CN106198551A - Method and device for detecting defects of power transmission line - Google Patents
Method and device for detecting defects of power transmission line Download PDFInfo
- Publication number
- CN106198551A CN106198551A CN201610630532.2A CN201610630532A CN106198551A CN 106198551 A CN106198551 A CN 106198551A CN 201610630532 A CN201610630532 A CN 201610630532A CN 106198551 A CN106198551 A CN 106198551A
- Authority
- CN
- China
- Prior art keywords
- image
- infrared
- sample
- transmission line
- detection
- 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
- 230000005540 biological transmission Effects 0.000 title claims abstract description 48
- 230000007547 defect Effects 0.000 title claims abstract description 40
- 238000000034 method Methods 0.000 title abstract description 13
- 238000001514 detection method Methods 0.000 claims abstract description 55
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims abstract description 13
- 238000013507 mapping Methods 0.000 claims abstract description 7
- 230000005611 electricity Effects 0.000 claims description 38
- 238000012549 training Methods 0.000 claims description 27
- 238000000605 extraction Methods 0.000 claims description 11
- 238000013480 data collection Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 239000000284 extract Substances 0.000 claims description 5
- 230000003252 repetitive effect Effects 0.000 claims description 3
- 238000004971 IR microspectroscopy Methods 0.000 claims 1
- 238000010801 machine learning Methods 0.000 description 8
- 239000012212 insulator Substances 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 210000001367 artery Anatomy 0.000 description 1
- 230000000875 corresponding effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000006116 polymerization reaction Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000001757 thermogravimetry curve Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The embodiment of the invention provides a method and a device for detecting defects of a power transmission line, relates to the field of power element detection, and can improve the accuracy of judging the defects of the power transmission line. The specific scheme comprises the following steps: determining a feature extractor through unsupervised learning, wherein the feature extractor is a mapping function between the infrared image and the data features of the infrared image; acquiring a target infrared image, and extracting the data characteristics of the target infrared image through the characteristic extractor; the target infrared image is an infrared image of the power transmission line; classifying and detecting the data characteristics of the target infrared image by using a classifier, and identifying a positive sample of the target infrared image; the positive sample is an infrared image sample of a power element on the power transmission line; and carrying out temperature analysis on the positive sample and outputting a defect judgment result. The invention is used for detecting the defects of the power transmission line.
Description
Technical field
The present invention relates to force device detection field, particularly relate to detection method and the device of a kind of transmission line of electricity defect.
Background technology
Power industry is the important foundation of national economy, and it provides basic motive for national economy and other departments, is state
Emphasis in family's Strategy for economic development.The geographical environment that transmission line of electricity passes through is complicated, away from main traffic artery, and power transmission line
Road is easily affected by factors such as natural disaster and artificial damages in longtime running, causes that wire strand breakage, insulator be dirty, shaft tower rust
The defects such as erosion, the early stage of this type of defect is often attended by shelf depreciation, local temperature rise, the phenomenon such as increases, it will have a strong impact on electrical network
Operation security.It is thus desirable to circuit is patrolled and examined, find in time to get rid of line defct.
In prior art, utilize helicopter/unmanned aerial vehicle platform, by various on transmission line of electricity of infrared detection technology
Equipment carries out on-line checking, is then based on image recognition technology and is analyzed acquired infrared image processing, so that it is determined that
Line defct.
But, in practical service environment, affected by complex environment equipment each on power transmission line feature, examined by infrared ray
Containing substantial amounts of noise and complicated background information in the image that survey technology obtains, cause detection side based on image recognition technology
Method is relatively low to force device recognition accuracy, there is the more serious False Rate of ratio, thus the accuracy of judgement to transmission line of electricity defect
Rate is not enough.
Summary of the invention
Embodiments of the invention provide detection method and the device of a kind of transmission line of electricity defect, it is possible to increase to transmission line of electricity
The judging nicety rate of defect.
In order to reach above-mentioned purpose, the present invention uses following solution:
First aspect, it is provided that the detection method of a kind of transmission line of electricity defect, including:
Determine that feature extractor, described feature extractor are the data of infrared image and infrared image by unsupervised learning
Mapping function between feature；
Obtain Infrared Targets image, and extracted the data spy obtaining described Infrared Targets image by described feature extractor
Levy；Wherein, described Infrared Targets image is the infrared image of transmission line of electricity；
Utilize grader that the data characteristics of described Infrared Targets image is carried out classification and Detection, identify described Infrared Targets
The positive sample of image；Wherein, described positive sample is the infrared image sample of force device on transmission line of electricity；
Described positive sample is carried out temperature analysis, exports defect estimation result.
