fast R-CNN
Class.
As further technical solution of the present invention are as follows: failure wire clamp is identified and exported；It specifically includes:
Wire clamp fault point samples pictures are trained, wire clamp fault category is obtained；
By extracting the encirclement frame of target wire clamp based on the convolution kernel in Fast R-CNN,
Fault location is carried out to the target wire clamp surrounded in frame, identifies fault type；
Export abort situation and fault type.
The present invention also proposes that a kind of power grid wire clamp detects and defect recognizing device automatically, comprising:
Image capture module, for obtaining wire clamp picture；
Wire clamp type training module obtains wire clamp type for being trained to wire clamp picture；
Wire clamp categorization module deletes choosing for carrying out classification to wire clamp picture；
Wire clamp fault detection module, for failure wire clamp to be identified and exported.
The invention has the benefit that
The present invention carries out Image Acquisition to the wire clamp on transmission line of electricity by image capture module, by using Faster R-
CNN technology is trained wire clamp picture, obtains wire clamp type, realizes the classification to wire clamp type, while by fault wire
Folder is trained identification, realizes the fault identification to the wire clamp device of transmission line of electricity and orients fault point while detecting and is out of order
Type.The present invention uses augmentation training technique, special for subsequent increased wire clamp by being trained to a part of wire clamp picture
Sign type can do augmentation training by full articulamentum, solve the time caused by needing re -training for new wire clamp feature
Long, model modification period slow defect improves detection recognition efficiency.
Detailed description of the invention
Fig. 1 is that a kind of power grid wire clamp proposed by the present invention detects and defect identification method flow chart automatically；
Fig. 2 is proposed by the present invention described to be identified to failure wire clamp and export flow chart；
Fig. 3 is Faster R-CNN work flow diagram proposed by the present invention；
Fig. 4 is Faster R-CNN network structure proposed by the present invention；
Fig. 5 is that a kind of power grid wire clamp proposed by the present invention detects and defect recognizing device structure chart automatically.
Specific embodiment
The present invention provides that a kind of power grid wire clamp detects automatically and new grid line occur in defect identification method, this method research
Press from both sides model augmentation training technique when target category.Faster R-CNN be using convolutional layer and pond layer as the early period of model and
Mid-term structure can broadly be referred to as characteristic extraction part, and what this part finally generated is the feature of a higher dimensional matrix form
Figure, does not exceed the four-dimension generally, is finally sent into full articulamentum and classifies, be directed to newly-increased classification and do augmentation instruction to full articulamentum
Practice, realizes the quick update and the automatic identification of power grid wire clamp defect of power grid wire clamp target detection model.
Target detection refers to position and the size for identifying interesting target in image and determining target；Classification is
Image is marked in the content for being mainly based upon image, it will usually have the label of one group of fixation, the model of foundation must be predicted
It is most suitable for the label of image out.
Training pattern be exactly using largely marked what be wire clamp image data go to allow network " association " what
It is the position of wire clamp and wire clamp.That is: data (picture marked) is inputted into neural network (each neuron elder generation input value
Weighted accumulation inputs output valve of the activation primitive as the neuron again) forward-propagating, obtain score；By " score " entrance loss
Function (regularization punishment, prevent overfitting) obtains error compared with expected value, multiple to be then and known by error judgment
Other degree (penalty values are the smaller the better)；Pass through backpropagation (each activation primitive in reversed derivation, loss function and neural network
Require, final purpose is to keep error minimum) determine gradient vector；Each weight is adjusted finally by gradient vector,
Error is set to tend to 0 or the adjusting of convergent trend to " score "；It repeats the above process until setting number or damages being averaged for error mistake
Value no longer declines (minimum point).
Full articulamentum (fully connected layers, FC) plays " classifier " in entire convolutional neural networks
Effect.If the operations such as convolutional layer, pond layer and activation primitive layer are if initial data to be mapped to hidden layer feature space,
" the distributed nature expression " that full articulamentum then plays the role of to acquire is mapped to sample labeling space.In CNN, Quan Lian
Connect often appear in it is last several layers of, for doing weighted sum to the feature that front is designed.Such as Faster R-CNN, the convolution sum of front
Pond, which is equivalent to, does Feature Engineering, and subsequent full connection, which is equivalent to, does characteristic weighing.
