CN106980863A - A kind of unit exception diagnostic model in transformer substation video monitoring - Google Patents
A kind of unit exception diagnostic model in transformer substation video monitoring Download PDFInfo
- Publication number
- CN106980863A CN106980863A CN201710180354.2A CN201710180354A CN106980863A CN 106980863 A CN106980863 A CN 106980863A CN 201710180354 A CN201710180354 A CN 201710180354A CN 106980863 A CN106980863 A CN 106980863A
- Authority
- CN
- China
- Prior art keywords
- edge
- sobel
- unit exception
- video monitoring
- diagnostic model
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/457—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by analysing connectivity, e.g. edge linking, connected component analysis or slices
Abstract
The present invention proposes the unit exception diagnostic model in a kind of transformer substation video monitoring, and transformer station's digital remote video monitor picture is identified processing, the substation equipment abnormity diagnosis model that the present invention is set up using the pattern edge Cleaning Principle based on Sobel operators using the model realization.After substation equipment breaks down, information gathering is carried out to substation field by camera, send the image information after collection to host computer, host computer uses the unit exception diagnostic model proposed by the present invention based on Sobel operators, failure is just automatically analyzed, and failure mode is reported, realize substation equipment automatic fault diagnosis function.
Description
Technical field
Patent of the present invention is related to the unit exception diagnostic model in a kind of transformer substation video monitoring, and more particularly to one kind is based on
The substation equipment abnormity diagnosis model that the pattern edge Cleaning Principle of Sobel operators is set up.
Background technology
With the development of information technology, transformer station's Remote Digital Video Monitoring System has been widely used for power system.
Video monitoring image is directly perceived, it is easy to handle, but if with the substantial amounts of increase of the points of monitoring, manually 24 hours differentiating or
The monitoring position of person's control camera is also difficult to the work that completes into one.Pattern recognition technique is to solve transformer station remotely to count
The effective way of word video monitoring.Although many Utilities Electric Co.s domestic at present are assembled with substantial amounts of remote in power plant, transformer station
Number of passes word video surveillance point, but mainly or by direct surveillance and manual patrol complete at present, these video points all have
For functions such as monitoring field apparatus, control remote cameras, but these video resources can not all be automatically processed at present.In order to more
Video resource is utilized well, the application function of transformer station's Remote Digital Video Monitoring System is improved, based on pattern recognition technique
Video monitoring system application study causes extensive concern.Using the reference format in digital video, to the picture of monitoring picture
Face carries out technical finesse, the feature extraction of image and the work such as match with standard picture feature is then carried out, so as to reality
The judgement of the working condition of existing power equipment.
The content of the invention
It is an object of the invention to provide a kind of unit exception diagnostic model for being applied to video monitoring in transformer station.
The present invention is by just analyzing the image that camera in transformer station is gathered, the exception class of automatic decision current device
Type, realizes abnormality diagnostic intellectuality.
The present invention uses Sobel algorithms, is refined by Sobel operators, carries out the steps such as Two-dimensional maximum-entropy image segmentation, real
The modeling of existing unit exception diagnostic model.
Brief description of the drawings
Fig. 1 is flow chart of the present invention.
Specific implementation method
Specific implementation method one：Below in conjunction with brief description of the drawings present embodiment, as shown in figure 1, collecting power transformation by camera
Photo during station failure, sends photo to host computer, host computer is using the fault diagnosis model based on Sobel algorithms to equipment event
Barrier situation is analyzed, and finally gives fault type.
Sobel edge detector principles：
To digital picture { f (x, y) } each pixel, investigate above and below it, the weighted difference of left and right adjoint point gray scale, close neighbour
The power of point is big.
Accordingly, Sobel operators are defined as follows：
-
+
-
Its convolution operator:
It is appropriate to take thresholding TH, make following judgement: s(i, j)>TH, (i, j) is step-like marginal point, and { s (i, j) } is edge
Image.
Sobel operators are easy to spatially realize, Sobel edge detectors not only produce preferable rim detection effect
Really, it is and affected by noise also smaller.When using big neighborhood, noiseproof feature can be more preferable, but can so increase amount of calculation,
And the edge drawn also can be accordingly thicker.Sobel operators are utilized above and below pixel, the intensity-weighted algorithm of left and right adjoint point, root
According to the detection that extreme value this phenomenon progress edge is reached at marginal point.Sobel operators have smoothing effect to noise, provide compared with
It is a kind of edge detection method more commonly used for accurate edge directional information.
