CN101833673A - Electric power switchgear switch state image recognition system - Google Patents
Electric power switchgear switch state image recognition system Download PDFInfo
- Publication number
- CN101833673A CN101833673A CN 201010174413 CN201010174413A CN101833673A CN 101833673 A CN101833673 A CN 101833673A CN 201010174413 CN201010174413 CN 201010174413 CN 201010174413 A CN201010174413 A CN 201010174413A CN 101833673 A CN101833673 A CN 101833673A
- Authority
- CN
- China
- Prior art keywords
- image
- switch
- state
- module
- gray
- 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
Abstract
The invention relates to an electric power switchgear switch state image recognition system comprising an off-line training module and an on-line detecting module. The off-line training module collects and trains a large number of switch samples to form a plurality of strong classifiers, and the strong classifiers are loaded in the on-line detecting module. The on-line detecting module transforms the collected switch images to switch gray images and denoises, utilizes the strong classifiers to respectively scan and search the gray images, and obtains the positions and state information of the switches. The invention can realize that the image collection, processing and recognition processes are completed in the on-line detecting module, and overcomes the dependence to communication bandwidth and equipment because videos collected by the traditional video monitoring systems are transmitted to remote background systems through a wired or wireless mode. The invention can automatically determine the positions and the states of the switches in the images, does not need manual operation, has high degree of automation, and is suitable for state recognition of various types of electric power switches.
Description
Technical field
The present invention relates to the power switch cabinet telemetry system of power matching network monitoring field, be specifically related to Flame Image Process, recognition system the real-time monitoring of electric power switchgear switch state based on image recognition.
Background technology
In electric system, switch cubicle is used very extensive, can indicate the power circuit running status in real time, to getting rid of power failure certain help is arranged.But because the electric system distributed areas are wide, when breaking down, can't at which place break down, can only check electric power machine room switch cabinet state, expend time in very much and manpower inefficiency by one place, an artificial place the very first time personnel that report.
At present if the state of long-range read switch cabinet will be connected to processor or controller unit by field bus technique with on-off circuit, read the state of power circuit with processor or controller, and come a state that reads is sent to long-range control center by wired or wireless mode.This implementation all comes with some shortcomings in reliability and dirigibility.And existing video monitoring system, terminal only plays the effect of gathering video usually, and image also needs to be transferred to far-end control center by wired or wireless mode, further carries out background process identification.Need higher communication bandwidth like this, require high equipment cost.
Existing existing power switch cabinet switch on off state realtime graphic recognition device adopts gray scale length of stroke statistic law to realize state recognition, this algorithm is simple, add and confine the position but need manually carry out target, automaticity is lower, and is difficult for expansion and is applicable to the other types power switch.
Summary of the invention
Technical matters to be solved by this invention is the deficiency at present popular power failure supervisory system, a kind of electric power switchgear switch state real time image collection processing and identification system has been proposed, it can realize the remote live of electric power switchgear switch state is monitored automatically, need not manually to carry out target localization, the automaticity height is applicable to all kinds power switch.
For solving the problems of the technologies described above, the present invention proposes a kind of electric power switchgear switch state realtime graphic recognition system, described image identification system comprises off-line training module and ONLINE RECOGNITION module, is applied in the realtime graphic processing and identification device of electric power switchgear switch state; Described off-line training module, based on the Adaboost algorithm the true sample and the dummy copy that comprise on off state are in a large number learnt, form a series of Weak Classifier, according to weight these Weak Classifiers are cascaded into some strong classifiers then, described strong classifier is represented different on off states respectively;
Described online detection module comprises image rough handling module and state recognition module;
Described image rough handling module, the switch image transitions that is used for collecting is switch gray-scale map and denoising, so that the state recognition module is handled;
Described state recognition module has been loaded described strong classifier, and utilizes described strong classifier respectively described switch gray-scale map to be carried out scanning search, obtains the position and the status information of switch.
The view data that collects is handled in real time, adopted the algorithm of target detection based on Adaboost, the sorter with three kinds of on off states detects image respectively, draws the on off state conclusion that comprises in the present image.
Preferably, in described off-line training module, the employed feature of each specific Weak Classifier defines with position in shape, the area-of-interest and scale-up factor; When extracting rectangular characteristic, reduce calculated amount with the integral image method.