Second aspect, it is provided that the detection device of a kind of transmission line of electricity defect, for performing the detection that first aspect is provided
Method.
The detection method of the transmission line of electricity defect that embodiments of the invention are provided and device, the method utilizing machine learning
Realizing the automatic study to Infrared Image Features, relative to the means of graphical analysis, the feature that machine learning obtains is to infrared figure
As information has more essential portraying, clearly embody the relation between Infrared image and transmission line of electricity defect, additionally engineering
Acquistion to feature also help dissimilar defect done classification and Detection, thus it is accurate to improve the judgement to transmission line of electricity defect
Really rate.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, embodiment will be described below
The accompanying drawing used required in is briefly described, it should be apparent that, the accompanying drawing in describing below is only some of the present invention
Embodiment, for those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to attached according to these
Figure obtains other accompanying drawing.
The detection method schematic flow sheet of the transmission line of electricity defect that Fig. 1 is provided by embodiments of the invention；
The structure of the detecting device schematic diagram of the transmission line of electricity defect that Fig. 2 is provided by embodiments of the invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Describe, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments wholely.Based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under not making creative work premise
Embodiment, broadly falls into the scope of protection of the invention.
Embodiment
Embodiments of the invention provide the detection method of a kind of transmission line of electricity defect, shown in Fig. 1, including following step
Rapid:
101, feature extractor is determined by unsupervised learning.
Feature extractor is the mapping function between infrared image and the data characteristics of infrared image.
The infrared image training sample set X={x of one not band class label is seta,xb,xc... }, by this collection
Conjunction is trained, and completes unsupervised learning process.
Unsupervised learning process can include following sub-step:
101-1, piecemeal operation will be carried out without label image, and extract at least one subimage block the most at random.
101-2, data at least one subimage block carry out pretreatment.
The overall light levels of image can't affect the analysis to objects in images, say, that can ignore image
The average brightness value of block, it is possible to deduct average brightness value to carry out average Regularization.Infrared image has smooth performance, logical
Often need each data sample is done respectively average Regularization in the first step, and divided by standard deviation, with Regularization.
Then use principal component analysis (English full name: principal component analysis, English abbreviation:
Or zero phase component analysis (English full name: zero-phase component analysis, English abbreviation: ZCA) is white PCA)
Change processes, to reach the effect of low-pass filtering.
PCA be a kind of can significant increase without supervise feature learning speed Data Dimensionality Reduction Algorithm.Image is used to train
Algorithm, because pixels tall adjacent in image is correlated with, input data have certain redundancy, and PCA algorithm can will input number
Be converted to a much lower vector approximation of dimension according to vector, and error is the least.
Albefaction is a pre-treatment step relevant to PCA, owing to having the strongest being correlated with in image between neighbor
Property, for training time input be redundancy, the purpose of albefaction be exactly reduce input redundancy.
101-3, sparse own coding algorithm is utilized to learn to obtain feature extractor.
Utilize sparse own coding algorithm to carry out learning characteristic to map, i.e. feature extractor.Feature extractor be infrared image with
Mapping function between the data characteristics of infrared image.
After data are carried out pretreatment, carry out learning characteristic by unsupervised-learning algorithm.Can be unsupervised-learning algorithm
Regard a flight data recorder as.It receives input, then produces an output.A Function Mapping function can be expressed as, a N
The input vector x of dimensioniIt is mapped as the characteristic vector of a K dimension.
Sparse own coding algorithm, with back-propagation algorithm (English full name: Backpropagation algorithm, English
Between claim: BP algorithm) train an automatic coding machine containing K concealed nodes, cost function is to reconstruct mean square error, and exists
One penalty term, major limitation concealed nodes so that it is keep a low activation value.Algorithm one weight matrix W (K × N of output
Dimension) and one group of base B (K dimension), feature extractor is: f (x)=g (Wx+b).Wherein, x is the input vector of N-dimensional, g (z)=1/ (1
+ e^ (-z)) it is sigmoid function, each the element evaluation to vector z.
102, obtain Infrared Targets image, and extracted the data characteristics obtaining Infrared Targets image by feature extractor.