The modeling process of Sobel models is as follows
1.Sobel is refined
Specific process step is as follows
(1) first by chromatic image be converted into gray-scale figure as
(2) noise is removed as making medium filtering to gray-scale figure
(3) Grayscale Edge figure is obtained as making the Sobel processing with decay factor to gray-scale figure
(4) the Sobel processing with decay factor is remake to gained Grayscale Edge figure
(5) Grayscale Edge figure subtracts the Sobel results, then the value of marginal point corresponding with negative loop is changed into zero, just
The edge graph refined
2.Sobel operator Two-dimensional maximum-entropy split plot designs
Algorithm steps are given below.
First, the overall segmentation threshold (t, s) of image is calculated with Two-dimensional maximum-entropy method.Then, Canny edges are used
Detective operators obtain image border image.To the every bit on the marginal portion of image, very big noise suppressed is carried out, to edge graph
As taking twice threshold T1And T2.We are less than T1Pixel grey scale be set to 0, obtain image A, then threshold value be less than T2Picture
Plain gray scale is set to 0, obtains image B.Image B threshold value is higher, eliminates most of noise, but also have lost useful edge
Information, and image A threshold value is relatively low, remains more information.On the basis of image B, with image A come supplemental image
B edge.We calculate the threshold value T of edge2Scope be 13 ~ 15.Finally in the case where s is constant, the threshold tried to achieve is utilized
Value T2Image split to obtain result.The method being combined herein using global threshold and local threshold carrys out segmentation figure picture.
The application of unit exception diagnostic model is as follows：
For substation equipment edge shape Deformation Types monitored device we need to set up die plate equipment edge coordinate letter
NumberP(x,y), i.e. device end shape function, this function needs refinement.According to die plate equipment edge shape function ± 5%
The edge coordinate function of the supervision equipment of range searching monitored pictureS(x,y),And refined.According toS(x,y)Region pressP (x,y)Figure stretching conversion is carried out, is set up after stretchingPS(x,y)Edge coordinate function.By rising for supervision equipment edge function
Point coordinates is designated as respectivelyS(x 0 ,y 0 ),PS(x 0 ,y 0 ), then the similitude of device end figure can be judged using following formulaD(x,y)：
The centroid calculation of administrative division map can also be used, i.e., monitor area is reduced by equipment template, passes through computing device figure
Centroid position change come discriminating device abnormal conditions.
If the border template image after segmentation is m × n, the pixel value beyond edge is set to the pixel value within 0, edge
The pixel value at edge is set to, in order to which more prominent barycenter changes, apparatus body is protruded using weight factor matrix λ.If the pixel of (i, j)
It is worth for f (i, j), then the center-of-mass coordinate M (x, y) of equipment is：
WhereinFor weight factor matrix λ elements：
。
Claims (4)
1. the unit exception diagnostic model in a kind of transformer substation video monitoring, it is characterised in that：By extracting from camera
The edge feature of fault in-situ photo, detects general principle using the pattern edge based on Sobel algorithms, utilizes both direction mould
Plate carries out neighborhood convolutional calculation with image digitization, and a direction template is used for detection level edge, and a direction template is used to examine
Survey vertical edge.Sobel operators relatively determine figure by calculating the approximate gradient value of figure gamma function by threshold value
Edge, on this basis, set up out the abnormity diagnosis model of equipment, identify substation equipment morphological state change.
2. a kind of according to claim 1, the unit exception diagnostic model in transformer substation video monitoring, it is characterised in that base
There are 2 technical requirements in Sobel algorithms, one is the approximate gradient value calculating of figure gamma function, and one is the selection of threshold value.Closely
Generally chosen like gradient value calculating methodT x , T y Both direction template, extracts edge feature in order to more accurate, also commonly usesT x , T y , T 45 , T 135 Four direction template；And the selection of threshold value has maximum entropy method (MEM).
