CN106022302A - Method of identifying faulted jumper yoke plate by USFPF characteristics - Google Patents

Abstract

The invention provides a method of identifying a faulted jumper yoke plate by USFPF characteristics. The method comprises following steps: step A, Otsu self-adaption threshold segmentation of an infrared gray scale image of a power transmission line is executed to acquire a suspected target image; step B, the connected domain separation of the suspected target image is executed to acquire a plurality of images; step C, the skeleton extraction of the plurality of images acquired by the step B is executed to acquire a plurality of skeleton images; step D, the USFPF characteristic of each of the above mentioned skeleton images is calculated, and the faulted jumper yoke plate is identified according to the USFPF characteristics. The method provided by the invention is advantageous in that the shape characteristic of the faulted jumper yoke plate is comprehensively considered, and by starting from the whole performance of the identified faulted jumper yoke plate, identification characteristics are selected in a selective manner, and therefore identification effect is improved.

Description

A kind of method by USFPF feature identification fault wire jumper yoke plate

Technical field

The invention belongs to transmission line malfunction wire jumper yoke plate identification technical field, be specifically related to one and pass through The method of USFPF feature identification fault wire jumper yoke plate.

Background technology

Society is constantly advancing, and power system is the heart providing power for society, but the most out of doors The transmission facility run is subject to the impact of the factors such as aging, vile weather, often breaks down, because of This finds the fault of transmission line of electricity the most in time, and to carry out maintenance be a study hotspot.Transmission line of electricity occurs Fault may result in superheating phenomenon, and infrared band is temperature sensitive, and therefore Leslie proposed fortune in 1949 With infrared measurement of temperature equipment Inspection transmission line malfunction.But owing to equipment fell behind at that time, the method does not has quilt Use widely.Along with the progress of science, infrared equipment is the most increasingly advanced, and has and do not contact, do not stop Electricity, the advantage such as remote, be used in transmission line faultlocating aspect widely.But use at digital picture Reason Intelligent Measurement to identify that the research of transmission line equipment fault also needs to more deep.

Electrical equipment operationally, due to the factor such as environment, conductive material, can produce various loss, These losses eventually exhale by the way of heat energy, and conductive equipment mostly is metal, and heat conductivility is relatively Good, so along with the working time is elongated, transmission facility will generate heat, cause the physical property of transmission facility to become Bad, transmission facility once breaks down, and its loss will drastically become big, and its temperature also can rise rapidly.Institute With, determine whether transmission facility generates heat seriously, it may be determined that whether it breaks down or there is hidden danger.? In high-voltage transmission equipment, wire jumper yoke plate, splicing sleeve and insulator are easiest to break down.The most existing many Scientific research personnel is by studying the recognizer of high-voltage transmission equipment heating region in thermal map, but directly passes through The research of the recognizer of infrared image utilization graphics research high voltage transmission line heating region is the most deep, The present invention is through being analyzed a large amount of infrared images, it is achieved that known in infrared image by iconology algorithm Wire jumper yoke plate fault in other transmission facility.

Summary of the invention

The invention provides a kind of method by USFPF feature identification fault wire jumper yoke plate, the method is comprehensive Consider the shape facility of fault wire jumper yoke plate, from the general performance situation of the fault wire jumper yoke plate identified, There is the selection identification feature of the property selected, make recognition effect be improved.

Realize technical scheme as follows:

A kind of method by USFPF feature identification fault wire jumper yoke plate, detailed process is:

Step A, carries out Otsu adaptive threshold fuzziness to the infrared hybrid optical system of transmission line of electricity, is doubted Like target image；

Step B, carries out connected domain separation to suspected target image, obtains multiple image；

Step C, the multiple image obtaining step B carries out skeletal extraction, obtains several skeleton image；

Step D, calculates the USFPF (U-shaped four-point feature) of each width in several skeleton image Feature, according to described USFPF feature identification fault wire jumper yoke plate.

Further, when step A of the present invention carries out Threshold segmentation, the threshold value used is that Otsu obtains dynamically Optimal threshold is adjusted up the value after 60-70 gray level.

Further, step B of the present invention carry out connected domain separation before, suspected target image is carried out unrestrained water Method filters.

Further, the present invention is before carrying out skeletal extraction, and the multiple image first obtained step B carries out swollen Swollen and holes filling processes.

