CN106569069A - Power transformer fault diagnosis method - Google Patents

Abstract

The invention discloses a power transformer fault diagnosis method comprising the following steps: building a fault standard reference sequence based on oil-dissolved gas samples of clear fault types; using an improved principal component analysis method to calculate the principal component of the fault standard reference sequence; calculating the entropy and weight of a standard reference sequence principal component matrix; calculating the gray correlation coefficient between a reference matrix and a to-be-tested sample based on a gray correlation analysis method; and calculating the weighted correlation degree between the reference matrix and the to-be-tested sample, and determining the fault type of the to-be-tested sample. The method has the following advantages: the efficiency of key information extraction is improved effectively through the improved principal component analysis method; the weight determined based on an entropy weight method is more objective and scientific; and the fault diagnosis method based on weighted gray correlation analysis effectively makes up for the defects of the traditional ratio method such as code deficiency and relatively low failure rate, and effectively improves the diagnosis accuracy of internal latent faults of power transformers.

Description

Power transformer fault diagnosis method

Technical Field

The invention relates to the technical field of power transformers, in particular to a power transformer fault diagnosis method.

Background

The power transformer is an important and expensive device in the power system, bears the functions of voltage transformation and electric energy distribution, and plays an especially important role in reliable operation of the power system. Various defect faults are inevitable under the combined action of multiple physical fields during the operation of the power transformer; the fault of the transformer is generally wide in spread range, and great loss is brought to power generation and power utilization units, so that great influence is caused, and therefore the method has important significance in finding and processing the fault of the transformer as soon as possible and establishing a high-efficiency and reliable fault diagnosis system.

Among the power transformer fault diagnosis methods, the Dissolved Gas Analysis (DGA) in oil is one of the simplest and most widely used methods. Based on the gas composition, content and gas production rate when the voltage transformer fails, the ratio can be used to diagnose most faults. However, the DGA-based ratio method has the defects of absolute coding boundary, limited coding types, low diagnosis accuracy, inconsistent diagnosis performance and the like; although the intelligent fault diagnosis method (including a support vector machine, a neural network, an expert system, a fuzzy theory and the like) improves the fault diagnosis precision to a certain extent, the intelligent fault diagnosis method still has better performance only in a local aspect.

Disclosure of Invention

The present invention is directed to solving at least one of the above problems.

Therefore, the invention aims to provide a power transformer fault diagnosis method.

In order to achieve the above object, an embodiment of the present invention discloses a power transformer fault diagnosis method, including the following steps: s1: establishing a fault sample library based on the gas sample of the dissolved gas in the oil with definite fault type; s2: carrying out standardization processing on the fault sample library by using a standardized formula, and establishing a fault standard reference sequence S; s3: normalizing a fault standard reference sequence S by using an averaging method, and establishing a reference sequence matrix Z; s4: calculating the eigenvalue and the eigenvector of the reference sequence matrix Z based on a principal component analysis method, and reserving a principal component vector F meeting a preset condition; s5: calculating the entropy value and the weight of the principal component vector F; s6: and calculating a gray correlation coefficient and a weighted correlation degree between the principal component vector F and the sequence X to be detected by using a gray correlation analysis method so as to determine the final fault type.

According to the power transformer fault diagnosis method provided by the embodiment of the invention, key information extraction is effectively improved by using improved principal component analysis; the weight determined based on the entropy weight method is more objective and scientific, the defects of coding loss, relatively low fault rate and the like existing in the traditional ratio method are effectively overcome by the fault diagnosis method based on weighted gray correlation analysis, and the accuracy of latent fault diagnosis inside the power transformer is effectively improved.

In addition, the power transformer fault diagnosis method according to the above embodiment of the present invention may further have the following additional technical features:

further, in step S2, the normalization process is performed by the following formula:

wherein,indicating gaugeDissolved gas volume in normalized oil, x_ijRepresenting a measure of dissolved gas in the oil; n is the number of gas samples dissolved in oil, and m is the number of gas components in the samples.

Further, in step S2, failure types including No Failure (NF), Partial Discharge (PD), low energy discharge (LD), high energy discharge (HD), low temperature superheat (LT), medium temperature superheat (MT), high temperature superheat (HT), low energy discharge and superheat (LTD), and high energy discharge and superheat (HTD) are selected according to the DLT 722-2000 dissolved gas analysis and judgment guide.

Further, in step S3, the values in the fault standard reference sequence S are normalized by the following formula, and a standard fault reference sequence matrix Z is established:

in the formula,to normalize matrix data, x'_ijIn order to further process the data after the normalized matrix by using an averaging method, p is the sample number of the q fault type, and m is the component number of the dissolved gas in the oil.

Further, in step S4, the correlation coefficient matrix R, the eigenvalue λ, and the eigenvector a are calculated for the matrix Z by the following formulas, respectively:

calculating a characteristic equation | λ I-R | ═ 0 by using a Jacobian method, and then obtaining a characteristic value λ and a characteristic vector a; wherein r is_ijFor the correlation coefficients that are solved after the sample is normalized,is as followsMean value of the characteristic quantity, x_ijThe matrix data obtained after the averaging processing.

Further, in step S4, the contribution rate K is calculated_rAnd cumulative contribution rate K_tAnd selecting the cumulative contribution rate K_tAnd (3) constructing a principal component matrix F when the feature vector is not less than the feature vector corresponding to the threshold, wherein the calculation formula of the cumulative contribution rate is as follows:

K_t>。

further, in step S5, F is processed by nonnegativity and matrix F is obtained^*The calculation formula of (a) is as follows:

F^*＝F+g

wherein g is max [ f ═ g [ [ f ]_ij]，f_ijRepresenting the amount of the principal component.

Further, in step S5, the entropy value e in each principal component_tInformation utility value d_tAnd weight ω_tCalculated using the following formula respectively:

d_t＝1-e_t

in the formula y_ijFor data in the non-negatively processed principal component matrix,f_ij' denotes the nonnegatively processed principal component matrix data.

Further, in step S6, a formula for calculating the gray correlation coefficient ξ and the degree γ by the gray correlation method is shown by the following equation:

Δ_oi(j)＝|X_o(j)-X_i(j)|

in the formula ξ_oiAs a principal component reference sequence X_oAnd comparison of sequence X_iThe coefficient of correlation between the two or more,to evaluate the minimum value in the data, andis the maximum value of the evaluation data; rho is a resolution coefficient, rho is more than or equal to 0 and less than or equal to 1, and the selection method comprises the following steps:

while remembering(i)＝Δv(i)/Δ_maxThe dynamic resolution coefficient takes the following values:

the calculation formula of the relevance degree with the weight is as follows:

when in useAnd judging that the sample data to be tested is the ith fault type.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

fig. 1 is a flowchart of a power transformer fault diagnosis method according to an embodiment of the present invention.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.

In the description of the present invention, it is to be understood that the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

These and other aspects of embodiments of the invention will be apparent with reference to the following description and attached drawings. In the description and drawings, particular embodiments of the invention have been disclosed in detail as being indicative of some of the ways in which the principles of the embodiments of the invention may be practiced, but it is understood that the scope of the embodiments of the invention is not limited correspondingly. On the contrary, the embodiments of the invention include all changes, modifications and equivalents coming within the spirit and terms of the claims appended hereto.

The invention is described below with reference to the accompanying drawings.

Fig. 1 is a flow chart of a power transformer fault diagnosis method according to an embodiment of the present invention. As shown in fig. 1, a method for diagnosing a fault of a power transformer includes the following steps:

s1: and collecting gas samples of dissolved gas in oil with definite fault types to establish a fault sample library.

S2: and carrying out normalization processing on the fault sample library by using a standardized formula, and establishing a fault standard reference sequence S.

In one embodiment of the invention, the normalization process is performed by the following formula:

wherein,represents the volume of dissolved gas in the normalized oil, x_ijRepresenting a measure of dissolved gas in the oil; n is the number of gas samples dissolved in oil, and m is the number of gas components in the samples.

