CN106874836A - A kind of cable connector running rate recognizing method based on Infrared Thermogram - Google Patents
A kind of cable connector running rate recognizing method based on Infrared Thermogram Download PDFInfo
- Publication number
- CN106874836A CN106874836A CN201611244033.6A CN201611244033A CN106874836A CN 106874836 A CN106874836 A CN 106874836A CN 201611244033 A CN201611244033 A CN 201611244033A CN 106874836 A CN106874836 A CN 106874836A
- Authority
- CN
- China
- Prior art keywords
- infrared thermogram
- cable connector
- running status
- infrared
- method based
- 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
Classifications
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/192—Recognition using electronic means using simultaneous comparisons or correlations of the image signals with a plurality of references
- G06V30/194—References adjustable by an adaptive method, e.g. learning
Abstract
The present invention relates to the state identification method in a kind of machine learning field, more particularly to a kind of cable connector running rate recognizing method based on Infrared Thermogram, ELM recognition methods of the invention can randomly generate hidden layer node parameters, then output is determined using the weights for obtaining, enormously simplify the complicated iterative process of traditional BP neural network, and recognition accuracy is high, cable detection field is met well to accuracy rate and the double requirements of speed.
Description
Technical field
It is more particularly to a kind of to be based on infrared thermal imagery the present invention relates to the state identification method in a kind of machine learning field
The cable connector running rate recognizing method of figure.
Background technology
Cable connector is one of weak link in electric power netting safe running, and the generation of fire incident is superheated to from cable connector, whole
Individual process development speed is slow, the time is more long, such as on June 18th, 2016, the explosion accident of Xi'an transformer station, and in October, 2016
13 days, Tokyo massive blackout accident etc., be all because cable temperature is too high, it is on fire to cause.The safe operation of power network, no
It is only closely bound up with daily life, also drastically influence a socio-economic activity for country, therefore cable fault
Early prevention and maintenance are highly important.But at present, the diagnosis of most of cables is periodically patrolled by patrol workman, is provided
Then inspection report is assessed by expert again, and assessment detection can only be responsible for by experienced person, otherwise, as a result often by mistake
Diagnosis, this method is fairly time consuming and somewhat expensive.In practical operation, because cable patrols and examines the substantial amounts of manpower inspection of needs
Look into, these manpowers are generally checked work appointment to outside contractor by most of Utilities Electric Co., and outside contractor submits to and checks report
After announcement, another connects a ground and assesses its validity for Utilities Electric Co., between experienced person need to complete multiple using the plenty of time
Miscellaneous appraisal, it is finally decided whether maintenance.
In order to overcome these to limit, it is ensured that power generation safe and highly efficient operation, cable status maintenance is proposed higher
It is required that, wherein infrared imagery technique has obvious superiority in cable condition monitoring.Infrared imaging is with Warm status
It is distributed as according to well whether being diagnosed to cable running status, it has does not stop transport, does not contact, at a distance, quickly, intuitively
Be imaged.Because the thermography of cable is the true description of Warm status and its Temperature Distribution under running status, and cable exists
Whether heat distribution under running status is normally to judge a key character whether in good condition, thus uses infrared imaging skill
Art can diagnose running status and its hidden danger defect by the analysis to cable thermography.Image processing techniques, pattern are known
Other technology is combined with infrared thermal imaging technique, and is applied in cable is patrolled and examined, for the reliability and fortune that improve electrical equipment
Row economic benefit, reduction maintenance cost have very important significance.
In recent years, infrared image identification research have developed rapidly.The thinking of binding pattern identification, can be represented using some
The characteristic of interesting part image carrys out training pattern, and the model that these are trained to can be included into thing to image to be processed
In the middle of the classification for first setting.At present, the image classification method based on machine learning is most widely used, in general, it
It is that some images using oneself through obtaining can most represent their characteristic to extract, these data are according to certain pattern
The key parameter of some common mathematical models is trained, then these Mathematical Modelings can just be believed according to treated unknown images
Breath is judged, and is referred in the middle of the classification for pre-setting.The image classification method of this pattern-recognition has two key factors,
One key factor is the structure of suitable images feature, and the measurement index of quality has two：One is that this feature is small and energy is accurate
The information that is included of description this image, especially information interested, and must be that the target only paid close attention to has, other
Background information does not have；Two is that this feature has good distinction, and characteristic spatially has certain clustering distribution, so
Be conducive to distinguishing.The key of wherein image procossing is to convert thereof into gray level image to be processed, many in terms of such gray scale
Algorithm can find application.Another crucial factor is exactly the selection of learner model, and more extensive method has BP now
Neural network algorithm and algorithm of support vector machine, some documents carry out biometric image classification with BP, but BP neural network is joined
Number is more, and training speed is slow, and requires mass data.