Further, the detailed process of step D of the present invention is as follows:

Fault wire jumper yoke plate is U-shaped feature, and U-shaped characteristic area must exist four effective nodes, including Two border vertices and two process flex points,

First, with eight in connected domain only one of which target pixel points as constraints, search out top, two borders Point；

Secondly, with one of them border vertices as starting point, then move on skeleton with the detection window of 3 × 3, Two points that gray scale grey scale change in the horizontal and vertical directions is maximum are defined as process flex point；

Window move grey scale change E that [u, v] produce (u, v):

$E(u,v)=\underset{x,y}{\Σ}w(x,y){\[f(x+u,y+v)-f(x,y)\]}^2---\left(1\right)$

Wherein, (u, v) is the grey scale change amount before and after window moves to E, and (x, y) is window function to w, and f (x+u, y+v) is Gray level image after translation, and have f (x+u, y+v)=f (x, y)+f_xu+f_yv+O(u²,v²), f_xFor x direction First differential f_yFor the first differential on y direction, O (u²,v²) it is Pei Yanuo (Peano) remainder.

So former grey scale change function can be converted into:

$E(u,v)=\[u,v\]\left[\begin{array}{cc}{f}_x^2& f_x{f}_y\\ f_x{f}_y& f_y^2\end{array}\right]\left[\begin{array}{c}u\\ v\end{array}\right]$

Minute movement amount approximate expression:

$E(u,v)\≅\[u,v\]M\left[\begin{array}{c}u\\ v\end{array}\right]$

Wherein M is 2*2 matrix, can be able to be obtained by image derivation:

$M=\underset{x,y}{\Σ}w(x,y)\left[\begin{array}{cc}{f}_x^2& f_x{f}_y\\ f_x{f}_y& f_y^2\end{array}\right]$

Finally, the process flex point by starting point, searched for the first time, the process flex point searched for the second time and end Point is designated as pt1 (x successively₁,y₁), pt2 (x₂,y₂), pt3 (x₃,y₃), pt4 (x₄,y₄)；Junction point pt1 (x₁,y₁) and pt2 (x₂,y₂) The slope of obtained straight line is designated as k₁, junction point pt2 (x₂,y₂) and pt3 (x₃,y₃Obtained by), the slope of straight line is designated as k₂, Junction point pt3 (x₃,y₃) and pt4 (x₄,y₄Obtained by), the slope of straight line is designated as k₃；

It is judged to that the condition of U-shaped heating region is as follows:

(1)sign(x₁-x₂)=sign (x₄-x₃)

(2)|k₁-k₂| ＞ tan (π/4)

(3)|k₂-k₃| ＞ tan (π/4)

(4)|k₁-k₃| ＜ tan (π/4)

Wherein, sign is sign function, sign (x₁-x₂)=sign (x₄-x₃) represent x₁-x₂And x₄-x₃Simultaneously greater than 0 or Person is less than 0, represents that starting point points to direction and the terminal sensing second time search of the process flex point searched for the first time To process flex point be same direction.

When meeting aforementioned four condition simultaneously, USFPF is characterized as 1, otherwise is then 0；When USFPF is special Levy when being 1, be identified as fault wire jumper yoke plate.

Beneficial effect

First, the present invention, by analyzing the fault wire jumper yoke plate in a large amount of infrared images, proposes a kind of based on bone The USFPF feature of the slope relative size that frame territory angle point and process flex point are constituted to identify fault wire jumper yoke plate, Fault wire jumper yoke plate can be identified efficiently.

Second, the present invention is when utilizing OSTU threshold segmentation method, and the threshold value used is that Otsu acquirement is dynamic State optimal threshold is adjusted up the value after 60-70 gray level, preferably can be partitioned into fault from infrared image Wire jumper yoke plate.

3rd, method provided by the present invention can identify fault wire jumper yoke plate efficiently, makes maintenance power train Unite convenient.

Accompanying drawing explanation

Fig. 1 is the overall flow schematic diagram of the present invention；

Fig. 2 is for identifying U-shaped feature process principle figure；

Fig. 3 is process flex point, border vertices and window moving process grey scale change situation；

Fig. 4 is fault wire jumper yoke plate figure.

Detailed description of the invention

Combine accompanying drawing below, the present invention is further elaborated.