Wherein, Infrared Targets image is the infrared image of transmission line of electricity.
From infrared image, randomly select an image block, utilize feature extractor, from the data choosing an image block
Learning is to some basic features, using these basic features as detector, with each subimage block of former infrared image is rolled up
Long-pending, obtain each subimage block respective convolution feature.
Optionally, after obtained each subimage block respective convolution feature by convolution, if utilizing these features straight
Connect to do and classify, the biggest amount of calculation will be faced.Therefore, in order to describe large-size images, the feature of diverse location is polymerized
Statistics, it is (English: pooling), pond is on the basis of convolution feature extraction, to each that the operation of this polymerization is pond
Convolution feature is carried out making even equalization, continue to zoom out concealed nodes for convolution intrinsic dimensionality, reduce the amount of calculation of grader.
It is to say, convolution feature is divided on several disjoint range, by the average of these regions or maximum feature
Obtain the convolution feature behind pond.These statistical natures not only have much lower dimension (comparing original convolution feature), with
Time would also avoid over-fitting.Feature through Chi Huahou just can be used to do classify.
103, utilize grader that the data characteristics of Infrared Targets image is carried out classification and Detection.
Grader passes through classification and Detection, identifies positive sample and the negative sample of Infrared Targets image, and wherein, positive sample is defeated
The infrared image sample of force device in electric line, can be described as the infrared image sample of foreground target, and negative sample is the red of background
The invalid sample such as outer image pattern.
In a kind of concrete implementation mode, use Softmax regression model that the number of drawbacks on transmission line of electricity is carried out
Classification and Detection.Training sample set (the x of one tape label is set(i),y(i))={ (x(1),y(1)),(x(2),y(2)),(x(3),y(3)) ..., wherein y(i)∈ 1,2 ... k}, k are the species number of transmission line of electricity defect.
For given sample input x, with assuming that function estimates probit p (y=j │ x) for each classification j, also
It is exactly to estimate the probability that each of x classification results occurs.The hypothesis function of Softmax is as follows:
Wherein, θ is model parameter.
As a example by insulator on power transmission line, as calculated input sample x, to belong to the other probit of insulator class big
When parameter Ψ (this parameter value can be determined according to cross-validation method), can differentiate that this subimage block is positive sample, i.e. insulator
Infrared image block.Otherwise, if less than parameter Ψ, it determines this subimage block is the infrared image sample etc. of negative sample, i.e. background
Invalid sample.
Optionally, after classification and Detection, it is possible to use labeled data collection carries out small parameter perturbations, plays correction feature extraction
The effect of device.
Grader is a big neutral net on the whole.Therefore, training obtain model initial parameters (utilize sparse from
Encoder training ground floor, utilizes the Softmax regression training second layer) after, parameter can be finely adjusted, in existing parameter
On the basis of use gradient to decline or conjugate gradient decent reduces the training error on labeled data collection.The work of fine setting
With being, labeled data collection may also be used for revising weights W, thus revises feature extractor f (x)=g (Wx+b), and utilization is repaiied
Feature extractor after just extracts feature, plays the purpose adjusting the feature extracted.
104, by PN study, the classification results of grader is estimated.
PN learns, and the constraint study of constraint negative i.e. certainly, English full name is positive constraint negative
constraint learning。
In a kind of concrete implementation mode, according to default constraints, by PN learning training grader.Work as classification
When testing result exists the sample with constraints contradiction to, the sample with constraints contradiction is added the training of grader
Sample set, when repetitive exercise grader is until constraints meets, the positive sample of output Infrared Targets image, with " passing through in Fig. 1
Positive sample after assessment " represent.
As a example by the positive sample subimage block as insulator.After utilizing Softmax to return classification, input sample x is judged to
Not Wei insulator, be positive sample.If this sample x meets constraints, the most i.e. can conclude that it is strictly foreground target exhausted
Edge.If this sample is unsatisfactory for constraints, the most again gives this sample label (representing with " sample correction " in Fig. 1), and add
Entering to training sample set, iteration is trained, until meeting constraints.
Constraints can be configured according to the practical situation of transmission line of electricity, and such as constraints specifically may be configured as:
Whether foreground target is at the near zone of transmission pressure.Because the linear target in infrared image is compared with it when being why arranged such
For its target the most notable.