Extreme Learning Machine (ELM) method can randomly generate hidden layer node parameters, then utilize
The weights for obtaining determine output, enormously simplify the complicated iterative process of traditional BP neural network, overcome infrared image because
Data volume causes greatly the slow limitation of the speed of service of traditional artificial neural network, SVMs, and cable inspection is met well
Survey field is to accuracy rate and the double requirements of speed.
The content of the invention
In order to overcome the shortcomings of existing recognition methods, the learner parameter for needing to set is reduced, and improves recognition accuracy,
The invention provides a kind of cable connector running rate recognizing method based on Infrared Thermogram, including training stage and identification rank
Section.In the training stage, the picture element matrix of different running status lower cable connector Infrared Thermograms is extracted first, and bring ELM into
Device is practised to be trained；In cognitive phase, cable connector Infrared Thermogram to be tested is entered with the ELM learners for having trained
Row identification, and it is identified result.
Cable connector running rate recognizing method based on Infrared Thermogram of the invention, comprises the following steps：
Step one：It is input into the Infrared Thermogram of different running status lower cable connectors；
Step 2：Infrared Thermogram to different running status lower cable connectors carries out Bit Plane Decomposition；
Step 3：Initial characteristicses matrix is set up, the characteristic parameter of each bit plane of Infrared Thermogram is extracted（Hu squares and Zernike
Square）, the characteristic parameter that will be extracted is brought into eigenmatrix；
Step 4：Infrared Thermogram characteristic parameter to different running status lower cable connectors carries out ELM learner training；
Step 5：The state recognition model that the cable connector Infrared Thermogram input of collection has been trained, and obtain corresponding
Cable connector running status result, i.e. running status are normal or abnormal.
As preferred scheme：
In order to reduce discrete the brought error of Infrared Thermogram, in above-mentioned steps three, the Zernike squares to extracting are returned
One change is processed.
In above-mentioned steps two, when carrying out Bit Plane Decomposition to Infrared Thermogram, first by Infrared Thermogram gray processing, will be defeated
The Infrared Thermogram of the different running status lower cable connectors for entering is transformed into the gray-scale map that pixel is 320*240, then enters again
Line position plane decomposition.
Cable connector running rate recognizing method of the invention can randomly generate hidden layer node parameters, then using obtaining
Weights determine output, enormously simplify the complicated iterative process of traditional BP neural network, and recognition accuracy is high, well
Cable detection field is met to accuracy rate and the double requirements of speed.
Brief description of the drawings
Fig. 1 is flow chart of the invention.
Fig. 2 is 8 bit plans after wherein one Infrared Thermogram of the embodiment of the present invention is decomposed.
In figure, 1 original image, 2 the 1st bit planes, 3 the 2nd bit planes, 4 the 3rd bit planes, 5 the 4th bit planes, 6
5th bit plane, 7 the 6th bit planes, 8 the 7th bit planes, 9 the 8th bit planes.
Specific embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those skilled in the art are not having
There is the every other embodiment made and being obtained under the premise of creative work, belong to the scope of protection of the invention.
The cable connector running rate recognizing method based on Infrared Thermogram of the embodiment of the present invention, comprises the following steps：
Step one：It is input into the Infrared Thermogram of different running status lower cable connectors.
