CN104331701A - Insulator external discharge mode identification method based on ultraviolet map - Google Patents

Abstract

The invention relates to an insulator external discharge pattern recognition method based on ultraviolet spectrum. It is characterized in that it includes the following steps: a, firstly, using an ultraviolet imager to conduct non-contact detection of the external discharge of the insulator, and forming an ultraviolet discharge image; b, secondly, performing grayscale processing on the obtained ultraviolet discharge image, and converting the color image to converted into a binary grayscale image; c, then, perform mathematical morphological filtering on the binary grayscale image to obtain a clear binary grayscale spectrum of ultraviolet light spots; d, carry out different Extracting characteristic parameters of variable torches, extracting 7 characteristic parameters of invariable torches; e, inputting 7 characteristic parameters of invariable torches into the neural network for discharge type pattern recognition, and outputting the recognition results. It is proved by trial that the method of the invention uses the ultraviolet detection spectrum to recognize the pattern of the external discharge of the insulator, and has the characteristics of high recognition degree and simple method.

Description

Pattern Recognition Method of Insulator External Discharge Based on Ultraviolet Spectrum

technical field

The invention relates to an insulator external discharge pattern recognition method based on ultraviolet spectrum.

Background technique

Insulators are the most widely used insulation equipment in power systems, and have a large number of applications in lines and substations. During operation, insulators will cause external discharge due to surface pollution, condensation, burrs, etc. Long-term external discharge will lead to a decline in their insulation performance, and even lead to serious damage, leading to grid accidents. There have been many accidents caused by external discharge of insulators in the domestic power grid, so strengthening the detection and analysis of external discharge of insulators is the key to avoiding such accidents. Ultraviolet imaging detection is a non-contact detection method for external discharge of electrical equipment developed in recent years and widely used in the field. At present, the sun-blind ultraviolet detection method is widely used, which has the characteristics of flexible operation, long detection distance, non-contact, high sensitivity, fast response speed, and no interference from sunlight. It is used in the external discharge detection of insulators, wires and other equipment. has been widely used. Through the ultraviolet detection of insulators, it is possible to visually detect whether an external discharge has occurred. Because of its convenient use and intuitive results, it is more and more widely used in power grids.

The current UV detection results are mostly presented in the form of UV pictures or videos, which can give an intuitive quantification of whether discharge occurs and the number of discharge photons, but cannot give further detailed results, such as the surface contamination of the insulator. Whether it is discharge or surface condensation discharge cannot be given, which also limits further analysis of on-site test results. Therefore, on the basis of obtaining accurate ultraviolet detection, further analysis of the detection image and obtaining more detailed detection results are the basis for further utilizing this method and effectively analyzing the detection results.

Contents of the invention

The purpose of the present invention is to provide a method for identifying the external discharge pattern of an insulator based on ultraviolet spectrum, which can accurately realize the identification of the discharge pattern by using ultraviolet image processing and extracting parameters.

An insulator external discharge pattern recognition method based on ultraviolet spectrum, which is particularly characterized in that it includes the following steps:

a, First, use the ultraviolet imager to detect the external discharge of the insulator in a non-contact manner, and form an ultraviolet discharge image;

b, secondly, grayscale processing is performed on the obtained ultraviolet discharge image, and the color picture is converted into a binary grayscale image;

c, then, perform mathematical morphological filtering on the binary grayscale image to obtain a clear binary grayscale spectrum of ultraviolet light spots;

d, extracting invariant torch characteristic parameters according to the filtered ultraviolet spot binary grayscale map, and extracting seven invariant torch characteristic parameters;

e, Input the 7 invariant torch characteristic parameters into the neural network for discharge type pattern recognition, and output the recognition result.

In step b, when converting a color image into a binary grayscale image, the original image is first digitized to form an image digital matrix, and then the white area in the matrix is assigned a value of "1", and the rest of the image is assigned a value of "0" to form a binary image. Value grayscale image.

In the feature parameter extraction stage in step d, 7 two-dimensional invariant moment parameters that are not sensitive to image translation, scaling, mirroring and rotation are extracted as feature parameters by using the invariant feature of the image.