It should be noted is that, the discriminant parameter Ψ that Softmax returns may be configured as a relatively small value, then
A fairly large number of positive sample will be produced.Therefore, in PN learning process, just have more sample and can return to training sample
Carry out iteration training, so can improve the detection efficiency of grader to a great extent.
105, align sample and carry out temperature analysis, export defect estimation result.
Relative temperature method is used to judge whether the force device identified exists thermal defect, i.e. according in equipment Thermogram picture
Relative temperature difference, calculate force device relative to temperature rise value, compare the temperature rise of normal condition, sentence based on temperature analysis output defect
Other result.
The detection method of the transmission line of electricity defect that embodiments of the invention are provided, utilizes the method for machine learning to realize right
The automatic study of Infrared Image Features, relative to the means of graphical analysis, the feature that machine learning obtains is to Infrared Image Information
Having more essential portraying, clearly embody the relation between Infrared image and transmission line of electricity defect, additionally machine learning obtains
Feature also help dissimilar defect done classification and Detection, thus improve the judging nicety rate to transmission line of electricity defect.
Further, embodiments of the invention introduce PN mechanism on the basis of exemplary depth convolutional neural networks, utilize
The Structural Characteristics existed between training sample and test sample progressively trains the grader of transmission line of electricity number of drawbacks, and
It is effectively improved the classification performance of grader.
Embodiments of the invention also provide for the detection device of a kind of transmission line of electricity defect, are used for performing institute in above-described embodiment
The detection method described.Detection step performed by device, identical with the step described in the embodiment corresponding to Fig. 1, this
Place is only briefly described.Shown in Fig. 2, detection device 20 includes:
Training unit 201, for determining feature extractor by unsupervised learning, feature extractor is that infrared image is with red
Mapping function between the data characteristics of outer image.
Feature extraction unit 202, is used for obtaining Infrared Targets image, and obtains Infrared Targets by feature extractor extraction
The data characteristics of image.Wherein, Infrared Targets image is the infrared image of transmission line of electricity.
Classification and Detection unit 203, for utilizing grader that the data characteristics of Infrared Targets image is carried out classification and Detection, knows
Do not go out the positive sample of Infrared Targets image.Wherein, positive sample is the infrared image sample of force device on transmission line of electricity.
Temperature analysis unit 204, is used for aligning sample and carries out temperature analysis, export defect estimation result.
Optionally, detection device 20 also includes certainly retraining negative constraint PN unit, for according to the constraint preset
Condition, by PN learning training grader.
Classification and Detection unit 203, specifically for existing and constraints in determining classification and Detection result when PN unit
During the sample of contradiction, the sample with constraints contradiction adding the training sample set of grader to, repetitive exercise grader is straight
To constraints meet time, output Infrared Targets image positive sample.
Optionally, classification and Detection unit 203, it is additionally operable to utilize the spy that labeled data collection correction training unit 201 determines
Levy extractor.
Optionally, training unit 201, specifically for carrying out piecemeal operation without label image, and therefrom extracts extremely at random
A few subimage block.The data of at least one subimage block are carried out pretreatment.Sparse own coding algorithm is utilized to learn
To feature extractor.
Optionally, feature extraction unit 202, specifically for randomly selecting an image block from infrared image, utilize spy
Levy extractor, from the data learning of an image block to basic feature, by basic feature with each subgraph of former infrared image
As block does convolution, obtain each subimage block respective convolution feature.
The detection device of the transmission line of electricity defect that embodiments of the invention are provided, utilizes the method for machine learning to realize right
The automatic study of Infrared Image Features, relative to the means of graphical analysis, the feature that machine learning obtains is to Infrared Image Information
Having more essential portraying, clearly embody the relation between Infrared image and transmission line of electricity defect, additionally machine learning obtains
Feature also help dissimilar defect done classification and Detection, thus improve the judging nicety rate to transmission line of electricity defect.
Further, embodiments of the invention introduce PN mechanism on the basis of exemplary depth convolutional neural networks, utilize
The Structural Characteristics existed between training sample and test sample progressively trains the grader of transmission line of electricity number of drawbacks, and
It is effectively improved the classification performance of grader.
Above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any familiar
Those skilled in the art, in the technical scope that the invention discloses, can readily occur in change or replace, should contain
Within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with scope of the claims.