Step 2：Infrared Thermogram to different running status lower cable connectors carries out Bit Plane Decomposition.Below with input
Different running status lower cable connectors Infrared Thermogram wherein as a example by declarative procedure：
First, by gray processing, Infrared Thermogram is transformed into the gray-scale map that pixel is 320*240, this gray-scale map is designated as, to gray-scale mapBit Plane Decomposition is carried out, can be obtainedIndividual bit plane, by taking wherein one Infrared Thermogram as an example, this reality
Apply example8 are set to, such as accompanying drawing 2 gives 8 bit plans of the figure.Each pixel is corresponding 0 or 1 liang in each bit plane
Individual value.TheIndividual bit plane () inRowIndividual pixel correspondence position, such as formula（1）It is shown
（1）
Single gray-scale map pixel valueCan be by every planar pixelRepresent, such as formula（2）It is shown
（2）
Step 3：The characteristic parameter of each bit plane of Infrared Thermogram is extracted, is divided into the following steps：
1st, initial characteristicses matrix A is set up
（3）
In formula,It is number of training, the present inventionIt is set to 100；It is initial characteristicses dimension, the present inventionSet
It is 16, i.e., 8 bit planes, each bit plane extracts two characteristic parameters, two characteristic parameters are respectively Hu squares and Zernike
Square.The initial value of each element is 0.
2nd, Hu central moments are calculated
For two-dimensional discrete image, i.e., each bit plan, calculate shown in Hu squares such as formula (4)：
(4)
The summation of the existing image object pixel of the zeroth order cubold of image, the quality or area only one of which of object, according to formula
(4), the zeroth order square of image is formula (5)：
(5)
The first moment of image can calculate the barycenter of target, including,, shown in target centroid such as formula (6)：
(6)
Central moment is used as origin analysis by target centroid, with location independence, bit planCentral moment be：
(7)
Normalized central moment is：
(8)
Wherein,.Formula（8）The normalization Hu central moments of as required bit plan.
（3）Calculate Zernike squares
Each bit planP ranks q weight Zernike squares be：
(9)
Wherein,It isConjugation.
(10)
Wherein,ForAxle withAngle,rPoint to the vector length of the origin of coordinates is represented,It is radial polynomial,
Expression formula it is as follows：
(11)
In order to reduce discrete the brought error of infrared thermal imagery, Zernike squares are normalized, after normalization
Shown in Zernike squares such as formula (12)：
(12)
Formula（12）The normalization Zernike squares of as required bit plan.
（4）The characteristic parameter that the above method is calculated is brought into eigenmatrix, the present embodiment is by calculated 100 groups
Characteristic parameter is brought into eigenmatrix A,
Wherein preceding 8 are classified as 100 groups of Hu central moment parameters of the 8 of Infrared Thermogram bit planes, 8 are classified as 100 groups of infrared thermal imageries afterwards
The Zernike square parameters of 8 bit planes of figure.
Step 4：Characteristic parameter to the Infrared Thermogram of different running status lower cable connectors carries out ELM learner instructions
Practice.
Extreme learning machine ELM (Extreme Learning Machine), ELM are a kind of new fast learning algorithms,
For neural networks with single hidden layer, ELM can be random initializtion input weight and biasing and be exported weight accordingly.ELM by
Input layer, hidden layer, output layer is constituted.After the characteristic parameter of infrared thermal imagery substitutes into ELM input layers, hidden layer calculating is carried out, then export
The running status result of identification.
Assuming that the characteristic parameter number of the Infrared Thermogram of training is, the Infrared Thermogram of the present embodiment setting training
Number of samples be 100, i.e.,It is 100, the characteristic parameter parameter matrix of each Infrared ThermogramRepresent,Represent theThe characteristic vector of individual Infrared Thermogram, i.e. parameter matrix OK,。
HaveThe output function of single hidden layer feedforward neural network of individual hidden neuron expresses formula：
(13)
Wherein,It is activation primitive,It is input weight,It is output weight,
It isThe biasing of individual Hidden unit,RepresentWithInner product,, the present invention sets hidden
Layer neuron numberIt is 5.In ELM algorithms, once the biasing of input weight and hidden layer is determined at random, the output square of hidden layer
Battle array is just now uniquely determined.
Step 5：The state recognition model that the cable connector Infrared Thermogram input of collection has been trained, and obtain phase
The cable connector running status result answered, i.e. running status is normal or abnormal.So far, all steps terminate, and complete identification.
Using in the once experiment that the above method is carried out, 50 groups of Infrared Thermograms of collection in worksite are tested in experiment,
The experiment average recognition accuracy of final result reaches more than 85%, realizes to effective identification of cable connector running status and prison
Control.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Ability
The those of ordinary skill in domain should be understood：It can still modify to the technical scheme described in above-described embodiment, or
Equivalent is carried out to which part technical characteristic；And these modifications or replacement, take off the essence of appropriate technical solution
From the spirit and scope of various embodiments of the present invention technical scheme.