Further, step d specifically includes the following steps:

According to the filtered ultraviolet speckle spectrum, the invariant torch characteristic parameters are extracted, and seven invariant torch characteristic parameters are extracted;

Specifically, the two-dimensional (p+q) moment of the digital image f(x,y) is defined as:

$m_{\mathrm{pq}}=\underset{x}{\mathrm{\Σ}}\underset{\mathrm{the\; y}}{\mathrm{\Σ}}{x}^p{\mathrm{the\; y}}^{q}f(x,\mathrm{the\; y});$ (Formula 1)

Where f(x, y) is a digital matrix formed after image digitization, x represents the abscissa of each point after image digitization, and y represents the ordinate of each point; (p, q)=0,1, 2. The summation is performed on all spatial coordinates x, y of the image, and the corresponding central moment is defined as:

${\mathrm{\μ}}_{\mathrm{pq}}=\underset{x}{\mathrm{\Σ}}\underset{\mathrm{the\; y}}{\mathrm{\Σ}}{(x-\stackrel{\‾}{x})}^p{(\mathrm{the\; y}-\stackrel{\‾}{\mathrm{the\; y}})}^{q}f(x,\mathrm{the\; y});$ (Formula 2)

in:

$\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 10</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ,</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 01</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

In the formula, m ₁₀ can be obtained by p=1, q=0 in formula (1); m ₀₀ can be obtained by p=0, q=0 in formula (1); m ₀₁ can be obtained by p=0 in formula (1), q = 1 to obtain;

The normalized (p+q) order central moment is defined as:

${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}^{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

where p,q=0,1,2,...,

$\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ;</span>$

where p+q=2,3,...,;

Select 7 two-dimensional invariant moment parameters that are not sensitive to image translation, scaling, mirroring and rotation, and the calculation formula is:

(1) φ ₁ = η ₂₀ + η ₀₂ ;

(2) φ ₂ = (η ₂₀ -η ₀₂ ) ² +4η ₁₁ ² ;

(3) φ ₃ = (η ₃₀ -3η ₁₂ ) ² + (3η ₂₁ -η ₀₃ ) ² ;

(4) φ ₄ = (η ₃₀ +η ₁₂ ) ² + (η ₂₁ +η ₀₃ ) ² ;

φ ₅ =(η ₃₀ -3η ₁₂ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]

(5) +(3η ₂₁ -η ₀₃ )(η ₂₁ +η ₀₃ )[3(n ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ];

φ ₆ =(η ₂₀ -η ₀₂ )[(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ]

(6) +4η ₁₁ (η ₃₀ +η ₁₂ )(η ₂₁ +η ₀₃ );

φ ₇ ＝(3η ₂₁ -η ₀₃ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]+

(7) (3η ₁₂ -η ₃₀ )(η ₂₁ +η ₀₃ )[3(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ].

Further, step d specifically includes the following steps: inputting seven invariant torch characteristic parameters into the neural network for discharge type pattern recognition, and judging whether the discharge belongs to surface pollution discharge, surface condensation discharge, or surface burr discharge.

Furthermore, the specific process of using neural network for pattern recognition is as follows: firstly, obtain the ultraviolet spectrum of the external discharge of typical defects in insulators in the laboratory, extract the invariant characteristic operator of the ultraviolet image of external discharge of typical defects, and use the neural network to The typical defect characteristic operator is trained to form a characteristic database. Secondly, for the unknown type of discharge ultraviolet image, after obtaining its invariant characteristic operator, it is input into the neural network for pattern recognition, and the recognition result is output. The output recognition result is dirty discharge, Condensation discharge or glitch discharge.

It is proved by trial that the method of the invention uses the ultraviolet detection spectrum to recognize the pattern of the external discharge of the insulator, and has the characteristics of high recognition degree and simple method.

Description of drawings

Accompanying drawing 1 is the schematic flowchart of the method for identifying the external discharge pattern of an insulator based on ultraviolet spectrum according to the present invention.

Detailed ways

As shown in Fig. 1, the present invention provides a kind of insulator external discharge pattern recognition method based on ultraviolet atlas, this method first carries out gray-scale conversion to the ultraviolet image that the spot detection obtains, and color picture is converted into binary gray-scale image; Then Perform mathematical morphological filtering on the binary grayscale image to obtain a clear binary grayscale spectrum of ultraviolet spots; extract invariant feature parameters from the ultraviolet spot spectrum, and extract 7 parameters that are not sensitive to image translation, scaling, mirroring and rotation The two-dimensional invariant moment parameters are used as characteristic parameters; the characteristic parameters are input into the neural network for discharge type pattern recognition. The method uses ultraviolet detection spectrum to recognize the pattern of insulator external discharge, which has the characteristics of high recognition degree and simple method. This pattern recognition method based on image parameters solves the difficulty that the current ultraviolet detection results cannot be used for discharge type pattern recognition.

Example 1:

The technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings.

Fig. 1 shows the schematic flow sheet of the GIS partial discharge pattern recognition method based on ultrasonic detection in the present invention, and described method comprises the following steps:

Step 1: Use an ultraviolet imager to conduct ultraviolet detection on the insulator, and obtain the original image of its ultraviolet discharge. The original image is a color image including discharge signals and other signals.

Step 2: Perform grayscale processing on the color original image, and convert the color image into a binary grayscale image.

Specifically, the original picture is first digitized to form an image digital matrix f(x, y), and then the white area in the matrix is assigned "1" and the rest is assigned "0" to form a binary grayscale image.

Step 3: Perform mathematical morphological filtering on the binary grayscale image to obtain a clear binary grayscale spectrum of ultraviolet light spots. Image mathematical morphological filtering is a commonly used image denoising algorithm, and it is an existing mature algorithm, so it will not be repeated here.

Step 4: Extract the characteristic parameters of the invariant torch according to the filtered ultraviolet spot spectrum, and extract 7 invariant torch characteristic parameters.

Specifically, the two-dimensional (p+q) moment of the digital image f(x,y) is defined as:

${\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; p</span>}}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; q</span>}}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

Where (p, q) = 0, 1, 2, ..., the summation is performed on all spatial coordinates x, y of the image. The corresponding central moment is defined as:

${\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; p</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; q</span>}}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

in:

$\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 10</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ,</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 01</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

The normalized (p+q) order central moment is defined as:

${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}^{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

where p,q=0,1,2,...,

$\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ;</span>$

where p+q=2,3,...,

Select 7 two-dimensional invariant moment parameters that are not sensitive to image translation, scaling, mirroring and rotation, and the calculation formula is:

1. φ ₁ = η ₂₀ + η ₀₂ ;

2.φ ₂ =(η ₂₀ -η ₀₂ ) ² +4η ₁₁ ² ;

3.φ ₃ =(η ₃₀ -3η ₁₂ ) ² +(3η ₂₁ -η ₀₃ ) ² ;

4.φ ₄ =(η ₃₀ +η ₁₂ ) ² +(η ₂₁ +η ₀₃ ) ² ;

φ ₅ =(η ₃₀ -3η ₁₂ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]

5. +(3η ₂₁ -η ₀₃ )(η ₂₁ +η ₀₃ )[3(n ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ];

φ ₆ =(η ₂₀ -η ₀₂ )[(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ]

6.+4η ₁₁ (η ₃₀ +η ₁₂ )(η ₂₁ +η ₀₃ );

φ ₇ ＝(3η ₂₁ -η ₀₃ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]+

7. (3η ₁₂ -η ₃₀ )(η ₂₁ +η ₀₃ )[3(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ];

The seven invariant torch operators represent the shape features of the ultraviolet spectrum, and their values do not change with the translation, scaling, mirroring and rotation of the image. Different discharge types have different shape features.

Further, it specifically includes the following steps:

According to the filtered ultraviolet speckle spectrum, the invariant torch characteristic parameters are extracted, and seven invariant torch characteristic parameters are extracted;

Specifically, the two-dimensional (p+q) moment of the digital image f(x,y) is defined as:

$m_{\mathrm{pq}}=\underset{x}{\mathrm{\&Sigma;}}\underset{\mathrm{the\; y}}{\mathrm{\&Sigma;}}{x}^p{\mathrm{the\; y}}^{q}f(x,\mathrm{the\; y});$ (Formula 1)

Where f(x, y) is a digital matrix formed after image digitization, x represents the abscissa of each point after image digitization, and y represents the ordinate of each point. (p, q) = 0, 1, 2, ..., the summation is performed on all spatial coordinates x, y of the image, and the corresponding central moment is defined as:

${\mathrm{\&mu;}}_{\mathrm{pq}}=\underset{x}{\mathrm{\&Sigma;}}\underset{\mathrm{the\; y}}{\mathrm{\&Sigma;}}{(x-\stackrel{\&OverBar;}{x})}^p{(\mathrm{the\; y}-\stackrel{\&OverBar;}{\mathrm{the\; y}})}^{q}f(x,\mathrm{the\; y});$ (Formula 2)

in:

$\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 10</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ,</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 01</span>}}}{{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

In the formula, m ₁₀ can be obtained by p=1, q=0 in formula (1); m ₀₀ can be obtained by p=0, q=0 in formula (1); m ₀₁ can be obtained by p=0 in formula (1), q=1 is obtained.

The normalized (p+q) order central moment is defined as:

${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}_{\mathrm{<span\; class="notranslate">\; pq</span>}}}{{{\mathrm{<span\; class="notranslate">\; \&mu;</span>}}^{\mathrm{<span\; class="notranslate">\; \&gamma;</span>}}}_{\mathrm{<span\; class="notranslate">\; 00</span>}}}<span\; class="notranslate">\; ;</span>$

where p,q=0,1,2,...,

$\mathrm{<span\; class="notranslate">\; \&gamma;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; q</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ;</span>$

where p+q=2,3,...,;

Select 7 two-dimensional invariant moment parameters that are not sensitive to image translation, scaling, mirroring and rotation, and the calculation formula is:

(1) φ ₁ = η ₂₀ + η ₀₂ ;

(2) φ ₂ = (η ₂₀ -η ₀₂ ) ² +4η ₁₁ ² ;

(3) φ ₃ = (η ₃₀ -3η ₁₂ ) ² + (3η ₂₁ -η ₀₃ ) ² ;

(4) φ ₄ = (η ₃₀ +η ₁₂ ) ² + (η ₂₁ +η ₀₃ ) ² ;

φ ₅ =(η ₃₀ -3η ₁₂ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]

(5) +(3η ₂₁ -η ₀₃ )(η ₂₁ +η ₀₃ )[3(n ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ];

φ ₆ =(η ₂₀ -η ₀₂ )[(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ]

(6) +4η ₁₁ (η ₃₀ +η ₁₂ )(η ₂₁ +η ₀₃ );

φ ₇ ＝(3η ₂₁ -η ₀₃ )(η ₃₀ +η ₁₂ )[(η ₃₀ +η ₁₂ ) ² -3(η ₂₁ +η ₀₃ ) ² ]+

(7) (3η ₁₂ -η ₃₀ )(η ₂₁ +η ₀₃ )[3(η ₃₀ +η ₁₂ ) ² -(η ₂₁ +η ₀₃ ) ² ].

Each of the above letters can be derived from Formula 1 and Formula 2, and the meanings of the letters are 0, 1, 2, 3, etc.

Step 5: Input the 7 constant torch characteristic parameters into the neural network for discharge type pattern recognition, and judge whether the discharge belongs to surface pollution discharge, surface condensation discharge and surface burr discharge.

Specifically, the process of using the neural network for pattern recognition is as follows: firstly, obtain the ultraviolet spectrum of the external discharge of typical insulator defects in the laboratory, extract the invariant characteristic operator of the ultraviolet image of the typical defect external discharge, and use the neural network to analyze the characteristics of typical defects Operators are trained to form a feature database. Secondly, for the unknown type of discharge ultraviolet image, after obtaining its invariant torch characteristic operator, it is input into the neural network for pattern recognition, and the recognition result is output.

Step 6: Output the identification result as pollution discharge, condensation discharge or glitch discharge.

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto, any changes or substitutions that can be easily imagined by those skilled in the art within the technical scope disclosed in the present invention, All should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims (6)

1., based on an exterior insulator discharge mode recognition methods for uv-spectrogram, it is characterized in that, comprise the steps:

A, first, utilizes ultraviolet imager to carry out non-contact detection to exterior insulator electric discharge, and forms EUV discharge image;

B, secondly, carries out gray proces by the EUV discharge image of acquisition, colour picture is converted into two-value gray level image;

C, then, carries out mathematical morphology filter to two-value gray level image, obtains ultraviolet hot spot two-value gray images clearly;

D, carries out constant torch characteristic parameter extraction according to the ultraviolet hot spot two-value gray images after filtering, extracts 7 constant torch characteristic parameters;

7 constant torch characteristic parameter input neural networks are carried out electric discharge type pattern-recognition by e, export recognition result.

2. as claimed in claim 1 based on the exterior insulator discharge mode recognition methods of uv-spectrogram, it is characterized in that: in step b when colour picture being converted into two-value gray scale picture, first by original image digitizing, form image digitization matrix, then be " 1 " by the white portion assignment in matrix, remainder assignment is " 0 ", forms two-value gray level image.

3. as claimed in claim 1 based on the exterior insulator discharge mode recognition methods of uv-spectrogram, it is characterized in that: in characteristic parameter extraction stage in steps d, utilize the constant torch feature of image, extract 7 to all insensitive two dimension invariant moment parameter of image translation, convergent-divergent, mirror image and rotation as characteristic parameter amount.

4., as claimed in claim 1 based on the exterior insulator discharge mode recognition methods of uv-spectrogram, it is characterized in that, steps d specifically comprises the steps:

Carry out constant torch characteristic parameter extraction according to the ultraviolet hot spot collection of illustrative plates after filtering, extract 7 constant torch characteristic parameters;

Concrete, two dimension (p+q) the rank square of digital picture f (x, y) is defined as:

$m_{\mathrm{pq}}=\underset{x}{\mathrm{\&Sigma;}}\underset{y}{\mathrm{\&Sigma;}}{x}^p{y}^{q}f(x,y);$ (formula 1)

The wherein character matrix of f (x, y) for being formed after image digitazation, x characterize image digitazation after each point horizontal ordinate, y characterize each point ordinate; (p, q)=0,1,2, sue for peace and to carry out on all volume coordinate x, the y of image, corresponding center square is defined as:

${\mathrm{\&mu;}}_{\mathrm{pq}}=\underset{x}{\mathrm{\&Sigma;}}\underset{y}{\mathrm{\&Sigma;}}{(x-\stackrel{\&OverBar;}{x})}^p{(y-\stackrel{\&OverBar;}{y})}^{q}f(x,y);$ (formula 2)

Wherein:

$\stackrel{\&OverBar;}{x}=\frac{m_{10}}{m_{00}},\stackrel{\&OverBar;}{y}=\frac{m_{01}}{m_{00}};$

M in formula ₁₀can be tried to achieve by p=1, q=0 in formula (1); m ₀₀can be tried to achieve by p=0, q=0 in formula (1); m ₀₁can be tried to achieve by p=0, q=1 in formula (1);

Normalization (p+q) center, rank square is defined as:

${\mathrm{\&eta;}}_{\mathrm{pq}}=\frac{{\mathrm{\&mu;}}_{\mathrm{pq}}}{{{\mathrm{\&mu;}}^{\mathrm{\&gamma;}}}_{00}};$

Wherein p, q=0,1,2 ...,

$\mathrm{\&gamma;}=\frac{p+q}{2}+1;$

Wherein p+q=2,3 ...;

Select to 7 to all insensitive two dimension invariant moment parameter of image translation, convergent-divergent, mirror image and rotation, its computing formula is:

(1)φ ₁＝η ₂₀+η ₀₂；

(2)φ ₂＝(η ₂₀-η ₀₂) ²+4η ₁₁ ²；

(3)φ ₃＝(η ₃₀-3η ₁₂) ²+(3η ₂₁-η ₀₃) ²；

(4)φ ₄＝(η ₃₀+η ₁₂) ²+(η ₂₁+η ₀₃) ²；

$\left(5\right)\begin{array}{c}{\mathrm{\&phi;}}_5=({\mathrm{\&eta;}}_{30}-3{\mathrm{\&eta;}}_{12})({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})[{({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})}^2-3{({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})}^2]\\ +(3{\mathrm{\&eta;}}_{21}-{\mathrm{\&eta;}}_{03})({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})[3{(n_{30}+{\mathrm{\&eta;}}_{12})}^2-{({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})}^2]\end{array};$

$\left(6\right)\begin{array}{c}{\mathrm{\&phi;}}_6=({\mathrm{\&eta;}}_{20}-{\mathrm{\&eta;}}_{02})[{({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})}^2-3{({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})}^2]\\ +4{\mathrm{\&eta;}}_{11}({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})\end{array};$

$\left(7\right)\begin{array}{c}{\mathrm{\&phi;}}_7=({3\mathrm{\&eta;}}_{21}-{\mathrm{\&eta;}}_{03})({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})[{({\mathrm{\&eta;}}_{30}+{\mathrm{\&eta;}}_{12})}^2-3{({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})}^2]+\\ (3{\mathrm{\&eta;}}_{12}-{\mathrm{\&eta;}}_{30})({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})[3{(n_{30}+{\mathrm{\&eta;}}_{12})}^2-{({\mathrm{\&eta;}}_{21}+{\mathrm{\&eta;}}_{03})}^2]\end{array}.$

5. as claimed in claim 1 based on the exterior insulator discharge mode recognition methods of uv-spectrogram, it is characterized in that, steps d specifically comprises the steps: 7 constant torch characteristic parameter input neural networks to carry out electric discharge type pattern-recognition, judges whether electric discharge belongs to surface filth electric discharge, the electric discharge of surperficial condensation, surface spikes electric discharge.

6. as claimed in claim 5 based on the exterior insulator discharge mode recognition methods of uv-spectrogram, it is characterized in that, the concrete process utilizing neural network to carry out pattern-recognition is: the uv-spectrogram obtaining insulator typical defect external discharge first in the lab, extract the constant torch feature operator of typical defect external discharge ultraviolet image, neural network is utilized to train typical defect feature operator, morphogenesis characters database, secondly for the electric discharge ultraviolet image of UNKNOWN TYPE, after obtaining its constant torch feature operator, input neural network, carry out pattern-recognition, export recognition result, output recognition result is contaminant flashover, condensation electric discharge or burr electric discharge.