Preferably, described image rough handling module comprises and is converted into gray-scale map unit and image denoising unit.Described conversion gray-scale map unit is converted into gray level image with the color switching image that collects, and described image denoising unit carries out medium filtering to gray level image, and the noise point that exists in the gray scale is removed the interference of having avoided noise spot that image recognition is brought.
Preferably, the state recognition module adopts the algorithm of target detection based on Adaboost, Da Xiao scanning window is searched for image in varing proportions, can be by the target that is of strong classifier, recognition unit by this target regional location and the kind of strong classifier to determine the state of certain switch.
Adopt the present invention, come the read switch state, and the on off state information that can cooperate video monitoring system will discern after the processing is sent to background system in real time by the distribution communication network, reach the purpose of remote monitoring by the mode of video monitoring and image recognition.The present invention can judge the position and the state of switch in the image voluntarily, need not manually-operated, and automaticity is higher; Also can combine the state of judging complicated switch with existing artificial selection method.The off-line training module that the present invention is only big with calculated amount is finished in far-end control center, and all processing and identification processes are all finished in end device, only need the on off state information after the processing and identification is sent in the far-end staff hand, also can transmit real-time video according to the background system needs.Overcome existing video monitoring system, terminal only plays gathers video and is transferred to far-end control center by wired or wireless mode, further carries out background process identification and the dependence to communication bandwidth and equipment cost of generation.
Simultaneously because the image recognition mode is adopted in monitoring, used camera switch cubicle at a distance it is monitored and the power equipment distance distant, between without any connection, thereby can guarantee system reliability.Supervisory system can be set up separately in addition, and the position can freely be selected, and the equipment installation and maintenance is all very flexible, and can not produce between the power equipment and influence each other, and installation and maintenance personnel's safety also can be fully protected.
Description of drawings
Below in conjunction with the drawings and specific embodiments technical scheme of the present invention is further described in detail.
Fig. 1 is the structured flowchart of power matching network system switching cabinet on off state image identification system.
Fig. 2 is three kinds of on off state original graph of switch cubicle disk rotary-type switch, conducting state (a), ground state (b), vacant state (c).
Fig. 3 is a sorter characteristic rectangle synoptic diagram.
Fig. 4 is the synoptic diagram of calculated product partial image.
Fig. 5 is the on off state in the intercepting a certain moment of switch cubicle, gets gradation of image figure.
Fig. 6 is the figure as a result that the switch state detector of application algorithm of the present invention obtains.
Embodiment
Be example with power matching network system switching cabinet disk rotary-type switch below, the specific embodiment of the present invention is described.
As Fig. 1 is the structured flowchart of electric power switchgear switch state image recognition system.Totally can be off-line training module 1 and online detection module 3.Off-line training module 1 is trained in the far-end background system and is formed the strong classifier 2 that comprises three kinds of on off state features respectively, and in the station terminal, deposition attitude identification module 5 uses before being stored in.The on off state image is sent to image rough handling unit 4 on the switch cubicle that collects, and after gray-scale map conversion and denoising, state recognition module 5 utilizes cascade classifier 2 respectively image to be searched for, and judges the state output of switch.
The picture in kind of power switch cabinet switch as shown in Figure 2, the disk rotary-type switch can present three kinds of different on off states at the different operating state: conducting state (a), ground state (b), vacant state (c).
The off-line training module is by the study to a large amount of true samples and dummy copy, forms a series of Weak Classifier, according to weight these Weak Classifiers is cascaded into strong classifier then.Theoretical proof, as long as each Weak Classifier classification capacity is than conjecture is good at random, when the Weak Classifier number trends towards when infinite, the error rate of strong classifier will be tending towards 0.Training method is, a given training set (x1, y1) ..., (xL, yL), wherein, xi is the training sample of input, yi is that the class formative of classification is true and false sample.When initialization, all training samples are all composed with an identical weight, then with carrying out the training of T wheel to training sample set by weak learning algorithm.After each takes turns the training end, from several simple classification devices, select that of least error, as a Weak Classifier hi, and the sample of failure to train composed with bigger weight, so that allow learning algorithm mainly difficult training sample be learnt in study afterwards.Like this, just can obtain a Weak Classifier sequence (h1, h2 ..., ht), wherein, the reasonable weight of classifying quality is bigger.Final classification function f (x) adopts a kind ofly has the ballot mode of weight to produce, and is about to a plurality of Weak Classifiers and stacks up by certain method and be combined into a strong classifier, promptly
The employed feature of each specific classification device defines with position in shape, the area-of-interest and scale-up factor.The value of rectangular characteristic is meant the difference of inner all the grey scale pixel value sums of two or more rectangles on the image.As shown in Figure 3, wherein 3 (a) are the positive samples pictures of training usefulness, 3 (b) are rectangular characteristic of a kind of on off state of extracting, its representative be the pixel grey scale in black rectangle zone the gray scale of white portion is big up and down than it, corresponding to the switch disk and indicate the white line part up and down.