Claims (3)
1. a kind of cable connector running rate recognizing method based on Infrared Thermogram, it is characterised in that：Comprise the following steps：
Step one：It is input into the Infrared Thermogram of different running status lower cable connectors；
Step 2：The Infrared Thermogram of the different running status lower cable connectors to being input into carries out Bit Plane Decomposition；
Step 3：Initial characteristicses matrix is set up, the characteristic parameter of each bit plane of Infrared Thermogram is extracted（Hu squares and Zernike
Square）, the characteristic parameter that will be extracted is brought into eigenmatrix；
Step 4：Infrared Thermogram characteristic parameter to different running status lower cable connectors carries out ELM learner training；
Step 5：The state recognition model that the cable connector Infrared Thermogram input of collection has been trained, and obtain corresponding
Cable connector running status result, i.e. running status are normal or abnormal.
2. the cable connector running rate recognizing method based on Infrared Thermogram according to claim 1, it is characterised in that：
In the step 3, the Zernike squares to extracting are normalized.
3. the cable connector running rate recognizing method based on Infrared Thermogram according to claim 1, it is characterised in that：
In the step 2, the Infrared Thermogram of the different running status lower cable connectors being input into is transformed into pixel for 320*240
Gray-scale map, Bit Plane Decomposition is then carried out again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611244033.6A CN106874836A (en) | 2016-12-29 | 2016-12-29 | A kind of cable connector running rate recognizing method based on Infrared Thermogram |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611244033.6A CN106874836A (en) | 2016-12-29 | 2016-12-29 | A kind of cable connector running rate recognizing method based on Infrared Thermogram |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106874836A true CN106874836A (en) | 2017-06-20 |
Family
ID=59165093
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611244033.6A Pending CN106874836A (en) | 2016-12-29 | 2016-12-29 | A kind of cable connector running rate recognizing method based on Infrared Thermogram |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106874836A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111351527A (en) * | 2020-04-01 | 2020-06-30 | 国网湖南省电力有限公司 | Intelligent cable explosion-proof box with pressure and gas monitoring function, system and application method thereof |
| CN111426406A (en) * | 2020-06-02 | 2020-07-17 | 国家电网有限公司 | Cable detection device based on temperature-sensitive color change of cable coating and detection method thereof |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101038668A (en) * | 2006-03-18 | 2007-09-19 | 辽宁师范大学 | Image searching method based on important bit-plane |
| CN201307149Y (en) * | 2008-10-24 | 2009-09-09 | 淄博智洋电气有限公司 | System for monitoring temperature of cable head and automatically positioning line fault of ring main unit |
| CN202904861U (en) * | 2012-11-27 | 2013-04-24 | 宁波电业局 | Temperature collector of electric cable intermediate connector |
| CN103886106A (en) * | 2014-04-14 | 2014-06-25 | 北京工业大学 | Remote sensing image safe-retrieval method based on spectral feature protection |
| CN104777410A (en) * | 2015-04-22 | 2015-07-15 | 东北电力大学 | Partial discharge pattern identification method for crosslinked polyethylene cable |
| CN104809722A (en) * | 2015-04-13 | 2015-07-29 | 国家电网公司 | Electrical device fault diagnosis method based on infrared thermography |
| CN105300528A (en) * | 2015-10-12 | 2016-02-03 | 国家电网公司 | Infrared image diagnosis method and infrared image diagnosis system for transformer station equipment |
| CN105471096A (en) * | 2014-09-12 | 2016-04-06 | 张万生 | Micro-network intelligent monitoring system based on multiple heterogeneous information fusion technology |
-
2016
- 2016-12-29 CN CN201611244033.6A patent/CN106874836A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101038668A (en) * | 2006-03-18 | 2007-09-19 | 辽宁师范大学 | Image searching method based on important bit-plane |
| CN201307149Y (en) * | 2008-10-24 | 2009-09-09 | 淄博智洋电气有限公司 | System for monitoring temperature of cable head and automatically positioning line fault of ring main unit |
| CN202904861U (en) * | 2012-11-27 | 2013-04-24 | 宁波电业局 | Temperature collector of electric cable intermediate connector |
| CN103886106A (en) * | 2014-04-14 | 2014-06-25 | 北京工业大学 | Remote sensing image safe-retrieval method based on spectral feature protection |