Technical solution general thought provided by the invention is as follows:
The present invention provides a kind of power grid wire clamp and detects automatically and defect identification method, model when by power grid wire clamp target category
Augmentation training technique will not train it after being collected into new data together with data before, and refer to new using these
The wire clamp that the data of collection are directed to newly-increased different characteristic does augmentation training, the model modification greatly shortened in this way to full articulamentum
Period.Power grid wire clamp is improved to detect automatically and the accuracy of defect recognition.
It is the core concept of the application above, in order to make those skilled in the art more fully understand application scheme, under
Face is described in further detail the application in conjunction with attached drawing.It should be understood that the specific spy in the embodiment of the present application and embodiment
Sign is the detailed description to technical scheme, rather than the restriction to technical scheme, the case where not conflicting
Under, the technical characteristic in the embodiment of the present application and embodiment can be combined with each other.
Embodiment one
As shown in Figure 1, being that a kind of power grid wire clamp proposed by the present invention detects and defect identification method flow chart automatically.
Referring to Fig.1, a kind of power grid wire clamp detects and defect identification method automatically, comprising the following steps:
Step 101, wire clamp picture is obtained by image capture module；
Step 102, wire clamp picture is trained and obtains wire clamp type；
Step 103, classification is carried out to wire clamp picture and deletes choosing；
Step 104, failure wire clamp is identified and is exported.
In the embodiment of the present invention, Image Acquisition is carried out to the wire clamp on transmission line of electricity by image capture module, by adopting
Wire clamp picture is trained with Faster R-CNN technology, obtains wire clamp type, realizes the classification to wire clamp type, simultaneously
By being trained identification to failure wire clamp, realizes the fault identification to the wire clamp device of transmission line of electricity and to orient fault point same
When detect fault type.The present invention uses augmentation training technique, by being trained to a part of wire clamp picture, for subsequent
Increased wire clamp characteristic type can do augmentation training by full articulamentum, solve and new wire clamp feature is needed to instruct again
Time caused by white silk is long, model modification period slow defect, improves detection recognition efficiency.
In a step 101, described that wire clamp picture is obtained by image capture module, it specifically includes: using unmanned plane and taking the photograph
As the style of shooting that device combines, the wire clamp on transmission line of electricity is shot to obtain wire clamp picture.
Specifically, shooting under different background by unmanned plane, the transmission line of electricity picture under different angle, increase the more of sample
Sample, to preferably improve the generalization ability of model.The picture of shooting is screened, the figure for not containing wire clamp is deleted
Piece.
In a step 102, described be trained to wire clamp picture obtains wire clamp classification；It specifically includes:
A certain number of wire clamp pictures are chosen as training sample；
Terminal clamp feature, which is extracted, using the convolution kernel in Fast R-CNN is trained acquisition wire clamp classification.
Processing training is carried out to the wire clamp picture of acquisition first, certain amount refers to the wire clamp picture obtained for the first time, these lines
Picture is pressed from both sides as training sample, wire clamp feature is obtained by the processing to wire clamp picture, so that it is determined that wire clamp type, for subsequent
The wire clamp picture obtained is shot, undetermined wire clamp type can be increased by the training of subsequent augmentation, not for certain amount
It is limited, can be determined according to wire clamp picture is obtained for the first time, a portion can also be chosen as training sample, specifically with reality
Subject to the design of border, optionally, mentioned using the wire clamp in parts of images as training sample and using the convolution kernel in Fast R-CNN
Wire clamp color, profile, Texture eigenvalue is taken out to be trained as feature vector.
Wherein, described be trained to wire clamp picture obtains wire clamp classification；Further include: for non-classified wire clamp picture into
The training of row augmentation, obtains wire clamp augmentation type.
When encountering the wire clamp image that do not collected, do not need that this hair can be used by all images all re -training one time
The training method of the bright model augmentation training based on full articulamentum augmentation developed.Specific method is exactly to the data newly collected
Augmentation training is done to full articulamentum for newly-increased classification, that is, carries out model fine tuning on former trained model.
In the embodiment of the present invention, it is described to wire clamp picture carry out classification delete choosing；It specifically includes:
Propose that network generates on wire clamp picture by region and surrounds frame；
Whether judge to surround in frame is target wire clamp；
If surrounding is target wire clamp in frame, point that its classification completes wire clamp picture is identified by the classifier in Fast R-CNN
Class.
