CN105242155A - Transformer fault diagnosis method based on entropy weight method and grey correlation analysis - Google Patents

Abstract

The invention discloses a transformer fault diagnosis method based on an entropy weight method and grey correlation analysis, belonging to the technical field of transformer fault diagnosis. The transformer fault diagnosis method comprises the steps of: carrying out standardization processing on transformer fault sample data; adopting the entropy weight method to determine weight of a transformer fault diagnosis index; and determining a fault type of a transformer through calculating Grey Euclid weighted correlation degree. The transformer fault diagnosis method makes full use of all information of gas in oil data, exerts the advantage that the grey correlation is applicable to a small sample and poor information system, avoids defects of partial correlation and information loss, and can effectively increase accuracy rate of transformer fault diagnosis as shown by results of example analysis.

Description

A kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis

Technical field

The present invention relates to a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis, belong to transformer fault diagnosis technical field.

Background technology

Power transformer is the major equipment in electric system, uses large in number and widely distributed, and its reliability service is the basis of power system safety and stability.Due to many-sided reason, the fault of transformer happens occasionally, and the fault diagnosis of power transformer is exactly to find out its inside Hidden fault timely and accurately.

Dissolved gas analysis (DGA) is the main method of current diagnosis transformer fault, and what be most widely used in DGA is IEC tri-ratio.Three-ratio method has simple, practical advantage, but there is the defect that coding lacks, encoded boundary is too absolute in actual applications, in addition when characteristic gas does not arrive demand value, can not use three-ratio method, therefore the accuracy rate of its fault diagnosis need to improve.Recently along with the development of artificial intelligence technology, rough set theory, neural network, Bayesian network, genetic algorithm and other intelligent methods are applied in transformer fault diagnosis.Although these algorithms improve the accuracy rate of transformer fault diagnosis, all need a large amount of fault sample data, but will to obtain a large amount of typical fault sample in the work of electrical network actual motion be very difficult.

Utilize the weight in entropy assessment determination Weighted Grey related degree model, belong to objective weighted model, the information of DGA data can be made full use of, eliminate the impact of subjective factor.Grey correlation analysis has advantage for the system of process small sample, poor information, and the incomplete rate of accurateness of diagnostic message is higher, therefore grey correlation analysis can be applied to transformer fault diagnosis.General related degree model takes equal Quan Huo expert to compose power usually when carrying out calculation of relationship degree, weight is comparatively large by subjective impact, and entropy is measuring of stochastic variable unascertained information, utilizes entropy assessment determination weight can reflect the state of system more objectively.The present invention proposes one and first carries out standardization to transformer fault sample data, then adopts the weight of entropy assessment determination transformer fault diagnosis index, then passes through the method for the fault type calculating grey Euclid weighted association degree determination transformer.Institute's extracting method can effectively improve transformer fault diagnosis accuracy rate.

Summary of the invention

Object: in order to overcome the deficiencies in the prior art, the invention provides a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis.

Technical scheme: for solving the problems of the technologies described above, the technical solution used in the present invention is:

Based on a Diagnosis Method of Transformer Faults for entropy assessment and grey correlation analysis, comprise the steps:

Step one: set up transformer fault state;

Step 2: sample data standardization;

Step 3: entropy assessment agriculture products weight;

Step 4: calculate grey Euclid weighted association degree;

Step 5: transformer fault type analysis.

Describedly set up transformer fault state, five kinds of crucial hydrocarbon gas H in being tested by oil chromatography ₂, CH ₄, C ₂h ₂, C ₂h ₄, C ₂h ₆as characteristic gas, transformer fault is divided into the double overheated eight kinds of malfunctions of non-fault, shelf depreciation, low-yield electric discharge, high-energy discharge, cryogenic overheating t ﹤ 300 DEG C, middle temperature overheated t ﹤ 700 DEG C, hyperthermia and superheating t ﹥ 700 DEG C and electric discharge.

Described sample data standardization, carries out nondimensionalization process to sample data, the mode of process be in each fault sample every gas content all divided by the maximal value of gas content in this fault sample.That is:

${x_i}^{\′}=\frac{x_i}{x_{\max }}$

In formula: x ' _ifor the value after fault sample standardization, x _ifor the content of gas in fault sample, x _maxfor the maximal value of gas content in fault sample.

Described entropy assessment agriculture products weight, if transformer has m kind fault mode, often kind of fault mode has n index, formed m × n exponent number according to matrix A '=[a ' _ij] _{m × n}, wherein a ' _ijrepresent the numerical value of a jth index in i-th kind of fault, carry out standardization to fault data, the mode of process is that under each malfunction, every gas content, all divided by the maximal value of gas content under this malfunction, obtains A=[a after standardization _ij] _{m × n}.The entropy of a jth index is:

$\begin{array}{cc}{E}_j=\mathrm{\α}\·\underset{i=1}{\overset{m}{\Σ}}{b}_{ij}{\mathrm{lnb}}_{ij}& j=1,2,\mathrm{...}n\end{array}$

Wherein, work as b _ijwhen=0, make b _ijlnb _ij=0, the entropy power of a jth index is:

$\begin{array}{cc}{\mathrm{\&omega;}}_j=(1-{\stackrel{\&OverBar;}{E}}^{35}){\mathrm{\&omega;}}_j^{\&prime;}+{\stackrel{\&OverBar;}{E}}^{35}{\mathrm{\&omega;}}_j^{\&prime;\&prime;}& E_j<1\end{array}$

ω _j＝0E _j＝1

Wherein ${\mathrm{\&omega;}}_j^{\&prime;}=\frac{1-E_j}{\underset{j=1}{\overset{n}{\&Sigma;}}(1-E_j)},{\mathrm{\&omega;}}_j^{\&prime;\&prime;}=\frac{1+\stackrel{\&OverBar;}{E}-E_j}{\underset{j=1,E_j\&NotEqual;1}{\overset{n}{\&Sigma;}}(1+\stackrel{\&OverBar;}{E}-E_j)},$ for all be not the mean value of the entropy of 1.

Described calculating grey Euclid weighted association degree, if transformer has m kind fault mode, often kind of fault mode has n index, then its reference sequence is designated as x ₁, x ₂..., x _m, compare ordered series of numbers and be designated as y ₁, y ₂..., y _t, wherein

x _i＝{x _i(1),x _i(2),…,x _i(n)}i＝1,2,…,m

y _j＝{y _j(1),y _j(2),…,y _j(n)}j＝1,2,…,t

Wherein x _i(k) and y _ik () represents x respectively _iand y _ia kth index.

Correlation coefficient is designated as:

${\mathrm{\&xi;}}_{ij}\left(k\right)=\frac{{\mathrm{\&Delta;}}_{min}+{\mathrm{\&rho;}}_{max}}{{\mathrm{\&Delta;}}_{ij}\left(k\right)+{\mathrm{\&rho;\&Delta;}}_{max}}$

Wherein: ${\mathrm{\&Delta;}}_{\min }=\underset{j}{\min }\underset{k}{\min }|x_i\left(k\right)-y_j\left(k\right)|,{\mathrm{\&Delta;}}_{\max }=\underset{j}{\max }\underset{k}{\max }|x_i\left(k\right)-y_j\left(k\right)|,$

Δ _ij(k)=| x _i(k)-y _j(k) |, get ρ=0.5.

Grey Euclid relation grade is:

$R_{ij}=1-\frac{1}{\sqrt{n}}{\&lsqb;\underset{k=1}{\overset{n}{\&Sigma;}}{\left({\mathrm{\&xi;}}_{ij}\right(k)-1)}^2\&rsqb;}^{1/2}$

The present invention considers the problem of weight, therefore slightly does to be out of shape to it:

$R_{ij}=1-{\&lsqb;\underset{k=1}{\overset{n}{\&Sigma;}}{\left({\mathrm{\&xi;}}_{ij}\right(k)-1)}^2{\mathrm{\&omega;}}_j\left(k\right)\&rsqb;}^{1/2}$

If the correlation coefficient ξ of a kth index _ij(k) and weighted association degree r _ijdifference be:

ε _ij(k)＝ξ _ij(k)-r _ij

Then

ξ _ij(k)＝ε _ij(k)+r _ij

Wherein $r_{ij}=\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&xi;}}_{ij}\left(k\right){\mathrm{\&omega;}}_j\left(k\right),$ Because $\underset{k=1}{\overset{n}{\&Sigma;}}\mathrm{\&omega;}\left(k\right)=1,$ So $\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}\mathrm{\&omega;}\left(k\right)=0.$

Above formula is substituted into obtain grey Euclid weighted association degree:

$R_{ij}=1-{\&lsqb;{(r_{ij}-1)}^2+\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}^2\left(k\right)\mathrm{\&omega;}\left(k\right)\&rsqb;}^{1/2}$

Described transformer fault type analysis, to the grey Euclid weighted association degree descending sort calculated, the maximum fault type of the degree of association is the fault type of follow-up transformer, and the sorting representationb follow-up transformer of the degree of association belongs to the arranging situation of various fault type possibility size.

Beneficial effect: a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis provided by the invention, compared to classic method, effectively prevent in traditional entropy assessment when entropy level off to 1 time, the problem of entropy power unreasonable distribution, and use grey Euclid weighted association degree to take into account the significance level of each index in index set, overcome the deficiency of index equal rights, make full use of the advantage of grey correlation analysis process small sample, poor information.

Accompanying drawing explanation

Fig. 1 is based on the Diagnosis Method of Transformer Faults process flow diagram of entropy assessment and grey correlation analysis;

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

A kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis of the present invention, as shown in Figure 1, comprises the steps:

Step one: set up transformer fault state;

Step 2: sample data standardization;

Step 3: entropy assessment agriculture products weight;

Step 4: calculate grey Euclid weighted association degree;

Step 5: transformer fault type analysis.

Specific implementation process is as follows:

In traditional grey correlation analysis, the calculating of the degree of association adopts equal rights or expert to compose power, its weight is made to there is certain subjectivity random, entropy assessment is a kind of method of the information determination weight according to achievement data, it calculates its entropy according to the difference of each index, then calculates the weight obtaining each index according to entropy.But in traditional entropy assessment when entropy level off to 1 time, even if there is very subtle difference in entropy, its entropy power all will have greatly changed, cause its weight not conform to the importance of index, and then cause the erroneous judgement of transformer fault.The present invention proposes the Diagnosis Method of Transformer Faults that a kind of entropy assessment and grey correlation analysis combine.Hereafter with 15 groups of transformer DGA data instances through pendant-core examination its fault type known, provide its method for diagnosing faults.

1) transformer fault state is set up

Five kinds of crucial hydrocarbon gas H in selecting oil chromatography to test herein ₂, CH ₄, C ₂h ₂, C ₂h ₄, C ₂h ₆as characteristic gas, transformer fault is divided into 8 kinds of malfunctions such as non-fault, shelf depreciation, low-yield electric discharge, specifically classifies in table 1.

Table 1 transformer fault is classified

2) sample data standardization

Nondimensionalization process is carried out to sample data, the mode of process be in each fault sample every gas content all divided by the maximal value of gas content in this fault sample.That is:

$x_i^{\&prime;}=\frac{x_i}{x_{\max }}$

In formula: x ' _ifor the value after fault sample standardization, x _ifor the content of gas in fault sample, x _maxfor fault

The maximal value of gas content in sample.

The present invention chooses 8 groups of transformer DGA data through pendant-core examination its fault type known as feature samples, and as shown in table 3, wherein sample 1 is unfaulty conditions, gas content is 20 DEG C, 101.3kPa time, the microlitre number dissolved in every 1L transformer oil.

Table 2 transformer fault feature samples

Data in his-and-hers watches 2 carry out standardization, and the feature samples after process is in table 3.

Transformer fault feature samples after table 3 standardization

3) entropy assessment agriculture products weight

If transformer has m kind fault mode, often kind of fault mode has n index, formed m × n exponent number according to matrix A '=[a ' _ij] _{m × n}, wherein a ' _ijrepresent the numerical value of a jth index in i-th kind of fault, carry out standardization to fault data, the mode of process is that under each malfunction, every gas content, all divided by the maximal value of gas content under this malfunction, obtains A=[a after standardization _ij] _{m × n}.The entropy of a jth index is:

$\begin{array}{cc}{E}_j=\mathrm{\&alpha;}\&CenterDot;\underset{i=1}{\overset{m}{\&Sigma;}}{b}_{ij}{\mathrm{lnb}}_{ij}& j=1,2,\mathrm{...}n\end{array}$

Wherein, work as b _ijwhen=0, make b _ijlnb _ij=0, the entropy power of a jth index is:

$\begin{array}{cc}{\mathrm{\&omega;}}_j=(1-{\stackrel{\&OverBar;}{E}}^{35}){\mathrm{\&omega;}}_j^{\&prime;}+{\stackrel{\&OverBar;}{E}}^{35}{\mathrm{\&omega;}}_j^{\&prime;\&prime;}& E_j<1\end{array}$

ω _j＝0E _j＝1

Wherein ${\mathrm{\&omega;}}_j^{\&prime;}=\frac{1-E_j}{\underset{j=1}{\overset{n}{\&Sigma;}}(1-E_j)},{\mathrm{\&omega;}}_j^{\&prime;\&prime;}=\frac{1+\stackrel{\&OverBar;}{E}-E_j}{\underset{j=1,E_j\&NotEqual;1}{\overset{n}{\&Sigma;}}(1+\stackrel{\&OverBar;}{E}-E_j)},$ for all be not the mean value of the entropy of 1.

The weights omega obtaining each index is:

ω＝[ω ₁,ω ₂,ω ₃,ω ₄,ω ₅]＝[0.0253,0.0408,0.5593,0.1651,0.2094]

4) grey Euclid weighted association degree is calculated

If transformer has m kind fault mode, often kind of fault mode has n index, then its reference sequence is designated as x ₁, x ₂..., x _m, compare ordered series of numbers and be designated as y ₁, y ₂..., y _t, wherein

x _i＝{x _i(1),x _i(2),…,x _i(n)}i＝1,2,…,m

y _j＝{y _j(1),y _j(2),…,y _j(n)}j＝1,2,…,t

Wherein x _i(k) and y _ik () represents x respectively _iand y _ia kth index;

Correlation coefficient is designated as:

${\mathrm{\&xi;}}_{ij}\left(k\right)=\frac{{\mathrm{\&Delta;}}_{\min }+{\mathrm{\&rho;}}_{max}}{{\mathrm{\&Delta;}}_{ij}\left(k\right)+{\mathrm{\&rho;\&Delta;}}_{\max }}$

Wherein: ${\mathrm{\&Delta;}}_{\min }=\underset{j}{\min }\underset{k}{\min }|x_i\left(k\right)-y_j\left(k\right)|,{\mathrm{\&Delta;}}_{\max }=\underset{j}{\max }\underset{k}{\max }|x_i\left(k\right)-y_j\left(k\right)|,$

Δ _ij(k)=| x _i(k)-y _j(k) |, get ρ=0.5;

Grey Euclid relation grade is:

$R_{ij}=1-\frac{1}{\sqrt{n}}{\&lsqb;\underset{k=1}{\overset{n}{\&Sigma;}}{\left({\mathrm{\&xi;}}_{ij}\right(k)-1)}^2\&rsqb;}^{1/2}$