| CN105471096A (en) * | 2014-09-12 | 2016-04-06 | 张万生 | Micro-network intelligent monitoring system based on multiple heterogeneous information fusion technology |
| CN104809722A (en) * | 2015-04-13 | 2015-07-29 | 国家电网公司 | Electrical device fault diagnosis method based on infrared thermography |
| CN104777410A (en) * | 2015-04-22 | 2015-07-15 | 东北电力大学 | Partial discharge pattern identification method for crosslinked polyethylene cable |
| CN105300528A (en) * | 2015-10-12 | 2016-02-03 | 国家电网公司 | Infrared image diagnosis method and infrared image diagnosis system for transformer station equipment |
Non-Patent Citations (2)
| Title |
|---|
| HAI-JUN RONG 等: "Aircraft recognition using modular extreme learning machine", 《NEUROCOMPUTING》 * |
| 赵珊 等: "基于位平面分布特征的图像检索算法", 《光子学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111351527A (en) * | 2020-04-01 | 2020-06-30 | 国网湖南省电力有限公司 | Intelligent cable explosion-proof box with pressure and gas monitoring function, system and application method thereof |
| CN111426406A (en) * | 2020-06-02 | 2020-07-17 | 国家电网有限公司 | Cable detection device based on temperature-sensitive color change of cable coating and detection method thereof |
| CN111426406B (en) * | 2020-06-02 | 2021-12-21 | 国家电网有限公司 | Cable detection device based on temperature-sensitive color change of cable coating and detection method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109118479B (en) | Capsule network-based insulator defect identification and positioning device and method | |
| CN111209434B (en) | Substation equipment inspection system and method based on multi-source heterogeneous data fusion | |
| CN111325347B (en) | Automatic danger early warning description generation method based on interpretable visual reasoning model | |
| CN107274451A (en) | Isolator detecting method and device based on shared convolutional neural networks | |
| CN106060008A (en) | Network invasion abnormity detection method | |
| CN112200263B (en) | Self-organizing federal clustering method applied to power distribution internet of things | |
| CN108921230A (en) | Method for diagnosing faults based on class mean value core pivot element analysis and BP neural network | |
| CN111209832A (en) | Auxiliary obstacle avoidance training method, equipment and medium for transformer substation inspection robot | |
| CN113255691A (en) | Method for detecting and identifying harmful bird species target of bird-involved fault of power transmission line | |
| CN115862073A (en) | Transformer substation harmful bird species target detection and identification method based on machine vision | |
| CN106874836A (en) | A kind of cable connector running rate recognizing method based on Infrared Thermogram | |
| Doaei et al. | Fault classification in electrofusion polyethylene joints by combined machine learning, thermal pulsing and IR thermography methods–A comparative study | |
| CN110084109A (en) | A kind of low-resolution face image recognition methods, device, electronic equipment and storage medium | |
| CN116338545A (en) | Method, system, equipment and medium for identifying metering error state of current transformer | |
| CN115859077A (en) | Multi-feature fusion motor small sample fault diagnosis method under variable working conditions | |
| CN115236272A (en) | Gas sensor fault diagnosis method and device under multi-working condition and storage medium | |
| Niu et al. | Implementation of network information security monitoring system based on adaptive deep detection | |
| CN114298144A (en) | Hydropower station equipment oil leakage identification method based on deep convolutional neural network | |
| Sharma et al. | Preventing car damage using CNN and computer vision | |
| CN112200030A (en) | Power system field operation action risk identification method based on graph convolution | |
| WANG et al. | Design and Implementation of Infrared Image Classification Algorithm for Defective Power Equipment Based on Deep Learning. | |
| CN116503398B (en) | Insulator pollution flashover detection method and device, electronic equipment and storage medium | |
| Azari et al. | Intelligent fault detection in power distribution systems using thermos-grams by ensemble classifiers | |
| Zhou et al. | Fault Diagnosis of Transmission Lines Based on Sketch Retrieval for Small Targets and Small Samples | |
| CN113487252B (en) | Enterprise safety early warning, prevention and control method and system based on big data analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170620 |