In one embodiment of the present invention, the failure types are selected according to the analysis and judgment guide rule of DLT 722 & 2000 dissolved gas in transformer oil, and include No Failure (NF), Partial Discharge (PD), low energy discharge (LD), high energy discharge (HD), low temperature superheat (LT), medium temperature superheat (MT), high temperature superheat (HT), low energy discharge and superheat (LTD) and high energy discharge and superheat (HTD).

S3: and normalizing the fault standard reference sequence S by using an averaging method, and establishing a reference sequence matrix Z.

In one embodiment of the invention, the values in the fault reference sequence S are normalized using the following equation and a standard fault reference sequence matrix Z is established:

in the formula,to normalize matrix data, x'_ijIn order to further process the data after the normalized matrix by using an averaging method, p is the sample number of the q fault type, and m is the component number of the dissolved gas in the oil. The normalization is carried out by adopting an averaging method, so that the difference between different index quantities can be effectively realized, the correlation information can be kept, and the original fault information can be reflected more accurately.

S4: and calculating the eigenvalue and the eigenvector of the reference sequence matrix Z based on a principal component analysis method, and reserving a principal component vector F meeting a preset condition.

In one embodiment of the present invention, the correlation coefficient matrix R, the eigenvalue λ and the eigenvector a are calculated for the matrix Z by the following formulas, respectively:

calculating a characteristic equation | λ I-R | ═ 0 by using a Jacobian method, and then obtaining a characteristic value λ and a characteristic vector a; wherein r is_ijFor the correlation coefficients that are solved after the sample is normalized,is the mean value of the characteristic quantities of the sample, x_ijThe matrix data obtained after the averaging processing.

In one embodiment of the invention, the contribution rate K is calculated_rAnd cumulative contribution rate K_tAnd selecting the cumulative contribution rate K_tAnd (3) constructing a principal component matrix F when the feature vector is not less than the feature vector corresponding to the threshold, wherein the calculation formula of the cumulative contribution rate is as follows:

K_t>。

s5: and calculating the entropy value and the weight of the principal component vector F.

In one embodiment of the invention, to avoid assignment during entropy calculation, F is processed nonnegatively and a matrix F is obtained^*The calculation formula of (a) is as follows:

F^*＝F+g

wherein g is max [ f ═ g [ [ f ]_ij]，f_ijRepresenting the amount of the principal component.

In one embodiment of the invention, the entropy value e in each principal component_tInformation utility value d_tAnd weight ω_tCalculated using the following formula respectively:

d_t＝1-e_t

in the formula y_ijFor data in the non-negatively processed principal component matrix,f_ij' denotes the nonnegatively processed principal component matrix data.

S6: and calculating a gray correlation coefficient and a weighted correlation degree between the principal component vector F and the sequence X to be detected by using a gray correlation analysis method so as to determine the final fault type.

In one embodiment of the present invention, a formula for calculating the gray correlation coefficient ξ and the degree γ by the gray correlation method is shown as follows:

Δ_oi(j)＝|X_o(j)-X_i(j)|

in the formula ξ_oiAs a principal component reference sequence X_oAnd comparison of sequence X_iThe coefficient of correlation between the two or more,to evaluate the minimum value in the data, andis the maximum value of the evaluation data; rho is a resolution coefficient, rho is more than or equal to 0 and less than or equal to 1,the selection method comprises the following steps:

while remembering(i)＝Δv(i)/Δ_maxThe dynamic resolution coefficient takes the following values:

the calculation formula of the relevance degree with the weight is as follows:

when in useAnd judging that the sample data to be tested is the ith fault type.

According to the power transformer fault diagnosis method, the key information can be further effectively extracted by adopting an improved principal component method, and redundant fault information is eliminated; then, the weight of each principal component is objectively calculated by using an entropy weight method, so that the reasonability and the scientificity of calculation are improved; and finally, accurately calculating the gray correlation coefficient among the sequences by using an improved gray correlation analysis method and a dynamic resolution coefficient, and effectively improving the accuracy of transformer fault diagnosis.

In addition, other structures and functions of the power transformer fault diagnosis method according to the embodiment of the present invention are known to those skilled in the art, and are not described in detail for reducing redundancy.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims (9)

1. A power transformer fault diagnosis method is characterized by comprising the following steps:

s1: establishing a fault sample library based on the gas sample of the dissolved gas in the oil with definite fault type;

s2: carrying out standardization processing on the fault sample library by using a standardized formula, and establishing a fault standard reference sequence S;

s3: normalizing a fault standard reference sequence S by using an averaging method, and establishing a reference sequence matrix Z;

s4: calculating the eigenvalue and the eigenvector of the reference sequence matrix Z based on a principal component analysis method, and reserving a principal component vector F meeting a preset condition;

s5: calculating the entropy value and the weight of the principal component vector F;

s6: and calculating a gray correlation coefficient and a weighted correlation degree between the principal component vector F and the sequence X to be detected by using a gray correlation analysis method so as to determine the final fault type.

2. The power transformer fault diagnosis method according to claim 1, wherein in step S2, the normalization process is performed by the following formula:

$x_{ij}^{*}=x_{ij}/\underset{j=1}{\overset{m}{\Σ}}{x}_{ij}(i=1,2,...,n;j=1,2...m)$

wherein,represents the volume of dissolved gas in the normalized oil, x_ijRepresenting a measure of dissolved gas in the oil; n is the number of gas samples dissolved in oil, and m is the number of gas components in the samples.

3. The method of claim 1, wherein in step S2, the fault types are selected according to the "analysis and judgment guide rule for dissolved gas in DLT 722-2000 transformer oil", and include No Fault (NF), Partial Discharge (PD), low energy discharge (LD), high energy discharge (HD), low temperature overheat (LT), medium temperature overheat (MT), high temperature overheat (HT), low energy discharge and overheat (LTD), and high energy discharge and overheat (HTD).

4. The power transformer fault diagnosis method according to claim 1, characterized in that in step S3, the values in the fault standard reference sequence S are normalized by the following formula, and a standard fault reference sequence matrix Z is established:

$x_{ij}^{,}=x_{ij}^{*}/\underset{i=1}{\overset{p}{\Σ}}{x}_{ij}^{*}(i=1,2,...,p;j=1,2,...,m)$

in the formula,to normalize the matrix data, x_i,_jFor further processing normalization by averagingAnd (3) the data after the matrix, p is the number of samples of the qth fault type, and m is the fraction of dissolved gas in the oil.

5. The power transformer fault diagnosis method according to claim 1, wherein in step S4, the correlation coefficient matrix R, the eigenvalue λ and the eigenvector a are calculated for the matrix Z by the following formulas, respectively:

$R=\left[\begin{array}{cccc}{r}_{11}& r_{12}& \mathrm{...}& r_{1p}\\ r_{21}& r_{22}& \mathrm{...}& r_{3p}\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ r_{n1}& r_{n2}& \mathrm{...}& r_{np}\end{array}\right]r_{ij}=\frac{\underset{k=1}{\overset{n}{\Σ}}(x_{ki}-{\stackrel{\‾}{x}}_i)(x_{kj}-{\stackrel{\‾}{x}}_j)}{\sqrt{\underset{k=1}{\overset{n}{\Σ}}{(x_{ki}-{\stackrel{\‾}{x}}_i)}^2\underset{k=1}{\overset{n}{\Σ}}{(x_{kj}-{\stackrel{\‾}{x}}_j)}^2}}$

calculating a characteristic equation | λ I-R | ═ 0 by using a Jacobian method, and then obtaining a characteristic value λ and a characteristic vector a; wherein r is_ijFor the correlation coefficients that are solved after the sample is normalized,is the mean value of the characteristic quantities of the sample, x_ijThe matrix data obtained after the averaging processing.