When detection, proposing that network generates on the image by region may be the encirclement frame of target, and the model after training can
Frames are surrounded to these to determine, detect whether that the image of various types of wire clamp can be effectively applicable to for target wire clamp
Detection can position picture position where terminal clamp, and identify its classification by the classifier in Fast R-CNN.
Referring to fig. 2, described failure wire clamp is identified and exports flow chart to be proposed by the present invention；
As shown in Fig. 2, failure wire clamp is identified and exported, specifically include:
Step 141, wire clamp fault point samples pictures are trained, obtain wire clamp fault category；
Step 142, by extracting the encirclement frame of target wire clamp based on the convolution kernel in Fast R-CNN；
Step 143, fault location is carried out to the target wire clamp surrounded in frame, identifies fault type；
Step 144, abort situation and fault type are exported.
For wire clamp fault location: the training process of wire clamp failure and detection are classified with wire clamp, due to giving birth to when wire clamp
When failure of becoming rusty or other failures, feature has differences with other wire clamps, therefore by based on the convolution kernel in Fast R-CNN
Extract its profile, Texture eigenvalue can detect that wire clamp falls to go here and there guilty culprit and surrounds frame, and orient abort situation.
Wire clamp fault detection based on Faster R-CNN is divided into two detection process: wire clamp classification and Detection and wire clamp failure
Positioning.Work flow diagram has the photo of most not as shown in Figure 1, for the magnanimity photo that unmanned plane transmission line of electricity is taken photo by plane
Wire clamp failure can be used to detect, therefore by wire clamp classification and Detection, a little magnanimity photos are subjected to intelligent screening and are classified, under
One step carries out fault detection and lays the foundation.The training positive sample of wire clamp classification and Detection module is various types of other wire clamp, in complexity
Energy conservation reaches identification in background.For the photo that classifying screen is selected, wire clamp fault detection and location is carried out, the training of the module is just
Sample is the fault point of wire clamp, therefore can be used for the fault condition that wire clamp gets rusty and carry out localization of fault, and in faulty wire clamp
It is out of order a little in figure with encirclement circle and shows fault type, accomplish the intelligent measurement of wire clamp failure.
Faster R-CNN work flow diagram is as shown in figure 3, Faster R-CNN method includes 2 CNN networks: region mentions
Discuss network RPN(Regional Proposal Network) and Fast R-CNN detection network.Wherein RPN is full convolutional network,
Its core concept is proposed using the direct generating region of convolutional neural networks, and a series of multiple dimensioned more length-width ratios are generated in picture
Candidate frame.Fast R-CNN is detected based on the RPN candidate frame extracted and is identified target therein.
Fig. 4 is Faster R-CNN network structure, and specific detection process is as follows: the image of input passes through convolution first
Layer is operated by convolution sum pondization, reduction image size and depth extraction characteristics of image, is formed in the last layer volume base special
Sign figure, characteristic pattern are the deep layer convolution feature of input picture, and the further feature of similar object is very close；Without similar object
Further feature is widely different, i.e., object has good separability on characteristic pattern.It is sliding that RPN network carries out window on characteristic pattern
It is dynamic, a series of candidate frames are extracted, and judge window for target or background.Network is detected finally by Fast R-CNN, it should
Network equally carries out feature extraction to characteristic pattern, and the candidate frame that the provincial characteristics of extraction and RPN are extracted is passed through ROI together
Pooling layers, pooling layers of ROI are collected simultaneously characteristic pattern and candidate frame, and carry out Region Feature Extraction, and after being sent to
The full articulamentum of continuous two layers simultaneously determines target category by softmax, while the position of image where displaying target, and to detection
Frame is modified, so that the position of detection block is more accurate.Faster R-CNN method can be classified and be positioned, and can be looked for simultaneously
The position of terminal clamp simultaneously identifies its type, batch processing can also be similarly carried out to wire clamp fault type, efficiently intelligence.
Embodiment two
Based on detecting inventive concept same as defect identification method automatically with power grid wire clamp a kind of in previous embodiment, the present invention
A kind of power grid wire clamp is also provided to detect automatically and defect recognizing device.
Referring to Fig. 5, a kind of power grid wire clamp detects automatically and defect recognizing device, comprising:
Image capture module 201, for obtaining wire clamp picture；
Wire clamp type training module 202 obtains wire clamp type for being trained to wire clamp picture；
Wire clamp categorization module 203 deletes choosing for carrying out classification to wire clamp picture；
Wire clamp fault detection module 204, for failure wire clamp to be identified and exported.