In the process of extracting rectangular characteristic, can use the method for integral image to reduce calculated amount.In the process of extracting rectangular characteristic, can use the method for integral image to reduce calculated amount.As Fig. 4, the some x, the integral image values at y place be its top and all pixels of the left side with.Like this by the integral image method, the pixel in the rectangle and can calculating arbitrarily by the value on its 4 summits.Therefore whole training process only need scan former figure one time.
At three kinds of different conditions of switch, train three sorters respectively, the two states in addition outside the current state all should participate in training as dummy copy, can improve the discrimination of algorithm like this.
Before the image that collects is discerned, need carry out pre-service to image.The quality of pretreating effect is the principal element that influences the whole test system performance, mainly is to be converted into gray-scale map and denoising.At first truncated picture is converted into gray-scale map by being converted into the gray-scale map unit, as shown in Figure 5, so that the identification of the Flame Image Process of back.
Then, by state recognition unit the image that is converted into gray-scale map is carried out state recognition.State recognition adopts three sorters of the characteristic of switch that contains different conditions respectively that has trained to find the rectangular area that comprises target in image, and judges on off state by the respective classified device.The recognition unit scanning window of size is in varing proportions searched for several times to image.In each scanning, when the rectangle frame of analyzing return during all by each layer of cascade classifier on the occasion of, representing this is a candidate target.After handling and collecting candidate's square frame, then average rectangle in the enough big combination of a series of numbers is made up and returned in these zones as detected target area.Can obtain the positional information of switch in image with this, and judge on off state by the respective classified device, the sorter that even detects switch is a conducting state, and then this switch should be in conducting state.
The on off state information that obtains can pass in the background system by communicator, with for further analysis.As Fig. 6 is to use the figure as a result that the switch state detector of algorithm of the present invention obtains, and what its training and examination phase all used is the analog switch model.
It should be noted last that, above embodiment is only unrestricted in order to technical scheme of the present invention to be described, although the present invention is had been described in detail with reference to preferred embodiment, those of ordinary skill in the art is to be understood that, can make amendment or be equal to replacement technical scheme of the present invention, and not breaking away from the spirit and scope of technical solution of the present invention, it all should be encompassed in the middle of the claim scope of the present invention.
Claims (4)
1. an electric power switchgear switch state image recognition system is characterized in that, described image identification system comprises off-line training module and online detection module;
Described off-line training module, based on the Adaboost algorithm the true sample and the dummy copy that comprise on off state are in a large number learnt, form a series of Weak Classifier, according to weight these Weak Classifiers are cascaded into some strong classifiers then, described strong classifier is represented different on off states respectively;
Described online detection module comprises image rough handling module and state recognition module;
Described image rough handling module, the switch image transitions that is used for collecting is switch gray-scale map and denoising, so that the state recognition module is handled;
Described state recognition module has been loaded described strong classifier, and utilizes described strong classifier respectively described switch gray-scale map to be carried out scanning search, obtains the position and the status information of switch.
2. electric power switchgear switch state image recognition system according to claim 1, it is characterized in that, in described off-line training module, the employed feature of each described specific Weak Classifier defines with position in shape, the area-of-interest and scale-up factor; When extracting rectangular characteristic, reduce calculated amount with the integral image method.
3. electric power switchgear switch state image recognition system according to claim 1 and 2, it is characterized in that, described image rough handling module comprises and is converted into gray-scale map unit and image denoising unit, described conversion gray-scale map unit is converted into gray level image with the color switching image that collects, described image denoising unit carries out medium filtering to gray level image, the noise point that exists in the gray scale is removed the interference of having avoided noise spot that image recognition is brought.