6. The power transformer fault diagnosis method according to claim 5, wherein in step S4, the contribution ratio K is calculated_rAnd cumulative contribution rate K_tAnd selecting the cumulative contribution rate K_tAnd (3) constructing a principal component matrix F when the feature vector is not less than the feature vector corresponding to the threshold, wherein the calculation formula of the cumulative contribution rate is as follows:

$K_r=100\%\×{\mathrm{\λ}}_r/\underset{i=1}{\overset{p}{\Σ}}{\mathrm{\λ}}_i,i=1,2,...,p$

$K_t=\underset{i=1}{\overset{m}{\&Sigma;}}{\mathrm{\&lambda;}}_i/\underset{i=1}{\overset{p}{\&Sigma;}}{\mathrm{\&lambda;}}_i,i=1,2,...,p;m<p$

K_t>。

7. the power transformer fault diagnosis method according to claim 1, characterized in that in step S5, F is processed through nonnegativity and matrix F is obtained^*The calculation formula of (a) is as follows:

F^*＝F+g

wherein g is max [ f ═ g [ [ f ]_ij]，f_ijRepresenting the amount of the principal component.

8. The power transformer fault diagnosis method according to claim 1, wherein in step S5, the entropy e of each principal component_tInformation utility value d_tAnd weight ω_tCalculated using the following formula respectively:

$e_t=-1/\left(\ln p\underset{j=1}{\overset{p}{\&Sigma;}}{y}_{ij}\ln y_{ij}\right)$

d_t＝1-e_t

${\mathrm{\&omega;}}_t=d_t/\underset{t=1}{\overset{p}{\&Sigma;}}{d}_t$

in the formula y_ijFor data in the non-negatively processed principal component matrix,f’_ijrepresenting the nonnegatively processed principal component matrix data.

9. The power transformer fault diagnosis method according to claim 1, wherein in step S6, the formula for calculating the gray correlation coefficient ξ and the degree γ by the gray correlation method is as follows:

${\mathrm{\&xi;}}_{oi}=\underset{i}{\min }\underset{j}{\min }{\mathrm{\&Delta;}}_{oi}\left(j\right)+\mathrm{\&rho;}\underset{i}{\min }\underset{j}{\min }{\mathrm{\&Delta;}}_{oi}\left(j\right)/({\mathrm{\&Delta;}}_{oi}\left(j\right)+\mathrm{\&rho;}\underset{i}{\max }\underset{j}{\max }{\mathrm{\&Delta;}}_{oi}(j\left)\right)$

Δ_oi(j)＝|X_o(j)-X_i(j)|

${\mathrm{\&Delta;}}_{min}=\underset{i}{min}\underset{j}{min}{\mathrm{\&Delta;}}_{oj}\left(i\right)$

${\mathrm{\&Delta;}}_{max}=\underset{i}{max}\underset{j}{max}{\mathrm{\&Delta;}}_{oj}\left(i\right)$

in the formula ξ_oiAs a principal component reference sequence X_oAnd comparison of sequence X_iThe coefficient of correlation between the two or more,to evaluate the minimum value in the data, andis the maximum value of the evaluation data; rho is a resolution coefficient, rho is more than or equal to 0 and less than or equal to 1, and the selection method comprises the following steps:

while remembering(i)＝Δv(i)/Δ_maxThe dynamic resolution coefficient takes the following values:

$\mathrm{\&rho;}=\left\{\begin{array}{cc}1.5\mathrm{\&epsiv;}\mathrm{\&Delta;}\left(i\right),& {\mathrm{\&Delta;}}_{\max }>3\mathrm{\&Delta;}v\left(i\right)\\ 2\mathrm{\&epsiv;}\mathrm{\&Delta;}\left(i\right),& 1.5\mathrm{\&epsiv;}\mathrm{\&Delta;}\left(i\right)\&le;{\mathrm{\&Delta;}}_{\max }\&le;3\mathrm{\&Delta;}v\left(i\right)\\ 0.8,& 0<{\mathrm{\&Delta;}}_{\max }\&le;2\mathrm{\&Delta;}v\left(i\right)\end{array}\right.$

the calculation formula of the relevance degree with the weight is as follows:

${\mathrm{\&gamma;}}_i=\underset{i=1}{\overset{p}{\&Sigma;}}\mathrm{\&omega;}\left(i\right){\mathrm{\&xi;}}_{oi}\left(i\right)$

when in useThen, the sample data to be tested is judged to be the ith typeThe type of failure.