Claims (10)
1. the detection method of a transmission line of electricity defect, it is characterised in that including:
Determine that feature extractor, described feature extractor are the data characteristics of infrared image and infrared image by unsupervised learning
Between mapping function；
Obtain Infrared Targets image, and extracted the data characteristics obtaining described Infrared Targets image by described feature extractor；
Wherein, described Infrared Targets image is the infrared image of transmission line of electricity；
Utilize grader that the data characteristics of described Infrared Targets image is carried out classification and Detection, identify described Infrared Targets image
Positive sample；Wherein, described positive sample is the infrared image sample of force device on transmission line of electricity；
Described positive sample is carried out temperature analysis, exports defect estimation result.
Detection method the most according to claim 1, it is characterised in that
Described detection method also includes: according to default constraints, described in the constraint PN learning training of constraint negative certainly
Grader；
Described utilize grader that the data characteristics of described Infrared Targets image is carried out classification and Detection, identify described Infrared Targets
The positive sample of image, including:
When classification and Detection result exists the sample with described constraints contradiction, by the sample with described constraints contradiction
Add the training sample set of described grader to, when grader described in repetitive exercise is until described constraints meets, export institute
State the positive sample of Infrared Targets image.
Detection method the most according to claim 1, it is characterised in that described detection method also includes:
Utilize feature extractor described in labeled data collection correction.
Detection method the most according to claim 1, it is characterised in that described determine feature extraction by unsupervised learning
Device, including:
Piecemeal operation will be carried out without label image, and extract at least one subimage block the most at random；
The data of at least one subimage block described are carried out pretreatment；
Sparse own coding algorithm is utilized to learn to obtain described feature extractor.
Detection method the most according to claim 1, it is characterised in that described extraction by described feature extractor obtains institute
State the data characteristics of Infrared Targets image, including:
From infrared image, randomly select an image block, utilize described feature extractor, from the data of one image block
Described basic feature, to basic feature, is done convolution with each subimage block of former infrared image, is obtained each subgraph by learning
As block respective convolution feature.
6. the detection device of a transmission line of electricity defect, it is characterised in that including:
By unsupervised learning, training unit, for determining that feature extractor, described feature extractor are that infrared image is with infrared
Mapping function between the data characteristics of image；
Feature extraction unit, is used for obtaining Infrared Targets image, and it is red to obtain described target by the extraction of described feature extractor
The data characteristics of outer image；Wherein, described Infrared Targets image is the infrared image of transmission line of electricity；
Classification and Detection unit, for utilizing grader that the data characteristics of described Infrared Targets image is carried out classification and Detection, identifies
Go out the positive sample of described Infrared Targets image；Wherein, described positive sample is the infrared image sample of force device on transmission line of electricity；
Temperature analysis unit, for described positive sample carries out temperature analysis, exports defect estimation result.
Detection device the most according to claim 6, it is characterised in that
Described detection device also includes certainly retraining negative constraint PN unit, for according to the constraints preset, passing through
Grader described in PN learning training；
Described classification and Detection unit, specifically for existing and described constraint when described PN unit determines in classification and Detection result
During the sample of condition contradiction to, the sample with described constraints contradiction is added the training sample set of described grader, iteration
When training described grader until described constraints meets, export the positive sample of described Infrared Targets image.
Detection device the most according to claim 6, it is characterised in that
Described classification and Detection unit, is additionally operable to the described feature extraction utilizing training unit described in labeled data collection correction to determine
Device.
Detection device the most according to claim 6, it is characterised in that
Described training unit, specifically for carrying out piecemeal operation without label image, and extracts at least one subgraph the most at random
As block；The data of at least one subimage block described are carried out pretreatment；Sparse own coding algorithm is utilized to learn to obtain described
Feature extractor.