The present invention relates to a kind of method by USFPF feature identification fault wire jumper yoke plate, the method is used for Patrolling in bat infrared image of unmanned plane identifies fault wire jumper yoke plate.

The process of the present invention patrols bat infrared image to liking unmanned plane, and this original image is to carry on unmanned plane Infrared camera gather gained.Use the Ostu thresholding method improved that infrared image is split, then use Unrestrained water law filters and separates each connected domain, uses expansion and Hole filling algorithms filter off some zonules and fill out Fill the hole in big region, extract the skeleton of each connected domain, skeleton image finds summit and process flex point, Calculate consecutive points and connected straight slope, by the relative size identification fault wire jumper yoke plate between this slope.

The detailed process of the inventive method is:

Step A, carries out Otsu adaptive threshold fuzziness to the infrared hybrid optical system of transmission line of electricity, removes the back of the body Scape, obtains suspected target image；

The segmentation image obtained by traditional Otsu method obtains segmentation threshold, and Otsu is obtained by the present invention Good threshold value is adjusted up 60-70 gray level, can obtain more preferable result.

Step B, carries out unrestrained water law filtering to suspected target image and connected domain separates, and obtains multiple image.

Step C, the multiple image obtaining step B expands and holes filling processes, then carries out skeleton Extract, obtain several skeleton image；

After suspicious type connected region being separated by unrestrained water law, in fact it could happen that following defect: 1. intermediate void, 2. border sawtooth, so needing to do isolated image to expand and the process of holes filling.The most again to it Extract its skeleton again.

Step D, calculates the USFPF (U-shaped four-point feature) of each width in several skeleton image Feature, according to described USFPF feature identification fault wire jumper yoke plate.

Four effective nodes, two border vertices and two processes must be there are and turn in one U-shaped characteristic area Point, when extracting USFPF feature, it is first determined border vertices, in border vertices eight in connection field Only one of which target pixel points, for process flex point, it is necessary at least there are inlet point and two targets of exit point Pixel.Therefore the detailed process extracting USFPF is:

First, with eight in connected domain only one of which target pixel points as constraints, search out top, two borders Point；

Wherein, with one of them border vertices as starting point, then move on skeleton with the detection window of 3 × 3 Detection can run into three kinds of situations, and (a) is that detection window moves on skeleton, does not has gray scale to become in the horizontal direction Changing, (b) is that detection window moves on skeleton, and horizontally and vertically direction all exists obvious grey scale change, (c) Move on skeleton for detection window, vertical direction does not has grey scale change.Window moves the ash that [u, v] produces Degree change:

$E(u,v)=\underset{x,y}{\Σ}w(x,y){\[f(x+u,y+v)-f(x,y)\]}^2$

Wherein, (x, y) is window function to w, and f (x+u, y+v) is the gray level image after translation, and has F (x+u, y+v)=f (x, y)+f_xu+f_yv+O(u²,v²)。

So former grey scale change function can be converted into:

$E(u,v)=\[u,v\]\left[\begin{array}{cc}{f}_x^2& f_x{f}_y\\ f_x{f}_y& f_y^2\end{array}\right]\left[\begin{array}{c}u\\ v\end{array}\right]$

Minute movement amount approximate expression:

$E(u,v)\≅\[u,v\]M\left[\begin{array}{c}u\\ v\end{array}\right]$

Wherein M is 2*2 matrix, can be able to be obtained by image derivation:

$M=\underset{x,y}{\Σ}w(x,y)\left[\begin{array}{cc}{f}_x^2& f_x{f}_y\\ f_x{f}_y& f_y^2\end{array}\right]$

Two points that gray scale grey scale change in the horizontal and vertical directions is maximum are defined as process flex point；

Finally, the process flex point by starting point, searched for the first time, the process flex point searched for the second time and end Point is designated as pt1 (x successively₁,y₁), pt2 (x₂,y₂), pt3 (x₃,y₃), pt4 (x₄,y₄)；Junction point pt1 (x₁,y₁) and pt2 (x₂,y₂) The slope of obtained straight line is designated as k₁, junction point pt2 (x₂,y₂) and pt3 (x₃,y₃Obtained by), the slope of straight line is designated as k₂, Junction point pt3 (x₃,y₃) and pt4 (x₄,y₄Obtained by), the slope of straight line is designated as k₃；

