CN103268501A - Image recognition method for intelligent transmission line patrol - Google Patents
Image recognition method for intelligent transmission line patrol Download PDFInfo
- Publication number
- CN103268501A CN103268501A CN2013102181026A CN201310218102A CN103268501A CN 103268501 A CN103268501 A CN 103268501A CN 2013102181026 A CN2013102181026 A CN 2013102181026A CN 201310218102 A CN201310218102 A CN 201310218102A CN 103268501 A CN103268501 A CN 103268501A
- Authority
- CN
- China
- Prior art keywords
- picture
- image
- transmission line
- electricity
- storehouse
- 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
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of remote monitoring, and discloses an image recognition method for intelligent transmission line patrol. The image recognition method comprises the following steps of: building a characteristic vector library, and using a classifier to carry out the initial training on the characteristic vector library, so that a classification surface is trained; and then judging the image characteristics of pictures collected in the line patrol process by classification through the classification surface, if the image characteristics of the pictures collected in the line patrol process are classified as a class for expressing defects, then judging that the pictures collected in the line patrol process have the defects, i.e., judging that a transmission line fails. Through the method provided by the invention, the defects of the collected pictures are recognized, so that the fault recognition of each picture by naked eyes is avoided, the intelligent transmission line patrol is realized, and the workload of workers is greatly reduced.
Description
Technical field
The present invention relates to the Remote Monitoring Technology field, particularly a kind of image-recognizing method for transmission line of electricity intelligence line walking.
Background technology
In recent years, along with the development of electric network composition and perfect, the construction of high pressure and extra high voltage network had obtained fast development.Because transmission line of electricity is exposed to the field all the year round, suffer the invasion and attack of natural weathers such as wind, snow, rain, mist, suffer the threat of disasteies such as flood, earthquake, rubble flow, also will suffer mountain stone exploitation, construction blasting, steal the damage of human factors such as breaking-up, the safe operation of transmission line of electricity is subjected to serious threat.It is the inevitable means that ensure the transmission line of electricity safe operation that transmission line of electricity is carried out running status and surrounding environment change tour (discovering device fault and the hidden danger that jeopardizes transmission line of electricity safety) and repair and maintenance.
Monitoring to transmission line of electricity at present mainly contains artificial line walking and helicopter line walking dual mode.Transmission line of electricity is long and distributional environment is complicated, and artificial line walking difficulty is big, the cycle is long.Helicopter line walking difficulty is less relatively, but expense height, cycle are long.In addition, artificial line walking and helicopter line walking are subjected to inside even from weather bigger, and inclement weather can't carry out the line walking monitoring, does not have ageing.In recent years, the development of the communication technology and sensing technology made the online monitoring of transmission line of electricity become possibility.The online monitoring of transmission line of electricity is to gather monitoring photo or video by electronic equipment, monitoring photo or video transfer to the monitoring backstage by transmission network again, and the staff on monitoring backstage judges by visual inspection monitoring photo or video whether transmission line of electricity breaks down.Though this on-line monitoring does not need artificial or helicopter carries out line walking, still needs manually to carry out fault distinguishing, workload is big, does not realize the intelligent line walking of transmission line of electricity fully.
Summary of the invention
The objective of the invention is to overcome the deficiency that existing needs in the prior art manually carry out fault distinguishing, a kind of image-recognizing method for transmission line of electricity intelligence line walking is provided, can realize that by the inventive method fault differentiates automatically, need not manually to carry out again fault and judge, realize transmission line of electricity intelligence line walking.
In order to realize the foregoing invention purpose, the invention provides following technical scheme:
A kind of image-recognizing method for transmission line of electricity intelligence line walking comprises and sets up proper vector storehouse step and picture defect recognition step, wherein,
Set up proper vector storehouse step: at first import samples pictures, described samples pictures comprises transmission line of electricity break down the defective picture under the situation of normal picture under the situation and transmission line of electricity that do not break down; Remove the noise in the samples pictures then; Extract the characteristics of image of every samples pictures at last, and the characteristics of image that all extract from samples pictures is stored in the proper vector storehouse, finish the foundation in proper vector storehouse;
Picture defect recognition step: at first adopt sorter that initial training is carried out in the proper vector storehouse, train classifying face; Import the picture of gathering in the line walking process then, and from the picture of gathering, extract characteristics of image; Then with in the characteristics of image input sorter that extracts, by classifying face judgements of classifying: if the characteristics of image of the picture of gathering is classified in a class of expressing defective, judges that then there is defective in the picture of gathering, otherwise judge the picture zero defect of collection; It is that transmission line of electricity breaks down that there is the implication of defective in the picture of described collection.