Detection device the most according to claim 6, it is characterised in that described feature extraction unit, specifically for from infrared
Image randomly selects an image block, utilizes described feature extractor, from the data learning of one image block to base
Eigen, does convolution by described basic feature with each subimage block of former infrared image, obtains each subimage block respective
Convolution feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610630532.2A CN106198551A (en) | 2016-08-01 | 2016-08-01 | Method and device for detecting defects of power transmission line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610630532.2A CN106198551A (en) | 2016-08-01 | 2016-08-01 | Method and device for detecting defects of power transmission line |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106198551A true CN106198551A (en) | 2016-12-07 |
Family
ID=57498340
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610630532.2A Pending CN106198551A (en) | 2016-08-01 | 2016-08-01 | Method and device for detecting defects of power transmission line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106198551A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107024506A (en) * | 2017-03-09 | 2017-08-08 | 深圳市朗驰欣创科技股份有限公司 | A kind of pyrogenicity defect inspection method and system |
| CN107179475A (en) * | 2017-05-27 | 2017-09-19 | 国网福建省电力有限公司 | A kind of unmanned plane device for protecting power high voltage circuit discharge fault and guard method |
| 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 |
| CN108734691A (en) * | 2018-03-22 | 2018-11-02 | 广东电网有限责任公司清远供电局 | A kind of transmission line of electricity defect image recognition methods |
| CN108734117A (en) * | 2018-05-09 | 2018-11-02 | 国网浙江省电力有限公司电力科学研究院 | Cable machinery external corrosion failure evaluation method based on YOLO |
| CN109447978A (en) * | 2018-11-09 | 2019-03-08 | 河北工业大学 | A kind of photovoltaic solar cell piece electroluminescent image defect classification method |
| CN110057918A (en) * | 2019-05-29 | 2019-07-26 | 山东大学 | Damage of composite materials quantitative identification method and system under strong noise background |
| CN110726724A (en) * | 2019-10-22 | 2020-01-24 | 北京百度网讯科技有限公司 | Defect detection method, system and device |
| CN112395936A (en) * | 2019-08-12 | 2021-02-23 | 湾流航空航天公司 | System and method for detecting defects in a structure subject to repetitive stress |
| CN113297979A (en) * | 2021-05-27 | 2021-08-24 | 南方电网电力科技股份有限公司 | Method and device for identifying heating state of power transmission wire connector |
| CN113469953A (en) * | 2021-06-10 | 2021-10-01 | 南昌大学 | Transmission line insulator defect detection method based on improved YOLOv4 algorithm |
| CN115272761A (en) * | 2022-07-26 | 2022-11-01 | 国网四川省电力公司电力科学研究院 | Power grid transmission line defect detection method based on collaborative representation bilateral self-encoder |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070036420A1 (en) * | 2003-02-18 | 2007-02-15 | Marena Systems Corporation | Methods for analyzing defect artifacts to precisely locate corresponding defects |
| CN104764748A (en) * | 2015-05-04 | 2015-07-08 | 成都唐源电气有限责任公司 | Method and system for positioning insulators and method and system for fault detection |
| CN104778474A (en) * | 2015-03-23 | 2015-07-15 | 四川九洲电器集团有限责任公司 | Classifier construction method for target detection and target detection method |
| CN105403312A (en) * | 2015-10-28 | 2016-03-16 | 国家电网公司 | Infrared imaging spectrum recognition and analysis method for power transmission and transformation equipment |
| CN105741291A (en) * | 2016-01-30 | 2016-07-06 | 西南交通大学 | Method for detecting faults of equipotential lines of high-speed railway overhead line system suspension devices |
-
2016
- 2016-08-01 CN CN201610630532.2A patent/CN106198551A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070036420A1 (en) * | 2003-02-18 | 2007-02-15 | Marena Systems Corporation | Methods for analyzing defect artifacts to precisely locate corresponding defects |
| CN104778474A (en) * | 2015-03-23 | 2015-07-15 | 四川九洲电器集团有限责任公司 | Classifier construction method for target detection and target detection method |
| CN104764748A (en) * | 2015-05-04 | 2015-07-08 | 成都唐源电气有限责任公司 | Method and system for positioning insulators and method and system for fault detection |
| CN105403312A (en) * | 2015-10-28 | 2016-03-16 | 国家电网公司 | Infrared imaging spectrum recognition and analysis method for power transmission and transformation equipment |
| CN105741291A (en) * | 2016-01-30 | 2016-07-06 | 西南交通大学 | Method for detecting faults of equipotential lines of high-speed railway overhead line system suspension devices |
Non-Patent Citations (1)
| Title |
|---|
| 王勇 等: "基于稀疏自编码深度神经网络的林火图像分类", 《计算机工程与应用》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107024506B (en) * | 2017-03-09 | 2020-06-26 | 深圳市朗驰欣创科技股份有限公司 | Pyrogenicity defect detection method and system |