One of embodiment one power grid wire clamp detects with the various change mode of defect identification method and specifically real automatically
A kind of power grid wire clamp that example is equally applicable to the present embodiment detects automatically and defect recognizing device, by aforementioned to a kind of grid line
The detailed description of folder automatic detection and defect identification method, those skilled in the art are clear that in the present embodiment a kind of
Power grid wire clamp detects automatically and defect recognizing device, so this will not be detailed here in order to illustrate the succinct of book.
Above-described embodiment is classical case of the invention, but embodiments of the present invention are not by the limit of above-described embodiment
System.Other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention etc.,
It should be equivalent substitute mode, be included within the scope of the present invention.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Finally it should be noted that: the above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof, institute
The those of ordinary skill in category field can still modify to a specific embodiment of the invention referring to above-described embodiment or
Equivalent replacement, these are applying for this pending hair without departing from any modification of spirit and scope of the invention or equivalent replacement
Within bright claims.
Claims (8)
1. a kind of power grid wire clamp detects automatically and defect identification method, which comprises the following steps:
Wire clamp picture is obtained by image capture module；
Wire clamp picture is trained and obtains wire clamp type；
Classification is carried out to wire clamp picture and deletes choosing；
Failure wire clamp is identified and exported.
2. the method according to claim 1, wherein described obtain wire clamp picture, tool by image capture module
Body includes: the style of shooting combined using unmanned plane and photographic device, is shot to obtain line to the wire clamp on transmission line of electricity
Press from both sides picture.
3. the method according to claim 1, wherein described be trained to wire clamp picture obtains wire clamp classification；
It specifically includes:
A certain number of wire clamp pictures are chosen as training sample；
Terminal clamp feature, which is extracted, using the convolution kernel in Fast R-CNN is trained acquisition wire clamp classification.
4. according to the method described in claim 3, it is characterized in that, described be trained to wire clamp picture obtains wire clamp classification；
Further include: augmentation training is carried out for non-classified wire clamp picture, obtains wire clamp augmentation type.
5. according to the method described in claim 3, it is characterized in that, the wire clamp feature includes wire clamp color, wire clamp profile, line
Press from both sides texture.
6. the method according to claim 1, wherein it is described to wire clamp picture carry out classification delete choosing；It specifically includes:
Propose that network generates on wire clamp picture by region and surrounds frame；
Whether judge to surround in frame is target wire clamp；
If surrounding is target wire clamp in frame, point that its classification completes wire clamp picture is identified by the classifier in Fast R-CNN
Class.
7. the method according to claim 1, wherein failure wire clamp is identified and is exported；It specifically includes:
Wire clamp fault point samples pictures are trained, wire clamp fault category is obtained；
By extracting the encirclement frame of target wire clamp based on the convolution kernel in Fast R-CNN,
Fault location is carried out to the target wire clamp surrounded in frame, identifies fault type；
Export abort situation and fault type.
8. any method proposes that a kind of power grid wire clamp detects and defect recognizing device automatically in -7 according to claim 1,
It is characterised by comprising:
Image capture module, for obtaining wire clamp picture；
Wire clamp type training module obtains wire clamp type for being trained to wire clamp picture；
Wire clamp categorization module deletes choosing for carrying out classification to wire clamp picture；
Wire clamp fault detection module, for failure wire clamp to be identified and exported.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513439.7A CN110378221A (en) | 2019-06-14 | 2019-06-14 | A kind of power grid wire clamp detects and defect identification method and device automatically |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513439.7A CN110378221A (en) | 2019-06-14 | 2019-06-14 | A kind of power grid wire clamp detects and defect identification method and device automatically |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110378221A true CN110378221A (en) | 2019-10-25 |