It is judged to that the condition of U-shaped heating region is as follows:

(1)sign(x₁-x₂)=sign (x₄-x₃)

(2)|k₁-k₂| ＞ tan (π/4)

(3)|k₂-k₃| ＞ tan (π/4)

(4)|k₁-k₃| ＜ tan (π/4)

Wherein, sign is sign function, sign (x₁-x₂)=sign (x₄-x₃) represent x₁-x₂And x₄-x₃Simultaneously greater than 0 or Person is less than 0, represents that starting point points to direction and the terminal sensing second time search of the process flex point searched for the first time To process flex point be same direction.

When meeting aforementioned four condition simultaneously, USFPF is characterized as 1, otherwise is then 0；When USFPF is special Levy when being 1, be identified as fault wire jumper yoke plate.

In sum, these are only presently preferred embodiments of the present invention, be not intended to limit the guarantor of the present invention Protect scope.All within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. made, Should be included within the scope of the present invention.

Claims (6)

1. the method passing through USFPF feature identification fault wire jumper yoke plate, it is characterised in that detailed process For:

Step A, carries out Otsu adaptive threshold fuzziness to the infrared hybrid optical system of transmission line of electricity, is doubted Like target image；

Step B, carries out connected domain separation to suspected target image, obtains multiple image；

Step C, the multiple image obtaining step B carries out skeletal extraction, obtains several skeleton image；

Step D, calculates the USFPF feature of each width in several skeleton image, according to described USFPF feature Identify fault wire jumper yoke plate.

The most according to claim 1 by the method for USFPF feature identification fault wire jumper yoke plate, its feature Being, when step A carries out Threshold segmentation, the threshold value used is that Otsu obtains dynamic optimal threshold to rise Save the value after 60-70 gray level.

The most according to claim 1 by the method for USFPF feature identification fault wire jumper yoke plate, its feature Be, step B carry out connected domain separation before, suspected target image is carried out the filtering of unrestrained water law.

The most according to claim 1 by the method for USFPF feature identification fault wire jumper yoke plate, its feature Being, before carrying out skeletal extraction, the multiple image first obtained step B expands and at holes filling Reason.

The most according to claim 1 by the method for USFPF feature identification fault wire jumper yoke plate, its feature Being, fault wire jumper yoke plate is U-shaped feature, and U-shaped characteristic area must exist four effective nodes, bag Including two border vertices and two process flex points, the detailed process of described step D is as follows:

First, with eight in connected domain only one of which target pixel points as constraints, search out top, two borders Point；

Secondly, with one of them border vertices as starting point, then move on skeleton with the detection window of N × N Dynamic, two points that gray scale grey scale change in the horizontal and vertical directions is maximum are defined as process flex point；

Finally, the process flex point by starting point, searched for the first time, the process flex point searched for the second time and end Point is designated as pt1 (x successively₁,y₁), pt2 (x₂,y₂), pt3 (x₃,y₃), pt4 (x₄,y₄)；Junction point pt1 (x₁,y₁) and pt2 (x₂,y₂) The slope of obtained straight line is designated as k₁, junction point pt2 (x₂,y₂) and pt3 (x₃,y₃Obtained by), the slope of straight line is designated as k₂, Junction point pt3 (x₃,y₃) and pt4 (x₄,y₄Obtained by), the slope of straight line is designated as k₃；

It is judged to that the condition of U-shaped heating region is as follows:

(1)sign(x₁-x₂)=sign (x₄-x₃)

(2)|k₁-k₂| ＞ tan (π/4)

(3)|k₂-k₃| ＞ tan (π/4)

(4)|k₁-k₃| ＜ tan (π/4)

Wherein, sign is sign function, sign (x₁-x₂)=sign (x₄-x₃) represent x₁-x₂And x₄-x₃Simultaneously greater than 0 or Person is less than 0, represents that starting point points to direction and the terminal sensing second time search of the process flex point searched for the first time To process flex point be same direction；

When meeting aforementioned four condition simultaneously, USFPF is characterized as 1, otherwise is then 0；When USFPF is special Levy when being 1, be identified as fault wire jumper yoke plate.

The most according to claim 1 by the method for USFPF feature identification fault wire jumper yoke plate, its feature Being, described N is equal to 3.