| CN107024506A (en) * | 2017-03-09 | 2017-08-08 | 深圳市朗驰欣创科技股份有限公司 | A kind of pyrogenicity defect inspection method and system |
| CN107179475A (en) * | 2017-05-27 | 2017-09-19 | 国网福建省电力有限公司 | A kind of unmanned plane device for protecting power high voltage circuit discharge fault and guard method |
| 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 |
| CN108734691A (en) * | 2018-03-22 | 2018-11-02 | 广东电网有限责任公司清远供电局 | A kind of transmission line of electricity defect image recognition methods |
| CN108734117A (en) * | 2018-05-09 | 2018-11-02 | 国网浙江省电力有限公司电力科学研究院 | Cable machinery external corrosion failure evaluation method based on YOLO |
| CN109447978A (en) * | 2018-11-09 | 2019-03-08 | 河北工业大学 | A kind of photovoltaic solar cell piece electroluminescent image defect classification method |
| CN109447978B (en) * | 2018-11-09 | 2021-11-02 | 河北工业大学 | Method for classifying defects of electroluminescence image of photovoltaic solar cell |
| CN110057918A (en) * | 2019-05-29 | 2019-07-26 | 山东大学 | Damage of composite materials quantitative identification method and system under strong noise background |
| CN110057918B (en) * | 2019-05-29 | 2020-11-06 | 山东大学 | Method and system for quantitatively identifying damage of composite material under strong noise background |
| CN112395936A (en) * | 2019-08-12 | 2021-02-23 | 湾流航空航天公司 | System and method for detecting defects in a structure subject to repetitive stress |
| CN110726724A (en) * | 2019-10-22 | 2020-01-24 | 北京百度网讯科技有限公司 | Defect detection method, system and device |
| CN113297979A (en) * | 2021-05-27 | 2021-08-24 | 南方电网电力科技股份有限公司 | Method and device for identifying heating state of power transmission wire connector |
| CN113469953A (en) * | 2021-06-10 | 2021-10-01 | 南昌大学 | Transmission line insulator defect detection method based on improved YOLOv4 algorithm |
| CN115272761A (en) * | 2022-07-26 | 2022-11-01 | 国网四川省电力公司电力科学研究院 | Power grid transmission line defect detection method based on collaborative representation bilateral self-encoder |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106198551A (en) | Method and device for detecting defects of power transmission line | |
| CN110827251B (en) | Power transmission line locking pin defect detection method based on aerial image | |
| CN109977812B (en) | Vehicle-mounted video target detection method based on deep learning | |
| CN112380952B (en) | Power equipment infrared image real-time detection and identification method based on artificial intelligence | |
| KR102263397B1 (en) | Method for acquiring sample images for inspecting label among auto-labeled images to be used for learning of neural network and sample image acquiring device using the same | |
| CN109118479B (en) | Capsule network-based insulator defect identification and positioning device and method | |
| CN110070530B (en) | Transmission line icing detection method based on deep neural network | |
| CN108830188B (en) | Vehicle detection method based on deep learning | |
| CN106595551B (en) | Ice covering thickness detection method in powerline ice-covering image based on deep learning | |
| CN109977921B (en) | Method for detecting hidden danger of power transmission line | |
| CN111368690B (en) | Deep learning-based video image ship detection method and system under influence of sea waves | |
| CN104809722A (en) | Electrical device fault diagnosis method based on infrared thermography | |
| CN109214308A (en) | A kind of traffic abnormity image identification method based on focal loss function | |
| CN105184271A (en) | Automatic vehicle detection method based on deep learning | |
| CN107316300A (en) | A kind of tire X-ray defect detection method based on depth convolutional neural networks | |
| CN106504233A (en) | Image electric power widget recognition methodss and system are patrolled and examined based on the unmanned plane of Faster R CNN | |
| CN113486865A (en) | Power transmission line suspended foreign object target detection method based on deep learning | |
| CN106991666B (en) | A kind of disease geo-radar image recognition methods suitable for more size pictorial informations | |
| CN107808133A (en) | Oil-gas pipeline safety monitoring method, system and software memory based on unmanned plane line walking | |
| CN113469953A (en) | Transmission line insulator defect detection method based on improved YOLOv4 algorithm | |
| CN109766936A (en) | Image change detection method based on information transmitting and attention mechanism | |
| CN101251896B (en) | Object detecting system and method based on multiple classifiers | |
| CN103824092A (en) | Image classification method for monitoring state of electric transmission and transformation equipment on line | |
| CN108932471B (en) | Vehicle detection method | |
| CN114863122B (en) | Intelligent high-precision pavement disease identification method based on artificial intelligence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161207 |