Family
ID=68250345
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910513439.7A Pending CN110378221A (en) | 2019-06-14 | 2019-06-14 | A kind of power grid wire clamp detects and defect identification method and device automatically |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110378221A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110826476A (en) * | 2019-11-02 | 2020-02-21 | 国网浙江省电力有限公司杭州供电公司 | Image detection method and device for identifying target object, electronic equipment and storage medium |
| CN110910360A (en) * | 2019-11-14 | 2020-03-24 | 腾讯云计算（北京）有限责任公司 | Power grid image positioning method and image positioning model training method |
| CN111047561A (en) * | 2019-11-22 | 2020-04-21 | 国网江西省电力有限公司电力科学研究院 | Method for identifying cracks of shed tortoise of composite insulator |
| CN111091544A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting breakage fault of side integrated framework of railway wagon bogie |
| CN111161244A (en) * | 2019-12-30 | 2020-05-15 | 华南理工大学 | Industrial product surface defect detection method based on FCN + FC-WXGboost |
| CN111402255A (en) * | 2020-04-10 | 2020-07-10 | 山东大学 | Side formwork failure detection method and system based on deep learning |
| CN111951229A (en) * | 2020-07-22 | 2020-11-17 | 国网安徽省电力有限公司电力科学研究院 | Small hardware fitting image data set training method based on full connection layer augmentation |
| CN112884738A (en) * | 2021-02-18 | 2021-06-01 | 国网江苏省电力有限公司扬州供电分公司 | Suspension clamp inclination fault detection method based on inclined frame marking |
| CN113657280A (en) * | 2021-08-18 | 2021-11-16 | 广东电网有限责任公司 | Power transmission line target defect detection warning method and system |
| CN113902739A (en) * | 2021-11-29 | 2022-01-07 | 广东电网有限责任公司肇庆供电局 | NUT wire clamp defect identification method, device and equipment and readable storage medium |
| CN114299359A (en) * | 2021-12-22 | 2022-04-08 | 山东浪潮科学研究院有限公司 | Method, equipment and storage medium for detecting transmission line fault |
| CN114299359B (en) * | 2021-12-22 | 2024-05-10 | 山东浪潮科学研究院有限公司 | Method, equipment and storage medium for detecting power transmission line faults |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018119684A1 (en) * | 2016-12-27 | 2018-07-05 | 深圳前海达闼云端智能科技有限公司 | Image recognition system and image recognition method |
| CN108280855A (en) * | 2018-01-13 | 2018-07-13 | 福州大学 | A kind of insulator breakdown detection method based on Fast R-CNN |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
| CN109872317A (en) * | 2019-02-22 | 2019-06-11 | 西南交通大学 | A kind of defect identification method based on power equipments defect identification learning model |
-
2019
- 2019-06-14 CN CN201910513439.7A patent/CN110378221A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018119684A1 (en) * | 2016-12-27 | 2018-07-05 | 深圳前海达闼云端智能科技有限公司 | Image recognition system and image recognition method |
| CN108280855A (en) * | 2018-01-13 | 2018-07-13 | 福州大学 | A kind of insulator breakdown detection method based on Fast R-CNN |
| CN109872317A (en) * | 2019-02-22 | 2019-06-11 | 西南交通大学 | A kind of defect identification method based on power equipments defect identification learning model |
| CN109765462A (en) * | 2019-03-05 | 2019-05-17 | 国家电网有限公司 | Fault detection method, device and the terminal device of transmission line of electricity |
Non-Patent Citations (1)
| Title |
|---|
| 常海涛;苟军年;李晓梅;: "Faster R-CNN在工业CT图像缺陷检测中的应用" * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110826476A (en) * | 2019-11-02 | 2020-02-21 | 国网浙江省电力有限公司杭州供电公司 | Image detection method and device for identifying target object, electronic equipment and storage medium |
| CN110910360A (en) * | 2019-11-14 | 2020-03-24 | 腾讯云计算（北京）有限责任公司 | Power grid image positioning method and image positioning model training method |
| CN111047561A (en) * | 2019-11-22 | 2020-04-21 | 国网江西省电力有限公司电力科学研究院 | Method for identifying cracks of shed tortoise of composite insulator |
| CN111091544A (en) * | 2019-12-12 | 2020-05-01 | 哈尔滨市科佳通用机电股份有限公司 | Method for detecting breakage fault of side integrated framework of railway wagon bogie |
| CN111161244B (en) * | 2019-12-30 | 2021-12-21 | 华南理工大学 | Industrial product surface defect detection method based on FCN + FC-WXGboost |
| CN111161244A (en) * | 2019-12-30 | 2020-05-15 | 华南理工大学 | Industrial product surface defect detection method based on FCN + FC-WXGboost |
| CN111402255A (en) * | 2020-04-10 | 2020-07-10 | 山东大学 | Side formwork failure detection method and system based on deep learning |