4. electric power switchgear switch state image recognition system according to claim 3, it is characterized in that, the state recognition module adopts the algorithm of target detection based on Adaboost, Da Xiao scanning window is searched for image in varing proportions, can be by the target that is of strong classifier, recognition unit by this target regional location and the kind of strong classifier to determine the state of certain switch.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010174413 CN101833673A (en) | 2010-05-18 | 2010-05-18 | Electric power switchgear switch state image recognition system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201010174413 CN101833673A (en) | 2010-05-18 | 2010-05-18 | Electric power switchgear switch state image recognition system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101833673A true CN101833673A (en) | 2010-09-15 |
Family
ID=42717736
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201010174413 Pending CN101833673A (en) | 2010-05-18 | 2010-05-18 | Electric power switchgear switch state image recognition system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101833673A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104103044A (en) * | 2014-07-09 | 2014-10-15 | 上海电力学院 | Tackle cable jumping on-line detecting method based on K-Mean algorithm |
| TWI459321B (en) * | 2011-12-29 | 2014-11-01 | ||
| CN104200219A (en) * | 2014-08-20 | 2014-12-10 | 深圳供电局有限公司 | Method and device for automatically identifying substation breaker and switch indicator |
| CN104331710A (en) * | 2014-11-19 | 2015-02-04 | 集美大学 | On-off state recognition system |
| CN105047450A (en) * | 2015-08-31 | 2015-11-11 | 中国西电电气股份有限公司 | High-voltage switch position indication interlocking system and method based on image recognition technology |
| CN105869164A (en) * | 2016-03-28 | 2016-08-17 | 国网浙江省电力公司宁波供电公司 | Method and system for detecting on/off state of switch |
| CN106339722A (en) * | 2016-08-25 | 2017-01-18 | 国网浙江省电力公司杭州供电公司 | Line knife switch state monitoring method and device |
| CN106570865A (en) * | 2016-11-08 | 2017-04-19 | 国家电网公司 | Digital-image-processing-based switch state detecting system of power equipment |
| CN106971182A (en) * | 2017-02-06 | 2017-07-21 | 王兴照 | Embedded electric power relay pressing plate is thrown and moves back state Intelligent Identify device and implementation method |
| CN111950606A (en) * | 2020-07-28 | 2020-11-17 | 北京恒通智控机器人科技有限公司 | Disconnecting link state identification method, device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060140455A1 (en) * | 2004-12-29 | 2006-06-29 | Gabriel Costache | Method and component for image recognition |
| CN101335465A (en) * | 2008-07-24 | 2008-12-31 | 华中科技大学 | Round disk rotation type switch state image recognition apparatus for electric switch cabinet |
| CN101504321A (en) * | 2008-12-31 | 2009-08-12 | 人民电器集团有限公司 | On-line monitoring system for switch cabinet state |
| CN101645597A (en) * | 2008-12-31 | 2010-02-10 | 人民电器集团有限公司 | Online detection system of switchgear |
-
2010
- 2010-05-18 CN CN 201010174413 patent/CN101833673A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060140455A1 (en) * | 2004-12-29 | 2006-06-29 | Gabriel Costache | Method and component for image recognition |
| CN101335465A (en) * | 2008-07-24 | 2008-12-31 | 华中科技大学 | Round disk rotation type switch state image recognition apparatus for electric switch cabinet |
| CN101504321A (en) * | 2008-12-31 | 2009-08-12 | 人民电器集团有限公司 | On-line monitoring system for switch cabinet state |
| CN101645597A (en) * | 2008-12-31 | 2010-02-10 | 人民电器集团有限公司 | Online detection system of switchgear |
Non-Patent Citations (1)
| Title |
|---|
| 《中国图象图形学报》 20091130 谢红跃 等 一种新的改进Adaboost鰯分类器训练算法 2411-2415 1-4 第14卷, 第11期 2 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI459321B (en) * | 2011-12-29 | 2014-11-01 | ||
| CN104103044A (en) * | 2014-07-09 | 2014-10-15 | 上海电力学院 | Tackle cable jumping on-line detecting method based on K-Mean algorithm |
| CN104200219B (en) * | 2014-08-20 | 2017-12-08 | 深圳供电局有限公司 | A kind of transformer station's disconnecting link bit switch position instruction automatic identifying method and device |
| CN104200219A (en) * | 2014-08-20 | 2014-12-10 | 深圳供电局有限公司 | Method and device for automatically identifying substation breaker and switch indicator |
| CN104331710A (en) * | 2014-11-19 | 2015-02-04 | 集美大学 | On-off state recognition system |
| CN104331710B (en) * | 2014-11-19 | 2018-01-02 | 集美大学 | On off state identifying system |
| CN105047450A (en) * | 2015-08-31 | 2015-11-11 | 中国西电电气股份有限公司 | High-voltage switch position indication interlocking system and method based on image recognition technology |
| CN105047450B (en) * | 2015-08-31 | 2018-01-09 | 中国西电电气股份有限公司 | A kind of high-voltage switch gear position instruction interlock system and method based on image recognition technology |
| CN105869164A (en) * | 2016-03-28 | 2016-08-17 | 国网浙江省电力公司宁波供电公司 | Method and system for detecting on/off state of switch |
| CN106339722A (en) * | 2016-08-25 | 2017-01-18 | 国网浙江省电力公司杭州供电公司 | Line knife switch state monitoring method and device |
| CN106570865A (en) * | 2016-11-08 | 2017-04-19 | 国家电网公司 | Digital-image-processing-based switch state detecting system of power equipment |
| CN106971182A (en) * | 2017-02-06 | 2017-07-21 | 王兴照 | Embedded electric power relay pressing plate is thrown and moves back state Intelligent Identify device and implementation method |
| CN111950606A (en) * | 2020-07-28 | 2020-11-17 | 北京恒通智控机器人科技有限公司 | Disconnecting link state identification method, device, equipment and storage medium |
| CN111950606B (en) * | 2020-07-28 | 2023-11-07 | 北京恒通智控机器人科技有限公司 | Knife switch state identification method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101833673A (en) | Electric power switchgear switch state image recognition system | |
| CN104951775B (en) | Railway highway level crossing signal region security intelligent identification Method based on video technique | |
| CN112381784A (en) | Equipment detecting system based on multispectral image | |
| CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
| WO2007022011A2 (en) | System and process for capturing processing, compressing, and displaying image information | |
| CN101335465A (en) | Round disk rotation type switch state image recognition apparatus for electric switch cabinet | |
| CN110852222A (en) | Campus corridor scene intelligent monitoring method based on target detection | |
| CN110175571A (en) | The intellectual monitoring of substation equipment state and recognition methods | |
| CN111814742A (en) | Knife switch state identification method based on deep learning | |
| CN105300528A (en) | Infrared image diagnosis method and infrared image diagnosis system for transformer station equipment | |
| CN111046728A (en) | Straw combustion detection method based on characteristic pyramid network | |
| CN108373001A (en) | A kind of intelligent industrial robot automation warehousing system | |
| CN110826577A (en) | High-voltage isolating switch state tracking identification method based on target tracking | |
| CN111680609B (en) | Foreign matter identification system and method based on image registration and target detection | |
| CN116805204B (en) | Intelligent plant monitoring method and system | |
| CN113205039A (en) | Power equipment fault image identification and disaster investigation system and method based on multiple DCNNs | |
| CN115761537A (en) | Power transmission line foreign matter intrusion identification method oriented to dynamic characteristic supplement mechanism | |
| CN116681979A (en) | Power equipment target detection method under complex environment | |
| CN109359545B (en) | Cooperative monitoring method and device under complex low-altitude environment | |
| CN112734637B (en) | Thermal infrared image processing method and system for monitoring temperature of lead | |
| CN116310922A (en) | Petrochemical plant area monitoring video risk identification method, system, electronic equipment and storage medium | |
| CN112633249A (en) | Embedded pedestrian flow detection method based on light deep learning framework | |
| CN116310274A (en) | State evaluation method for power transmission and transformation equipment | |
| CN217181103U (en) | Monitoring system of high-voltage switch cabinet | |
| CN111402223B (en) | Transformer substation defect problem detection method using transformer substation video image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20100915 |