3.Sobel operators are digital picture functionsf(x,y)Partial derivative,T x , T y The digital gradient of both direction template is approximately square
Journey is generally as follows description：
。
4. gradient magnitude value：
The general principle of algorithm is, if selected thresholdM(x,y)If,g(x,y) >M(x,y), then it is marginal point, in figureg (x,y) >M(x,y)Point set constitutes edge image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710180354.2A CN106980863A (en) | 2017-03-24 | 2017-03-24 | A kind of unit exception diagnostic model in transformer substation video monitoring |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710180354.2A CN106980863A (en) | 2017-03-24 | 2017-03-24 | A kind of unit exception diagnostic model in transformer substation video monitoring |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106980863A true CN106980863A (en) | 2017-07-25 |
Family
ID=59339586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710180354.2A Pending CN106980863A (en) | 2017-03-24 | 2017-03-24 | A kind of unit exception diagnostic model in transformer substation video monitoring |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106980863A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108549309A (en) * | 2018-05-17 | 2018-09-18 | 青海黄河上游水电开发有限责任公司光伏产业技术分公司 | A kind of monitoring system and monitoring method |
| CN109102669A (en) * | 2018-09-06 | 2018-12-28 | 广东电网有限责任公司 | A kind of transformer substation auxiliary facility detection control method and its device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060181740A1 (en) * | 2004-12-08 | 2006-08-17 | Byung-Gyu Kim | Block artifact phenomenon eliminating device and eliminating method thereof |
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
| CN105303158A (en) * | 2015-08-31 | 2016-02-03 | 国家电网公司 | Line-fitting video intelligent analysis algorithm for disconnecting switch of transformer station |
| CN105915645A (en) * | 2016-06-20 | 2016-08-31 | 上海博英信息科技有限公司 | Substation video monitoring automatic alarm system |
-
2017
- 2017-03-24 CN CN201710180354.2A patent/CN106980863A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060181740A1 (en) * | 2004-12-08 | 2006-08-17 | Byung-Gyu Kim | Block artifact phenomenon eliminating device and eliminating method thereof |
| CN101957325A (en) * | 2010-10-14 | 2011-01-26 | 山东鲁能智能技术有限公司 | Substation equipment appearance abnormality recognition method based on substation inspection robot |
| CN105303158A (en) * | 2015-08-31 | 2016-02-03 | 国家电网公司 | Line-fitting video intelligent analysis algorithm for disconnecting switch of transformer station |
| CN105915645A (en) * | 2016-06-20 | 2016-08-31 | 上海博英信息科技有限公司 | Substation video monitoring automatic alarm system |
Non-Patent Citations (1)
| Title |
|---|
| 张金玉: "梯度边缘检测", 《红外热波检测及其图像序列处理技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108549309A (en) * | 2018-05-17 | 2018-09-18 | 青海黄河上游水电开发有限责任公司光伏产业技术分公司 | A kind of monitoring system and monitoring method |
| CN109102669A (en) * | 2018-09-06 | 2018-12-28 | 广东电网有限责任公司 | A kind of transformer substation auxiliary facility detection control method and its device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018028103A1 (en) | Unmanned aerial vehicle power line inspection method based on characteristics of human vision | |
| Ke et al. | A review of methods for automatic individual tree-crown detection and delineation from passive remote sensing | |
| CN102982341B (en) | Self-intended crowd density estimation method for camera capable of straddling | |
| CN104966304B (en) | Multi-target detection tracking based on Kalman filtering and nonparametric background model | |
| Qin et al. | A hierarchical building detection method for very high resolution remotely sensed images combined with DSM using graph cut optimization | |
| CN113324864B (en) | Pantograph carbon slide plate abrasion detection method based on deep learning target detection | |
| CN101179713A (en) | Method of detecting single moving target under complex background | |
| CN103268480A (en) | System and method for visual tracking | |
| CN105976383A (en) | Power transmission equipment fault diagnosis method based on limit learning machine image recognition | |
| Chen et al. | Bottom-up image detection of water channel slope damages based on superpixel segmentation and support vector machine | |
| CN105740945A (en) | People counting method based on video analysis | |
| CN101957325A (en) | Substation equipment appearance abnormality recognition method based on substation inspection robot | |
| CN105069818A (en) | Image-analysis-based skin pore identification method | |
| CN109359577B (en) | System for detecting number of people under complex background based on machine learning | |
| CN109886170B (en) | Intelligent detection, identification and statistics system for oncomelania | |
| CN104240264A (en) | Height detection method and device for moving object | |
| Ke et al. | Active contour and hill climbing for tree crown detection and delineation | |
| CN116844147A (en) | Pointer instrument identification and abnormal alarm method based on deep learning | |
| Haiyang et al. | Earthquake-collapsed building extraction from LiDAR and aerophotograph based on OBIA | |
| CN106980863A (en) | A kind of unit exception diagnostic model in transformer substation video monitoring | |
| CN105023272A (en) | Crop leaf insect pest detection method and system | |
| CN113177439B (en) | Pedestrian crossing road guardrail detection method | |
| CN104123569B (en) | Video person number information statistics method based on supervised learning | |
| CN114120141A (en) | All-weather remote sensing monitoring automatic analysis method and system thereof | |
| CN105427279A (en) | Grassland drought status monitoring system based on and machine vision and Internet of things, grassland drought status monitoring method |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170725 |
|
| WD01 | Invention patent application deemed withdrawn after publication |