Preferably, in the described picture defect recognition step, described sorter is the svm classifier device.
Preferably, above-mentioned image-recognizing method for transmission line of electricity intelligence line walking also comprises the incremental learning step, and the incremental learning step comprises: the support vector that obtains in the storehouse initial training of storage proper vector, composition characteristic storehouse; The picture input svm classifier device of gathering in the line walking process is classified after the judgement, identify whether classification results is correct: if classification error, whether the characteristics of image of then further judging the picture of gathering belongs to support vector, belong to support vector and then this characteristics of image is added into feature database, again feature database is trained, obtain new classifying face.
In the classification deterministic process, if classifying face immobilizes, the characteristics of image that is classified the new picture near or the classifying face of classifying face so may be judged by accident, and then causes new picture defect recognition inaccurate.In the classification deterministic process, carry out incremental learning, optimize classifying face, can ensure the correctness that classification is judged.
Compared with prior art, beneficial effect of the present invention:
The present invention is used for the image-recognizing method of transmission line of electricity intelligence line walking, at first set up the characteristics of image storehouse, the characteristics of image of the picture that the line walking process is collected and the characteristics of image in the characteristics of image storehouse comparison of classifying then, if the characteristics of image of the picture that the line walking process collects is classified in a class of expressing defective, judge that then there is defective in the picture that the line walking process collects, judge that namely transmission line of electricity breaks down.By the inventive method the picture that collects is carried out defect recognition, has avoided the artificial naked eyes that pass through that each pictures is carried out Fault Identification, realized transmission line of electricity intelligence line walking completely, greatly the minimizing of degree staff's workload.
The inventive method also comprises the incremental learning step, continues to optimize classifying face in the classification deterministic process, has further improved the accuracy of picture defect recognition.
Description of drawings:
Fig. 1 is used for the process flow diagram of the image-recognizing method of transmission line of electricity intelligence line walking for the present invention.
Fig. 2 is used for the process flow diagram of the image-recognizing method incremental learning step of transmission line of electricity intelligence line walking for the present invention.
Fig. 3 represents synoptic diagram for 4 chain code shapes of insulator.
Embodiment
The present invention is described in further detail below in conjunction with testing example and embodiment.But this should be interpreted as that the scope of the above-mentioned theme of the present invention only limits to following embodiment, all technology that realizes based on content of the present invention all belong to scope of the present invention.
A kind of image-recognizing method for transmission line of electricity intelligence line walking that present embodiment is enumerated is at first set up the proper vector storehouse, and is adopted the svm classifier device that initial training is carried out in the proper vector storehouse, obtains classifying face; Then the picture that collects is carried out defect recognition, namely, with the characteristics of image of the picture that collects by the classifying face judgement of classifying, if the characteristics of image of the picture that collects is classified in a class of expressing defective, judge that then there is defective in the picture that collects, it is that transmission line of electricity breaks down that there is the implication of defective in the picture that collects, comprises breakdown of conducting wires, as on disconnected thigh, broken lot, intersection, the lead hanger being arranged; The iron tower fault is as damaged, disappearance; Insulator breakdown breaks as damaged, thunderbolt vestige, full skirt; Passage and surrounding enviroment are unusual, as construction, danger stacking etc.
With reference to figure 1, set up in the step of proper vector storehouse, at first import samples pictures, samples pictures comprises that any fault does not take place transmission line of electricity (is that the lead of expressing in the samples pictures, an iron rake with three to six teeth, insulator are all without any fault, passage and surrounding enviroment are also without any unusual) normal picture under the situation and the transmission line of electricity defective picture under the situation that breaks down, adopt image denoising algorithm or algorithm for image enhancement that samples pictures is carried out pre-service then, remove some noises in the samples pictures, making needs the feature of extraction more outstanding.And then extract the characteristics of image of every samples pictures, and with the characteristics of image storage of extracting, composition characteristic vector storehouse.For example, extract the characteristics of image of lead, at first carry out the Hough change detection and go out straight line, carry out the sift change detection at straight line again and go out proper vector, and characteristics of image is stored as the proper vector in the proper vector storehouse.For another example, adopt the shape counting method to extract the characteristics of image of insulator.The shape number is based on a kind of boundary shape of chain code and describes, and is a kind of normalized differential code in essence.The shape number has translation, rotation and yardstick unchangeability, and for a given border, its shape has uniqueness.As shown in Figure 3, represent synoptic diagram for 4 chain code shapes of insulator.From limit, left side lowermost end, chain code is 1110303232(1), the first order difference sign indicating number is 0033133130, its shape number is 0003313313.The characteristics of image that will extract from all samples pictures is stored in the proper vector storehouse at last, finishes the foundation in proper vector storehouse.