| CN111951229A (en) * | 2020-07-22 | 2020-11-17 | 国网安徽省电力有限公司电力科学研究院 | Small hardware fitting image data set training method based on full connection layer augmentation |
| CN112884738A (en) * | 2021-02-18 | 2021-06-01 | 国网江苏省电力有限公司扬州供电分公司 | Suspension clamp inclination fault detection method based on inclined frame marking |
| CN113657280A (en) * | 2021-08-18 | 2021-11-16 | 广东电网有限责任公司 | Power transmission line target defect detection warning method and system |
| CN113902739A (en) * | 2021-11-29 | 2022-01-07 | 广东电网有限责任公司肇庆供电局 | NUT wire clamp defect identification method, device and equipment and readable storage medium |
| CN113902739B (en) * | 2021-11-29 | 2022-03-22 | 广东电网有限责任公司肇庆供电局 | NUT wire clamp defect identification method, device and equipment and readable storage medium |
| CN114299359A (en) * | 2021-12-22 | 2022-04-08 | 山东浪潮科学研究院有限公司 | Method, equipment and storage medium for detecting transmission line fault |
| CN114299359B (en) * | 2021-12-22 | 2024-05-10 | 山东浪潮科学研究院有限公司 | Method, equipment and storage medium for detecting power transmission line faults |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110378221A (en) | A kind of power grid wire clamp detects and defect identification method and device automatically | |
| CN110598736B (en) | Power equipment infrared image fault positioning, identifying and predicting method | |
| CN111368690B (en) | Deep learning-based video image ship detection method and system under influence of sea waves | |
| CN107133943A (en) | A kind of visible detection method of stockbridge damper defects detection | |
| CN110909690A (en) | Method for detecting occluded face image based on region generation | |
| CN109712127B (en) | Power transmission line fault detection method for machine inspection video stream | |
| CN109886153B (en) | Real-time face detection method based on deep convolutional neural network | |
| CN111582092B (en) | Pedestrian abnormal behavior detection method based on human skeleton | |
| CN112541393A (en) | Transformer substation personnel detection method and device based on deep learning | |
| CN112801019B (en) | Method and system for eliminating re-identification deviation of unsupervised vehicle based on synthetic data | |
| CN114972312A (en) | Improved insulator defect detection method based on YOLOv4-Tiny | |
| CN110334685A (en) | Flame detecting method, fire defector model training method, storage medium and system | |
| CN109086657B (en) | A kind of ear detection method, system and model based on machine learning | |
| CN110618129A (en) | Automatic power grid wire clamp detection and defect identification method and device | |
| Wang et al. | Railway insulator detection based on adaptive cascaded convolutional neural network | |
| CN115147383A (en) | Insulator state rapid detection method based on lightweight YOLOv5 model | |
| CN113689382B (en) | Tumor postoperative survival prediction method and system based on medical images and pathological images | |
| CN111126155A (en) | Pedestrian re-identification method for generating confrontation network based on semantic constraint | |
| CN108073873A (en) | Human face detection and tracing system based on high-definition intelligent video camera | |
| CN116630668A (en) | Method for identifying wearing abnormality of safety helmet in quick lightweight manner | |
| Ying et al. | An improved defect detection method for substation equipment | |
| Mou et al. | State recognition of electric control cabinet switches based on cnns | |
| CN115909192A (en) | Pedestrian detection method based on improved EfficientDet | |
| CN115601674A (en) | Power transmission project smoke hidden danger identification method and device and storage medium | |
| CN114359578A (en) | Application method and system of pest and disease damage identification intelligent terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200413 Address after: 210000 good faith Avenue, Jiangning economic and Technological Development Zone, Nanjing, Jiangsu Province, No. 19 Applicant after: NARI GROUP Corp. Applicant after: ELECTRIC POWER RESEARCH INSTITUTE OF STATE GRID SHANDONG ELECTRIC POWER Co. Applicant after: ANHUI NANRUI JIYUAN POWER GRID TECHNOLOGY Co.,Ltd. Applicant after: STATE GRID CORPORATION OF CHINA Address before: 230088 No. 90 science Avenue, hi tech Zone, Anhui, Hefei Applicant before: ANHUI NANRUI JIYUAN POWER GRID TECHNOLOGY Co.,Ltd. Applicant before: ELECTRIC POWER RESEARCH INSTITUTE OF STATE GRID SHANDONG ELECTRIC POWER Co. Applicant before: STATE GRID CORPORATION OF CHINA |
|
| TA01 | Transfer of patent application right | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |