CN112734713B

CN112734713B - Transformer state detection method, device, computer equipment and storage medium

Info

Publication number: CN112734713B
Application number: CN202011637727.2A
Authority: CN
Inventors: 张凡; 张玉焜; 朱筱瑜; 汲胜昌; 毛光辉; 季坤; 丁国成; 张晨晨
Original assignee: State Grid Corp of China SGCC; Xian Jiaotong University; Electric Power Research Institute of State Grid Anhui Electric Power Co Ltd
Current assignee: State Grid Corp of China SGCC; Xian Jiaotong University; Electric Power Research Institute of State Grid Anhui Electric Power Co Ltd
Priority date: 2020-12-31
Filing date: 2020-12-31
Publication date: 2024-02-09
Anticipated expiration: 2040-12-31
Also published as: CN112734713A

Abstract

The application relates to a transformer state detection method, a transformer state detection device, computer equipment and a storage medium. The method comprises the following steps: and acquiring acoustic image information of the transformer to be detected, and carrying out gray processing on the acoustic image information to obtain a corresponding gray image to be detected. And extracting texture characteristic values of the gray level image to be detected, checking the texture characteristic values according to the theoretical characteristic values to generate a checking result, and further determining the mechanical state of the transformer to be detected according to the checking result. By adopting the method, the visual acoustic image information of the transformer to be detected is acquired, the texture characteristic value of the corresponding acoustic information is accurately checked, the fault or normal state of the transformer to be detected is rapidly distinguished, and the accuracy of the mechanical fault detection result of the transformer is improved.

Description

Transformer state detection method, device, computer equipment and storage medium

Technical Field

The present disclosure relates to the field of power technologies, and in particular, to a method and apparatus for detecting a transformer state, a computer device, and a storage medium.

Background

With the development of power technology and the wide application of power transformers as important power equipment for realizing remote transmission and power distribution in power systems, the requirements on the safety and stability of the power transformers are increasingly raised. The vibration form of the transformer oil tank changes along with the micro-change of the internal mechanical structural parts, so that the noise radiated outwards from the surface of the oil tank also carries corresponding mechanical fault information, and the method can be used for determining whether the transformer has corresponding mechanical faults.

The windings are used as important mechanical components in the transformer oil tank, and the proportion of mechanical faults is high. Because the windings vibrate in the leakage magnetic field due to the fact that alternating electromagnetic force is born, the windings are deformed and other mechanical defects are caused by long-term vibration effects, the change of mechanical states is irreversible, and finally the short-circuit resistance of the transformer is reduced. Therefore, conventionally, the winding of the transformer oil tank is detected by using, for example, a low-voltage pulse method, a frequency response analysis method, a short-circuit impedance method, and the like, so as to obtain the mechanical state of the winding, and further determine whether the transformer has a corresponding mechanical fault.

However, at present, for the mechanical state of the transformer winding, the adopted detection modes such as a low-voltage pulse method, a frequency response analysis method, a short-circuit impedance method and the like need to be electrically connected with the power system, when the detection is performed, the operation of the whole power system is easily affected, meanwhile, as the power system relates to various power equipment, the detection result obtained according to the traditional detection mode is easily interfered greatly, and the detection result of the mechanical fault of the transformer is still inaccurate.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a transformer state detection method, apparatus, computer device, and storage medium that can improve accuracy of a mechanical failure detection result of a transformer.

A method for detecting a state of a transformer, the method comprising:

acquiring acoustic image information of a transformer to be detected;

carrying out gray processing on the acoustic image information to obtain a corresponding gray image to be detected;

extracting texture characteristic values of the gray level image to be detected;

according to the theoretical characteristic value, verifying the texture characteristic value to generate a verification result;

and determining the mechanical state of the transformer to be detected according to the verification result.

In one embodiment, the mechanical state of the transformer includes a normal state and a fault state; the method further comprises the steps of:

acquiring first sample acoustic image information of a sample transformer after load current adjustment in a normal state;

acquiring second sample acoustic image information of the sample transformer after load current adjustment in a fault state;

and carrying out gray processing on the first sample acoustic image information and the second sample acoustic image information to generate a corresponding positive sample training set and a corresponding negative sample training set.

In one embodiment, the fault states include winding loosening fault states with different loosening degrees and insulating gasket falling fault states with different falling numbers; the obtaining second sample acoustic image information of the sample transformer after load current adjustment in the fault state includes:

acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under winding loosening fault states with different loosening degrees;

or acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under the falling fault states of insulating gaskets with different falling numbers.

In one embodiment, the extracting the texture feature value of the gray-scale image to be detected includes:

acquiring a central pixel point of the gray level image to be detected;

acquiring gray values of a preset number of pixel points adjacent to the central pixel point from the detected gray image;

according to a preset rotation function, performing circular right shift on the preset number of pixel points to obtain texture feature vectors corresponding to the preset number of pixel points;

and determining a minimum texture feature vector from texture feature vectors corresponding to the preset number of pixel points, and determining the minimum texture feature vector as a texture feature value of the gray image to be detected.

In one embodiment, before the verifying the texture feature value according to the theoretical feature value and generating the verification result, the method further includes:

and randomly determining theoretical characteristic values from the positive sample training set or the negative sample training set.

In one embodiment, the verifying the texture feature value according to the theoretical feature value, to generate a verification result, includes:

calculating the difference degree of the theoretical characteristic value and the texture characteristic value;

and generating a verification result of the texture characteristic value according to the difference degree and a preset difference threshold value.

In one embodiment, the theoretical feature values include a first theoretical feature value determined from the positive sample training set and a second theoretical feature value determined from the negative sample training set; and determining the mechanical state of the transformer to be detected according to the verification result, wherein the determining comprises the following steps:

when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined to be larger than the preset difference threshold value according to the verification result, determining that the mechanical state of the transformer to be detected is a fault state;

or when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined to be smaller than the preset difference threshold value according to the verification result, determining that the mechanical state of the transformer to be detected is a fault state.

A transformer condition detection apparatus, the apparatus comprising:

the acoustic image information acquisition module is used for acquiring acoustic image information of the transformer to be detected;

the gray image to be detected generation module is used for carrying out gray processing on the acoustic image information to obtain a corresponding gray image to be detected;

the texture characteristic value extraction module is used for extracting texture characteristic values of the gray level image to be detected;

the verification result generation module is used for verifying the texture characteristic value according to the theoretical characteristic value to generate a verification result;

and the mechanical state determining module is used for determining the mechanical state of the transformer to be detected according to the verification result.

A computer device comprising a memory storing a computer program and a processor which when executing the computer program performs the steps of:

acquiring acoustic image information of a transformer to be detected;

A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:

acquiring acoustic image information of a transformer to be detected;

In the transformer state detection method, the transformer state detection device, the computer equipment and the storage medium, the corresponding gray level image to be detected is obtained by acquiring the acoustic image information of the transformer to be detected and carrying out gray level processing on the acoustic image information. The texture characteristic value of the gray level image to be detected is extracted, the texture characteristic value is verified according to the theoretical characteristic value, a verification result is generated, and then the mechanical state of the transformer to be detected is determined according to the verification result. The method realizes the acquisition of visual acoustic image information of the transformer to be detected and the accurate verification of texture characteristic values of corresponding acoustic information, so as to realize the rapid distinction of faults or normal states of the transformer to be detected, and improve the accuracy of mechanical fault detection results of the transformer.

Drawings

FIG. 1 is a diagram of an application environment of a transformer state detection method in one embodiment;

FIG. 2 is a flow chart of a transformer status detection method according to an embodiment;

FIG. 3 is a schematic diagram of an acoustic imaging apparatus of a transformer state detection method in one embodiment;

FIG. 4 is a schematic diagram of a gray scale image to be detected obtained by performing gray scale processing on acoustic image information in one embodiment;

FIG. 5 is a flow chart of extracting texture feature values of a gray image to be detected according to an embodiment;

FIG. 6 is a diagram of rotational invariance of a partial binary pattern in one embodiment;

FIG. 7 is a flow chart of a transformer status detection method according to another embodiment;

FIG. 8 is a block diagram of a transformer status detection device in one embodiment;

FIG. 9 is a block diagram of a transformer status detection device according to another embodiment;

fig. 10 is an internal structural view of a computer device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

The transformer state detection method provided by the application can be applied to an application environment shown in fig. 1. Wherein the transformer to be tested 102 communicates with the server 104 via a network. The server 104 obtains the corresponding gray image to be detected by acquiring the acoustic image information of the transformer to be detected 102 and performing gray processing on the acoustic image information. Further, the server 104 extracts the texture feature value of the gray level image to be detected, verifies the texture feature value according to the theoretical feature value, generates a verification result, and determines the mechanical state of the transformer 102 to be detected according to the verification result. The transformer 102 to be tested is a static electrical device for converting ac voltage and current to transmit ac power, and is implemented according to electromagnetic induction principle, including but not limited to a power transformer, a test transformer, a transformer for instruments, a special purpose transformer, etc., and the server 104 may be implemented by a stand-alone server or a server cluster composed of a plurality of servers.

In one embodiment, as shown in fig. 2, a method for detecting a state of a transformer is provided, and the method is applied to the server in fig. 1 for illustration, and includes the following steps:

Step S202, acquiring acoustic image information of a transformer to be detected.

Specifically, acoustic imaging equipment is utilized to acquire acoustic signals on the surface of the oil tank of the transformer to be detected, and corresponding acoustic image information is obtained. The obtained acoustic image information of the oil tank surface of the transformer to be detected can represent the internal mechanical state of the transformer to be detected, and the acoustic image information of the transformer winding in different states, such as a normal state and a fault state, is different.

Further, an acoustic imaging device shown in fig. 3 is adopted to collect acoustic signals on the surface of the oil tank of the transformer to be detected, wherein the acoustic imaging device is specifically based on a microphone array measurement technology, the signal phase difference of sound waves in a certain space reaching each microphone is measured, the position of the sound source is determined according to a phased array principle, the amplitude of the sound source is measured, the distribution of the sound source in space is displayed in an image mode, a spatial sound field distribution cloud image, namely an acoustic image, is obtained, and the color and the brightness of the image represent the intensity of sound.

The array of the acoustic imaging device may include, but is not limited to, an S-type array as shown in fig. 3, the acquisition front end performs acoustic signal acquisition on the surface of the oil tank of the transformer to be detected based on the S-type array as shown in fig. 3, and performs recognition processing on the acquired acoustic signal through sound source recognition software to obtain acoustic image information.

Step S204, gray processing is carried out on the acoustic image information, and a corresponding gray image to be detected is obtained.

Specifically, the acoustic image information is subjected to gray scale processing by adopting an average value method, and the acoustic image information is converted into a gray scale image to obtain a corresponding gray scale image to be detected.

Further, as shown in fig. 4, there is provided a schematic diagram of a gray image to be detected obtained by gray processing of acoustic image information, referring to fig. 4, by converting each obtained acoustic image information into a gray image and compressing the gray image into an image with a pixel size of 64×64, wherein the average method is to average three values of the same pixel by RGB, average three brightness in a color image, and obtain a gray value, and specifically, the method is calculated by adopting the following formula (1):

RGB represents three colors of red, green and blue, x and y represent the number of pixels of a pixel, and different colors can be synthesized according to different ratios by using the three colors of red, green and blue, if r=g=b, a gray color is represented, only intensity information is represented, no color information is represented, the corresponding value of r=g=b is called a gray value, and the conversion process of an image is called graying process.

Step S206, extracting texture characteristic values of the gray level image to be detected.

Specifically, a central pixel point of a gray image to be detected is obtained, and gray values of a preset number of pixel points adjacent to the central pixel point are obtained from the gray image to be detected. And then, according to a preset rotation function, carrying out circular right shift on the preset number of pixel points to obtain texture feature vectors corresponding to the preset number of pixel points, determining a minimum texture feature vector from the texture feature vectors corresponding to the preset number of pixel points, and further determining the minimum texture feature vector as a texture feature value of the gray level image to be detected.

Further, a local binary pattern operator (i.e., LBP operator, local Binary Patterns) is adopted to extract the texture feature vector of the gray image to be detected, the detection window is divided into a plurality of small areas, one pixel in each area is compared with the gray value of the adjacent preset number of pixels, if the surrounding pixel value is greater than the central pixel value, the position of the pixel is marked as 1, otherwise, the position of the pixel is marked as 0. And then, according to a preset rotation function, circularly right-shifting the preset number of pixels in the field to obtain texture feature vectors corresponding to the preset number of pixels.

After the preset number of pixels in the adjacent area are compared, binary modes corresponding to the preset number can be generated, and the LBP value of the pixel in the center of the window can be obtained. Whereas an LBP operator can generate different binary patterns, an LBP operator containing P sample points for a circular region of radius R will generate 2 ^P And (3) reducing the dimension of the mode type of the generated LBP operator along with the increase of the number of pixel points in the field, determining the minimum texture feature vector from the texture feature vectors subjected to dimension reduction, and determining the minimum texture feature vector as the texture feature value of the gray level image to be detected.

And step S208, verifying the texture characteristic value according to the theoretical characteristic value to generate a verification result.

Specifically, the difference degree of the theoretical characteristic value and the texture characteristic value is calculated, a preset difference threshold value is obtained, and then a verification result of the texture characteristic value is generated according to the difference degree and the preset difference threshold value. The verification result comprises the difference degree of the theoretical characteristic value and the texture characteristic value, wherein the difference degree is larger than a preset difference threshold value, or the difference degree of the theoretical characteristic value and the texture characteristic value is smaller than the preset difference threshold value.

Further, the following formula (2) corresponding to chi-square verification is adopted to realize verification of texture characteristic values according to theoretical characteristic values:

wherein T is a theoretical characteristic value, A is a texture characteristic value and χ is ² Is the degree of difference between the theoretical eigenvalue and the textural eigenvalue. For chi-square verification, a plurality of theoretical characteristic values can be set, the results of the plurality of chi-square values obtained through calculation are synthesized, and the degree of difference between the theoretical characteristic values and the texture characteristic values is determined.

In one embodiment, the theoretical feature value includes a first theoretical feature value determined from a positive sample training set and a second theoretical feature value determined from a negative sample training set, and by calculating a first difference degree between the first theoretical feature value and the texture feature value by using the above formula (2), and comparing the first difference degree with a preset difference threshold value, a verification result that the difference degree between the first theoretical feature value and the texture feature value is greater than the preset difference threshold value, or the difference degree between the first theoretical feature value and the texture feature value is less than the preset difference threshold value is obtained.

Similarly, for the second theoretical feature value, by calculating a second difference degree between the second theoretical feature value and the texture feature value by adopting the formula (2), and comparing the second difference degree with a preset difference threshold value, a verification result that the difference degree between the second theoretical feature value and the texture feature value is greater than the preset difference threshold value or the difference degree between the second theoretical feature value and the texture feature value is less than the preset difference threshold value can be obtained.

Step S210, determining the mechanical state of the transformer to be detected according to the verification result.

Specifically, for the first theoretical characteristic value, when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state. And conversely, when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a normal state.

Further, for the second theoretical characteristic value, when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, the mechanical state of the transformer to be detected is determined to be a normal state. And conversely, when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than the preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state.

In one embodiment, the first theoretical feature value is determined from the positive sample training set, and the positive sample training set includes the first sample acoustic image information of the sample transformer after load current adjustment in a normal state, when the difference degree between the texture feature value and the first theoretical feature value is greater than a preset difference threshold value, it is indicated that the difference between the texture feature value and the first theoretical feature value is greater, that is, the difference degree is greater than the first sample acoustic image information of the sample transformer after load current adjustment in a normal state, and then it is determined that the mechanical state of the transformer to be detected is a fault state.

And when the difference degree of the texture characteristic value and the first theoretical characteristic value is smaller than a preset difference threshold value, the fact that the difference between the texture characteristic value and the first theoretical characteristic value is smaller, namely the texture characteristic value and the first sample acoustic image information of the sample transformer after load current adjustment in a normal state, is indicated, and if the difference degree is smaller, the mechanical state of the transformer to be detected is determined to be the normal state.

Similarly, since the second theoretical characteristic value is determined from the negative sample training set, the negative sample training set includes the second sample acoustic image information of the sample transformer after load current adjustment in the fault state. And when the difference degree of the texture characteristic value and the second theoretical characteristic value is smaller than a preset difference threshold value, the fact that the difference between the texture characteristic value and the second theoretical characteristic value is smaller is indicated, namely, the texture characteristic value and the second sample acoustic image information of the sample transformer after load current adjustment in a fault state are smaller, and the mechanical state of the transformer to be detected is determined to be the fault state.

And when the difference degree of the texture characteristic value and the second theoretical characteristic value is larger than a preset difference threshold value, the fact that the difference between the texture characteristic value and the second theoretical characteristic value is larger, namely second sample acoustic image information after load current adjustment is carried out on the texture characteristic value and the sample transformer in a fault state, is indicated, and if the difference degree is larger, the mechanical state of the transformer to be detected is determined to be a normal state.

In the transformer state detection method, the corresponding gray level image to be detected is obtained by acquiring the acoustic image information of the transformer to be detected and carrying out gray level processing on the acoustic image information. The texture characteristic value of the gray level image to be detected is extracted, the texture characteristic value is verified according to the theoretical characteristic value, a verification result is generated, and then the mechanical state of the transformer to be detected is determined according to the verification result. The method realizes the acquisition of visual acoustic image information of the transformer to be detected and the accurate verification of texture characteristic values of corresponding acoustic information, so as to realize the rapid distinction of faults or normal states of the transformer to be detected, and improve the accuracy of mechanical fault detection results of the transformer.

In one embodiment, as shown in fig. 5, extracting texture feature values of a gray image to be detected includes the following steps:

step S502, a center pixel point of a gray image to be detected is obtained.

Specifically, a local binary pattern operator, namely an LBP operator, is adopted to obtain central pixel points of all gray images to be detected, wherein the central pixel points of all gray images to be detected belong to pixel points which can be determined in advance, and the central pixel points can be determined in advance according to the original size of the images to be detected. The local binary pattern operator is a texture description operator and is used for measuring and extracting the texture information of the image part.

Step S504, obtaining gray values of a preset number of pixels adjacent to the center pixel from the detected gray image.

Specifically, by acquiring a preset number P, such as a gray value of p=8 pixels, adjacent to the center pixel from the gray image to be detected.

Step S506, according to the preset rotation function, the preset number of pixels are circularly shifted to the right, and the texture feature vector corresponding to the preset number of pixels is obtained.

Specifically, the gray value of the central pixel point is compared with the gray values of the adjacent preset number of pixel points, the gray value of the central pixel point is set as a threshold value, a neighborhood is selected to be compared clockwise or anticlockwise, if a certain pixel is larger than the central pixel point, the position of the pixel point is marked as 1, and otherwise, the position of the pixel point is marked as 0.

Wherein the following formula (3) is used to represent the corresponding labels of the respective pixel points:

if a pixel is greater than the central pixel value, the position of the pixel point is marked as 1, otherwise, the position is marked as 0.

Further, the expression of the preset rotation function is as shown in the following formula (4):

wherein,the LBP operator, i.e., the local binary pattern operator, representing rotational invariance, ROR (x, i) is a rotation function representing right shifting x-cycles by i bits, as shown in FIG. 6, providing a rotational invariance diagram of a local binary pattern, as can be seen with reference to FIG. 6.

In one embodiment, when the LBP operator is used to extract the texture feature vector, after comparing the preset number of pixels in the neighborhood, a binary pattern corresponding to the preset number can be generated, so as to obtain the LBP value of the pixel in the center of the window. Whereas an LBP operator can generate different binary patterns, an LBP operator containing P sample points for a circular region of radius R will generate 2 ^P The pattern, along with the increase of the number of pixels in the field, the binary pattern type increases sharply, so that the pattern type of the generated LBP operator needs to be reduced in dimension.

The following formula (5) is adopted to realize the dimension reduction processing of the mode type of the generated LBP operator:

wherein p is the number of pixels in the acquired field, U (Gp) is used for judging whether the mode is equivalent mode, s (g _p-1 -g _c )、s(g ₀ -g _c ) Etc. are used to represent gray scale invariance. If U (Gp) is less than or equal to 2, it can be attributed to the equivalent mode, g in formula (5) _p To surround the pixel point g _c And gray values corresponding to p pixel points distributed equidistantly.

In one embodiment, the texture T is defined as a joint distribution of image pixel gray levels within a local neighborhood of a black and white texture image: t=t (g) _c ,g ₀ ,…g _p-1 ) G is then _c Gray value g corresponding to center pixel of local neighborhood _p (p=1, 2.,. P-1) corresponds to a pixel gray level of P equal division within a circularly symmetric neighborhood of radius R.

Further, by integrating the center pixel point g in the neighborhood _c Gray value of (2) and other points g in the circularly symmetric region _p Is subtracted from the gray value of (1) and assumes a difference g _p -g _c And g _c Independent of each other, T.apprxeq.t (g ₀ -g _c ,g ₁ -g _c ,…,g _p-1 -g _c ) And signed difference g _p -g _c Independent of the average light source variation, the gray scale invariance is represented by using only the sign of the difference shown in the following equation (6):

T≈t(s(g ₀ -g _c ),s(g ₁ -g _c ),…,s(g _p-1 -g _c ))； (6)

step S508, determining the minimum texture feature vector from the texture feature vectors corresponding to the preset number of pixel points, and determining the minimum texture feature vector as the texture feature value of the gray image to be detected.

Specifically, a minimum texture feature vector is determined from texture feature vectors corresponding to a preset number of pixel points after dimension reduction, and the minimum texture feature vector is determined to be a texture feature value of a gray image to be detected.

In this embodiment, the gray value of the preset number of pixels adjacent to the center pixel is obtained from the detected gray image by obtaining the center pixel of the gray image to be detected. And then, according to a preset rotation function, carrying out circular right shift on the preset number of pixel points to obtain texture feature vectors corresponding to the preset number of pixel points, determining a minimum texture feature vector from the texture feature vectors corresponding to the preset number of pixel points, and further determining the minimum texture feature vector as a texture feature value of the gray level image to be detected. The texture characteristic value extraction is carried out by adopting the local binary pattern operator with rotation invariance, so that good display effect of the extracted texture characteristic is ensured, calculation errors and the like caused by poor display effect can be avoided when the difference degree calculation and the verification are further carried out according to the texture characteristic, and the accuracy of judging the mechanical state of the transformer to be detected according to the verification result is improved.

In one embodiment, as shown in fig. 7, a method for detecting a state of a transformer is provided, which specifically includes the following steps:

step S702, obtaining acoustic image information of the transformer to be detected, and carrying out gray processing on the acoustic image information to obtain a corresponding gray image to be detected.

Specifically, acoustic imaging equipment is used for acquiring acoustic signals on the surface of an oil tank of a transformer to be detected, corresponding acoustic image information is obtained, gray processing is carried out on the acoustic image information by an average value method, the acoustic image information is converted into gray images, and corresponding gray images to be detected are obtained.

In step S704, a texture feature value of the gray-scale image to be detected is extracted.

Step S706, acquiring first sample acoustic image information of the sample transformer after load current adjustment in a normal state.

Specifically, by acquiring a sample transformer which has been determined to be in a normal state, and performing load current adjustment on the sample transformer when the sample transformer is in the normal state, specifically including taking 0.8A as a step length, acoustic images of the sample transformer are acquired in states in which the load currents are adjusted to 4A, 4.8A, 5.6A, 6.4A, 7.2A, and 8A, respectively. And the horizontal distance between the microphone and the oil tank surface of the sample transformer is 1m, and acoustic images of 50Hz, 100Hz, 200Hz and 300Hz are respectively acquired under the conditions that the load currents are adjusted to be 4A, 4.8A, 5.6A, 6.4A, 7.2A and 8A, so that the acoustic image information of the first sample is obtained.

Step S708, obtaining second sample acoustic image information of the sample transformer after load current adjustment in the fault state.

Specifically, if the fault states of the sample transformer include winding loosening fault states with different loosening degrees and insulating gasket falling fault states with different falling numbers, the second sample acoustic image information may include: the sample transformer carries out the acoustic image information of each second sample after the load current adjustment for a plurality of times under the winding loosening fault state with different loosening degrees, or carries out the acoustic image information of each second sample after the load current adjustment for a plurality of times under the insulating gasket shedding fault state with different shedding numbers.

The method comprises the steps of setting winding loosening faults for sample transformers according to winding loosening fault states of different loosening degrees, loosening a screw rod of a fixed winding by using a torque wrench for 3 gears, wherein the gears are respectively 4 N.m, 8 N.m and 12 N.m, obtaining acoustic images with 0.8A as step sizes and respectively adjusting load currents to be 4A, 4.8A, 5.6A, 6.4A, 7.2A and 8A, and obtaining acoustic image information of a second sample, wherein the acoustic images are 50Hz, 100Hz, 200Hz and 300 Hz.

Similarly, for the falling fault states of the insulating spacers with different falling numbers, the acoustic images of 50Hz, 100Hz, 200Hz and 300Hz are obtained by respectively removing 1, 2, 3 and 4 insulating spacers in the sequence from left to right and obtaining the acoustic image information of the second sample under the state that the load current is adjusted to be 4A, 4.8A, 5.6A, 6.4A, 7.2A and 8A.

Step S710, gray processing is performed on the first sample acoustic image information and the second sample acoustic image information to generate a corresponding positive sample training set and negative sample training set.

Specifically, gray processing is performed on the acoustic image information of each first sample to obtain a corresponding positive sample training set, gray processing is performed on the acoustic image information of each second sample to generate a corresponding negative sample training set. The negative sample training set comprises second sample acoustic image information under winding loosening fault states with different loosening degrees and under insulation gasket falling fault states with different falling numbers.

Step S712, randomly determining a theoretical feature value from the positive sample training set or the negative sample training set.

Specifically, the theoretical feature values include a first theoretical feature value determined from a positive sample training set and a second theoretical feature value determined from a negative sample training set.

In step S714, a degree of difference between the theoretical feature value and the texture feature value is calculated.

Specifically, a first degree of difference between the first theoretical feature value and the texture feature value and a second degree of difference between the second theoretical feature value and the texture feature value are calculated.

Step S716, generating a verification result of the texture characteristic value according to the difference degree and a preset difference threshold value.

Specifically, by acquiring a preset difference threshold and comparing the first difference degree with the preset difference threshold, a verification result that the first difference degree of the first theoretical characteristic value and the texture characteristic value is larger than the preset difference threshold or the first difference degree of the first theoretical characteristic value and the texture characteristic value is smaller than the preset difference threshold can be obtained.

And similarly, comparing the second difference degree with a preset difference threshold value to obtain a verification result that the second difference degree of the second theoretical characteristic value and the texture characteristic value is larger than the preset difference threshold value or smaller than the preset difference threshold value.

Step S718, determining the mechanical state of the transformer to be detected according to the verification result.

Specifically, when a first difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state. And conversely, when the first difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than the preset difference threshold value, determining that the mechanical state of the transformer to be detected is a normal state.

And likewise, when the second difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a normal state. And conversely, when the second difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than the preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state.

In the transformer state detection method, the corresponding gray level image to be detected is obtained by acquiring the acoustic image information of the transformer to be detected and carrying out gray level processing on the acoustic image information. And extracting texture characteristic values of the gray level image to be detected. The method comprises the steps of obtaining first sample acoustic image information of a sample transformer after load current adjustment in a normal state and second sample acoustic image information of the sample transformer after load current adjustment in a fault state, and carrying out gray processing on the first sample acoustic image information and the second sample acoustic image information to generate a corresponding positive sample training set and a corresponding negative sample training set. The theoretical characteristic value is randomly determined from the positive sample training set or the negative sample training set, the texture characteristic value is further verified according to the theoretical characteristic value, a verification result is generated, and the mechanical state of the transformer to be detected is further determined according to the verification result. The method realizes the acquisition of visual acoustic image information of the transformer to be detected and the accurate verification of texture characteristic values of corresponding acoustic information, so as to realize the rapid distinction of faults or normal states of the transformer to be detected, and improve the accuracy of mechanical fault detection results of the transformer.

It should be understood that, although the steps in the flowcharts related to the above embodiments are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least a part of the steps in the flowcharts related to the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages performed is not necessarily sequential, but may be performed alternately or alternately with at least a part of the steps or stages in other steps or other steps.

In one embodiment, as shown in fig. 8, there is provided a transformer state detection apparatus including: an acoustic image information acquisition module 802, a gray image to be detected generation module 804, a texture feature value extraction module 806, a verification result generation module 808, and a mechanical state determination module 810, wherein:

the acoustic image information acquiring module 802 is configured to acquire acoustic image information of the transformer to be detected.

The gray image to be detected generating module 804 is configured to perform gray processing on the acoustic image information to obtain a corresponding gray image to be detected.

The texture feature value extracting module 806 is configured to extract texture feature values of the gray-scale image to be detected.

And the verification result generation module 808 is configured to verify the texture feature value according to the theoretical feature value, and generate a verification result.

The mechanical state determining module 810 is configured to determine a mechanical state of the transformer to be detected according to the verification result.

In the transformer state detection device, the corresponding gray level image to be detected is obtained by acquiring the acoustic image information of the transformer to be detected and carrying out gray level processing on the acoustic image information. The texture characteristic value of the gray level image to be detected is extracted, the texture characteristic value is verified according to the theoretical characteristic value, a verification result is generated, and then the mechanical state of the transformer to be detected is determined according to the verification result. The method realizes the acquisition of visual acoustic image information of the transformer to be detected and the accurate verification of texture characteristic values of corresponding acoustic information, so as to realize the rapid distinction of faults or normal states of the transformer to be detected, and improve the accuracy of mechanical fault detection results of the transformer.

In one embodiment, as shown in fig. 9, there is provided a transformer state detection apparatus including:

the to-be-detected gray level image generating module 902 is configured to obtain acoustic image information of the to-be-detected transformer, and perform gray level processing on the acoustic image information to obtain a corresponding to-be-detected gray level image.

The texture feature value extraction module 904 is configured to extract texture feature values of the gray-scale image to be detected.

The first sample acoustic image information obtaining module 906 is configured to obtain first sample acoustic image information of the sample transformer after load current adjustment in a normal state.

The second sample acoustic image information obtaining module 908 is configured to obtain second sample acoustic image information of the sample transformer after load current adjustment in the fault state.

The positive/negative training set generating module 910 is configured to perform gray processing on the first sample acoustic image information and the second sample acoustic image information, and generate a positive training set and a negative training set.

The theoretical feature value determining module 912 is configured to randomly determine a theoretical feature value from the positive sample training set or the negative sample training set.

The difference degree calculating module 914 is configured to calculate a difference degree between the theoretical feature value and the texture feature value.

And the verification result generating module 916 is configured to generate a verification result of the texture feature value according to the difference degree and the preset difference threshold.

The mechanical state determining module 918 is configured to determine a mechanical state of the transformer to be detected according to the verification result.

In the transformer state detection device, the corresponding gray level image to be detected is obtained by acquiring the acoustic image information of the transformer to be detected and carrying out gray level processing on the acoustic image information. And extracting texture characteristic values of the gray level image to be detected. The method comprises the steps of obtaining first sample acoustic image information of a sample transformer after load current adjustment in a normal state and second sample acoustic image information of the sample transformer after load current adjustment in a fault state, and carrying out gray processing on the first sample acoustic image information and the second sample acoustic image information to generate a corresponding positive sample training set and a corresponding negative sample training set. The theoretical characteristic value is randomly determined from the positive sample training set or the negative sample training set, the texture characteristic value is further verified according to the theoretical characteristic value, a verification result is generated, and the mechanical state of the transformer to be detected is further determined according to the verification result. The method realizes the acquisition of visual acoustic image information of the transformer to be detected and the accurate verification of texture characteristic values of corresponding acoustic information, so as to realize the rapid distinction of faults or normal states of the transformer to be detected, and improve the accuracy of mechanical fault detection results of the transformer.

In one embodiment, the texture feature value extraction module is configured to:

acquiring a central pixel point of a gray level image to be detected; acquiring gray values of a preset number of pixel points adjacent to the central pixel point from the detected gray image; according to a preset rotation function, performing circular right shift on the preset number of pixel points to obtain texture feature vectors corresponding to the preset number of pixel points; and determining a minimum texture feature vector from texture feature vectors corresponding to the preset number of pixel points, and determining the minimum texture feature vector as a texture feature value of the gray image to be detected.

In this embodiment, since the local binary pattern operator with rotation invariance is adopted to extract the texture feature value, the good display effect of the extracted texture feature is ensured, and when the difference degree calculation and the verification are further performed according to the texture feature, calculation errors and the like caused by poor display effect can be avoided, and the accuracy of judging the mechanical state of the transformer to be detected according to the verification result is improved.

In one embodiment, the second sample acoustic image information acquisition module is further configured to:

acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under winding loosening fault states with different loosening degrees; or acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under the falling fault states of insulating gaskets with different falling numbers.

In one embodiment, the verification result generation module is further configured to:

calculating the difference degree of the theoretical characteristic value and the texture characteristic value; and generating a verification result of the texture characteristic value according to the difference degree and a preset difference threshold value.

In one embodiment, the mechanical state determination module is further configured to:

when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state; or when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state.

For specific limitations of the transformer state detection device, reference may be made to the above limitations of the transformer state detection method, and no further description is given here. The above-described respective modules in the transformer state detection apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 10. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing the acoustic image information, the texture characteristic value and the verification result. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a transformer state detection method.

It will be appreciated by those skilled in the art that the structure shown in fig. 10 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:

acquiring acoustic image information of a transformer to be detected;

In one embodiment, the processor when executing the computer program further performs the steps of:

and carrying out gray processing on the acoustic image information of each first sample and the acoustic image information of each second sample to generate a corresponding positive sample training set and a corresponding negative sample training set.

acquiring a central pixel point of a gray level image to be detected;

theoretical characteristic values are randomly determined from the positive sample training set or the negative sample training set.

when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined according to the verification result and is larger than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state;

or when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined according to the verification result and is smaller than a preset difference threshold value, determining that the mechanical state of the transformer to be detected is a fault state.

In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:

acquiring acoustic image information of a transformer to be detected;

In one embodiment, the computer program when executed by the processor further performs the steps of:

acquiring a central pixel point of a gray level image to be detected;

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, or the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.

Claims

1. A method for detecting a state of a transformer, the method comprising:

acquiring acoustic image information of a transformer to be detected;

calculating the difference degree of the theoretical characteristic value and the texture characteristic value, and generating a verification result of the texture characteristic value according to the difference degree and a preset difference threshold value;

determining the mechanical state of the transformer to be detected according to the verification result;

the mechanical state of the transformer includes a normal state and a fault state; the fault states comprise winding loosening fault states with different loosening degrees and insulating gasket falling fault states with different falling numbers; the method further comprises the steps of:

acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under winding loosening fault states with different loosening degrees, or acquiring acoustic image information of each second sample of the sample transformer after multiple load current adjustment under insulating gasket shedding fault states with different shedding numbers;

gray processing is carried out on the first sample acoustic image information and the second sample acoustic image information, and a corresponding positive sample training set and a corresponding negative sample training set are generated; the positive sample training set and the negative sample training set are used for randomly determining theoretical characteristic values;

the theoretical characteristic values comprise first theoretical characteristic values determined from the positive sample training set and second theoretical characteristic values determined from the negative sample training set; and determining the mechanical state of the transformer to be detected according to the verification result, wherein the determining comprises the following steps:

when the difference degree between the first theoretical characteristic value and the texture characteristic value is determined to be larger than the preset difference threshold value according to the verification result, determining that the mechanical state of the transformer to be detected is a fault state; or when the difference degree between the second theoretical characteristic value and the texture characteristic value is determined to be smaller than the preset difference threshold value according to the verification result, determining that the mechanical state of the transformer to be detected is a fault state.

2. The method according to claim 1, wherein the extracting the texture feature value of the gray-scale image to be detected comprises:

acquiring a central pixel point of the gray level image to be detected;

3. The method according to claim 2, further comprising, before said verifying the texture feature value according to the theoretical feature value, generating a verification result:

4. The method of claim 1, wherein the acquiring acoustic image information of the transformer to be inspected comprises: acquiring acoustic signals on the surface of the oil tank of the transformer to be detected by using acoustic imaging equipment, and identifying the acquired acoustic signals by using sound source identification software to obtain corresponding acoustic image information; the acoustic image information is used for reflecting the internal mechanical state of the transformer to be detected, wherein the acoustic image information of the transformer winding in different states is different, and the states of the transformer winding comprise a normal state and a fault state.

5. The method according to claim 2, wherein the acquiring the center pixel of the gray-scale image to be detected comprises:

and acquiring the central pixel point of each gray level image to be detected by adopting a local binary pattern operator.

6. The method according to claim 1, wherein the way of calculating the degree of difference of theoretical feature values from the texture feature values comprises:

calculating a chi-square value of the theoretical characteristic value and the texture characteristic value in a chi-square verification mode;

and setting a plurality of theoretical characteristic values for chi-square verification, and comprehensively calculating the results of the plurality of chi-square values to determine the degree of difference between the theoretical characteristic values and the texture characteristic values.

7. A transformer condition detection apparatus, the apparatus comprising:

the verification result generation module is used for calculating the difference degree of the theoretical characteristic value and the texture characteristic value and generating a verification result of the texture characteristic value according to the difference degree and a preset difference threshold value;

The mechanical state determining module is used for determining the mechanical state of the transformer to be detected according to the verification result;

the mechanical state of the transformer includes a normal state and a fault state; the fault states comprise winding loosening fault states with different loosening degrees and insulating gasket falling fault states with different falling numbers; the apparatus further comprises:

the first sample acoustic image information acquisition module is used for acquiring first sample acoustic image information of the sample transformer after load current adjustment in a normal state;

the second sample acoustic image information acquisition module is used for acquiring each second sample acoustic image information of the sample transformer after carrying out multiple load current adjustment under winding loosening fault states with different loosening degrees or acquiring each second sample acoustic image information of the sample transformer after carrying out multiple load current adjustment under insulating gasket shedding fault states with different shedding numbers;

the positive/negative sample training set generation module is used for carrying out gray processing on the first sample acoustic image information and the second sample acoustic image information to generate a corresponding positive sample training set and a negative sample training set; the positive sample training set and the negative sample training set are used for randomly determining theoretical characteristic values;

The theoretical characteristic values comprise first theoretical characteristic values determined from the positive sample training set and second theoretical characteristic values determined from the negative sample training set; the mechanical state determination module is further configured to:

8. The apparatus of claim 7, wherein the texture feature value extraction module is further configured to:

acquiring a central pixel point of the gray level image to be detected; acquiring gray values of a preset number of pixel points adjacent to the central pixel point from the detected gray image; according to a preset rotation function, performing circular right shift on the preset number of pixel points to obtain texture feature vectors corresponding to the preset number of pixel points; and determining a minimum texture feature vector from texture feature vectors corresponding to the preset number of pixel points, and determining the minimum texture feature vector as a texture feature value of the gray image to be detected.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.