The picture that collects is carried out at first importing new picture in the defect recognition step, and described new picture is the picture that collects in the transmission line of electricity intelligence line walking process.Extract the characteristics of image of new picture then.In the characteristics of image that will from new picture, extract the again input sorter, by the classifying face that obtains in the initial training of the proper vector storehouse judgement of classifying.Sorter adopts the SVM(support vector machine), SVM needs a spot of sample namely to can converge to global optimum, and computing velocity is very fast.SVM is divided into two classes with proper vectors all in the proper vector storehouse, two category feature vectors are classified face, and (classifying face is not a plane, a but lineoid on the N dimension space) separates, the characteristics of image of the defective picture in the samples pictures is divided into a class, namely express a class of defective, the characteristics of image of normal picture is divided into another kind of.After the characteristics of image input svm classifier device of new picture, judge whether the characteristics of image of new picture is classified in a class of expressing defective.If the characteristics of image of new picture is classified in a class of expressing defective, judge that then there is defective in new picture, judge that namely there is fault in transmission line of electricity, otherwise judge that new picture does not have defective that namely any fault does not take place transmission line of electricity.
In the svm classifier deterministic process, if classifying face immobilizes, the characteristics of image that is classified the new picture near or the classifying face of classifying face so may be judged by accident, and then causes new picture defect recognition inaccurate.Therefore the accuracy in order to ensure that classification is judged as a kind of optimal way, is carried out incremental learning in the classification deterministic process, continues to optimize classifying face.
With reference to figure 2, in the incremental learning process, the support vector that obtains in the storehouse initial training of storage proper vector is (in the svm classifier deterministic process, satisfy the proper vector of KKT constraint, namely satisfy Y (i) * f (Xi) 〉=proper vector of 1 condition is support vector, classification under Y (i) the expression sample, f (Xi) presentation class surface function), the composition characteristic storehouse; The picture gathered in line walking process input svm classifier device classify judge after, identify manually whether classification results correct, if classification is correct, then carries out the classification of next pictures and judge.If classification error judges further then whether the characteristics of image of the picture of gathering in the line walking process belongs to support vector, if belong to support vector, then this characteristics of image is added into feature database, again feature database is trained, obtain new classifying face.If the characteristics of image of the picture of gathering in the line walking process does not belong to support vector, then do not add feature database.
Each pictures that collects in the line walking process classified carry out incremental learning after judging, the characteristics of image of the wrongheaded picture of classification is carried out the support vector judgement, judge whether characteristics of image belongs to support vector, and the characteristics of image that will belong to support vector adds the proper vector storehouse, again train, obtain new classifying face, continue to optimize classifying face like this, can further improve the accuracy that classification is judged.Classifying face is through continuing to optimize, and the error rate that follow-up classification is judged will be more and more lower, even inerrancy.
The equipment (for example insulator) that is used on the transmission line of electricity might be replaced with the new product different with original structure, carry out incremental learning, the characteristics of image of new product is added into the sample storehouse, when the new picture of gathering relates to new product, just can carries out correct defect recognition.
Above-described embodiment is most preferred embodiment of the present invention, and is not used in the restriction scope of the invention.
Claims (3)
1. an image-recognizing method that is used for transmission line of electricity intelligence line walking is characterized in that, comprise and set up proper vector storehouse step and picture defect recognition step, wherein,
Set up proper vector storehouse step: at first import samples pictures, described samples pictures comprises transmission line of electricity break down the defective picture under the situation of normal picture under the situation and transmission line of electricity that do not break down; Remove the noise in the samples pictures then; Extract the characteristics of image of every samples pictures at last, and the characteristics of image that all extract from samples pictures is stored in the proper vector storehouse, finish the foundation in proper vector storehouse;
Picture defect recognition step: at first adopt sorter that initial training is carried out in the proper vector storehouse, train classifying face; Import the picture of gathering in the line walking process then, and from the picture of gathering, extract characteristics of image; Then with in the characteristics of image input sorter that extracts, by classifying face judgements of classifying: if the characteristics of image of the picture of gathering is classified in a class of expressing defective, judges that then there is defective in the picture of gathering, otherwise judge the picture zero defect of collection; It is that transmission line of electricity breaks down that there is the implication of defective in the picture of described collection.
2. the image-recognizing method for transmission line of electricity intelligence line walking according to claim 1 is characterized in that in the described picture defect recognition step, described sorter is the svm classifier device.
3. the image-recognizing method for transmission line of electricity intelligence line walking according to claim 2 is characterized in that also comprise the incremental learning step, described incremental learning step is: the support vector that obtains in the storehouse initial training of storage proper vector, composition characteristic storehouse; The picture input svm classifier device of gathering in the line walking process is classified after the judgement, identify whether classification results is correct: if classification error, whether the characteristics of image of then further judging the picture of gathering belongs to support vector, belong to support vector and then this characteristics of image is added into feature database, again feature database is trained, obtain new classifying face.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102181026A CN103268501A (en) | 2013-06-04 | 2013-06-04 | Image recognition method for intelligent transmission line patrol |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102181026A CN103268501A (en) | 2013-06-04 | 2013-06-04 | Image recognition method for intelligent transmission line patrol |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103268501A true CN103268501A (en) | 2013-08-28 |
Family
ID=49012128
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013102181026A Pending CN103268501A (en) | 2013-06-04 | 2013-06-04 | Image recognition method for intelligent transmission line patrol |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103268501A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105447530A (en) * | 2016-01-05 | 2016-03-30 | 国网四川省电力公司检修公司 | Power transmission line hidden risk and fault detection method based on image identification technology |
| CN105718842A (en) * | 2014-12-02 | 2016-06-29 | 中国科学院沈阳自动化研究所 | Machine vision-based detection method for transmission line strand breakage fault |
| CN106355187A (en) * | 2016-09-07 | 2017-01-25 | 西华大学 | Application of visual information to electrical equipment monitoring |
| CN106803290A (en) * | 2017-01-16 | 2017-06-06 | 国网山东省电力公司龙口市供电公司 | Transmission line makes an inspection tour recording method and device |
| CN106877237A (en) * | 2017-03-16 | 2017-06-20 | 天津大学 | A kind of method of insulator missing in detection transmission line of electricity based on Aerial Images |
| CN109239550A (en) * | 2018-09-04 | 2019-01-18 | 国网山东省电力公司青岛供电公司 | A kind of line insulation situation judgment method |
| CN109489629A (en) * | 2018-12-07 | 2019-03-19 | 国网四川省电力公司电力科学研究院 | A kind of safety monitoring method of electric power line pole tower |
| CN111583196A (en) * | 2020-04-22 | 2020-08-25 | 北京智芯微电子科技有限公司 | Monitoring system and monitoring method for power transmission line |
| CN111654106A (en) * | 2020-06-11 | 2020-09-11 | 国家电网有限公司华东分部 | Power grid dispatching system based on image recognition technology in deep learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102005755A (en) * | 2010-10-29 | 2011-04-06 | 昆明理工大学 | Extra-high voltage direct current transmission line boundary element forming method based on support vector machine |
| CN103091603A (en) * | 2013-01-14 | 2013-05-08 | 华北电力大学 | Breakdown intelligent classification and positioning method of electric transmission line |
-
2013
- 2013-06-04 CN CN2013102181026A patent/CN103268501A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102005755A (en) * | 2010-10-29 | 2011-04-06 | 昆明理工大学 | Extra-high voltage direct current transmission line boundary element forming method based on support vector machine |
| CN103091603A (en) * | 2013-01-14 | 2013-05-08 | 华北电力大学 | Breakdown intelligent classification and positioning method of electric transmission line |
Non-Patent Citations (5)
| Title |
|---|
| JIAN HUANG EL AT.: "《Supoort vector machine with genetic algorithm for machinery fault diagnosis of high voltage circuit breaker》", 《ELSEVIER：MEASUREMENT》, 31 December 2011 (2011-12-31), pages 1018 - 1027 * |
| 刘坐乾: "《基于改进的支持向量机的模拟电路故障诊断方法研究》", 《中国优秀硕士学位论文全文数据库（电子期刊）》, 30 June 2012 (2012-06-30) * |
| 柴毅等: "《半监督增量式SVM 在故障诊断中的应用#》", 《中国科技论文在线》, 14 February 2012 (2012-02-14) * |
| 王华忠等: "《基于支持向量机的故障诊断方法》", 《华东理工大学学报》, vol. 30, 29 February 2004 (2004-02-29), pages 179 - 182 * |
| 王延峰: "《SVM增量学习算法研究》", 《黑龙江科技信息》, no. 9, 31 December 2012 (2012-12-31), pages 2 - 3 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105718842A (en) * | 2014-12-02 | 2016-06-29 | 中国科学院沈阳自动化研究所 | Machine vision-based detection method for transmission line strand breakage fault |
| CN105718842B (en) * | 2014-12-02 | 2018-12-07 | 中国科学院沈阳自动化研究所 | A kind of machine vision detection method of power transmission line line-broken malfunction |
| CN105447530A (en) * | 2016-01-05 | 2016-03-30 | 国网四川省电力公司检修公司 | Power transmission line hidden risk and fault detection method based on image identification technology |
| CN106355187A (en) * | 2016-09-07 | 2017-01-25 | 西华大学 | Application of visual information to electrical equipment monitoring |
| CN106803290A (en) * | 2017-01-16 | 2017-06-06 | 国网山东省电力公司龙口市供电公司 | Transmission line makes an inspection tour recording method and device |
| CN106877237A (en) * | 2017-03-16 | 2017-06-20 | 天津大学 | A kind of method of insulator missing in detection transmission line of electricity based on Aerial Images |
| CN109239550A (en) * | 2018-09-04 | 2019-01-18 | 国网山东省电力公司青岛供电公司 | A kind of line insulation situation judgment method |
| CN109239550B (en) * | 2018-09-04 | 2020-11-27 | 国网山东省电力公司青岛供电公司 | Line insulation condition judgment method |
| CN109489629A (en) * | 2018-12-07 | 2019-03-19 | 国网四川省电力公司电力科学研究院 | A kind of safety monitoring method of electric power line pole tower |
| CN111583196A (en) * | 2020-04-22 | 2020-08-25 | 北京智芯微电子科技有限公司 | Monitoring system and monitoring method for power transmission line |
| CN111583196B (en) * | 2020-04-22 | 2021-09-07 | 北京智芯微电子科技有限公司 | Monitoring system and monitoring method for power transmission line |
| CN111654106A (en) * | 2020-06-11 | 2020-09-11 | 国家电网有限公司华东分部 | Power grid dispatching system based on image recognition technology in deep learning |
| CN111654106B (en) * | 2020-06-11 | 2021-08-24 | 国家电网有限公司华东分部 | Power grid dispatching system based on image recognition technology in deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103268501A (en) | Image recognition method for intelligent transmission line patrol | |
| WO2022077605A1 (en) | Wind turbine blade image-based damage detection and localization method | |
| CN103260016B (en) | A kind of transmission line long-distance Intelligent line patrolling method | |
| CN102590823B (en) | Rapid extraction and reconstruction method for data power line of airborne LIDAR | |
| CN105809679B (en) | Mountain railway side slope rockfall detection method based on visual analysis | |
| CN108037133A (en) | A kind of power equipments defect intelligent identification Method and its system based on unmanned plane inspection image | |
| Li et al. | Sewer pipe defect detection via deep learning with local and global feature fusion | |
| Wu et al. | Detection of bird nests in overhead catenary system images for high-speed rail | |
| CN109359697A (en) | Graph image recognition methods and inspection system used in a kind of power equipment inspection | |
| CN107276805A (en) | A kind of sample predictions method, device and electronic equipment based on IDS Framework | |
| CN106355580B (en) | Method and device for detecting toppling of tower | |
| CN105574165B (en) | A kind of grid operating monitoring information identification classification method based on cluster | |
| CN106447227A (en) | Urban power grid abnormal state analyzing method and system | |
| CN110136131A (en) | A kind of zero resistance insulator detection method based on deep learning | |
| CN106326808A (en) | Method for detecting bird nests in power transmission line poles based on unmanned plane images | |
| CN105303162A (en) | Target proposed algorithm-based insulator recognition algorithm for aerial images | |
| CN111695506B (en) | Early warning method and system for wind-induced foreign matter short-circuit fault of transmission line | |
| CN111145159B (en) | Method and device for extracting inspection key component points | |
| CN109460787A (en) | IDS Framework method for building up, device and data processing equipment | |
| CN109919921A (en) | Based on the influence degree modeling method for generating confrontation network | |
| CN106250920A (en) | The insulator state detection merged based on multicharacteristic information and diagnostic method | |
| Song et al. | Intrusion detection of foreign objects in high-voltage lines based on YOLOv4 | |
| CN113298077A (en) | Transformer substation foreign matter identification and positioning method and device based on deep learning | |
| CN105528790A (en) | Transmission line small part identification method | |
| CN106569098A (en) | Power transmission and distribution line safety patrol inspection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 610041, 30, 1, 3003, 138 Tianfu two street, Chengdu hi tech Zone, Sichuan, China Applicant after: Chengdu Si Han Science and Technology Co., Ltd. Address before: West high tech Zone Fucheng Road in Chengdu city of Sichuan province 610041 No. 399 Tianfu New Valley No. 5 Building 8 floor Applicant before: Chengdu Sihan Technology Co., Ltd. |
|
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20130828 |
|
| RJ01 | Rejection of invention patent application after publication |