The present invention considers the problem of weight, therefore slightly does to be out of shape to it:

$R_{ij}=1-{\&lsqb;\underset{k=1}{\overset{n}{\&Sigma;}}{\left({\mathrm{\&xi;}}_{ij}\right(k)-1)}^2{\mathrm{\&omega;}}_j\left(k\right)\&rsqb;}^{1/2}$

If the correlation coefficient ξ of a kth index _ij(k) and weighted association degree r _ijdifference be:

ε _ij(k)＝ξ _ij(k)-r _ij

Then

ξ _ij(k)＝ε _ij(k)+r _ij

Wherein $r_{ij}=\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&xi;}}_{ij}\left(k\right){\mathrm{\&omega;}}_j\left(k\right),$ Because $\underset{k=1}{\overset{n}{\&Sigma;}}\mathrm{\&omega;}\left(k\right)=1,$ So $\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}\mathrm{\&omega;}\left(k\right)=0.$

Sample to be tested data are: H ₂content is 1565.0 μ L/L, CH ₄content is 93.0 μ L/L, C ₂h ₂content is 0 μ L/L, C ₂h ₄content is 34.0 μ L/L, C ₂h ₆content is 47.0 μ L/L.Carrying out the data after standardization to it is 1,0.0594,0,0.0217,0.0300.

Above formula is substituted into obtain grey Euclid weighted association degree:

$R_{ij}=1-{\&lsqb;{(r_{ij}-1)}^2+\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}^2\left(k\right)\mathrm{\&omega;}\left(k\right)\&rsqb;}^{1/2}$

Obtaining grey Euclid weighted association degree R is:

R＝[R ₁,R ₂,R ₃,R ₄,R ₅,R ₆,R ₇,R ₈,R ₉]＝[0.7337,0.9765,0.6627,0.4325,0.8932,0.8502,0.8007,0.8629]

5) transformer fault type analysis

To the grey Euclid weighted association degree descending sort calculated, the maximum fault type of the degree of association is the fault type of follow-up transformer, and the sorting representationb follow-up transformer of the degree of association belongs to the arranging situation of various fault type possibility size.

The degree of association calculated from above formula, the degree of association of sample to be tested and sample 2 is maximum, therefore transformer there occurs partial discharges fault, and diagnostic result is true to life.

In order to further illustrate accuracy and the reliability of the inventive method, the DGA data choosing 15 groups of its malfunctions known, as sample to be tested, carry out Analysis on Fault Diagnosis, in table 4 to it, fault type sequence number in table is consistent with the sequence number in table 1, represents the fault type of each sample.

Table 415 group transformer fault sample

The present invention adopts IEC three-ratio method respectively, general grey correlation analysis carries out fault diagnosis to above-mentioned 15 groups of samples to be tested, its analysis result is in table 5, sample sequence number in table, fault sequence number are consistent with table 4, the result sequence number diagnosed out when often kind of method is consistent with the fault sequence number that the 2nd arranges, then illustrate that its fault diagnosis result is accurately, on the contrary then inaccurate.

The diagnostic result of table 53 kind of method

As can be seen from above-mentioned diagnostic result, IEC three-ratio method, general grey correlation analysis, its Accurate Diagnosis sample number of the inventive method are respectively 9,12,14, and rate of correct diagnosis is respectively 60%, 80%, 93.33%.The accuracy rate of computing method of the present invention is the highest as can be seen here, and the method can improve the accuracy rate of transformer fault diagnosis.

The above is only the preferred embodiment of the present invention; be noted that for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (6)

1., based on a Diagnosis Method of Transformer Faults for entropy assessment and grey correlation analysis, it is characterized in that, comprise the steps:

Step one: set up transformer fault state;

Step 2: sample data standardization;

Step 3: entropy assessment agriculture products weight;

Step 4: calculate grey Euclid weighted association degree;

Step 5: transformer fault type analysis.

2. a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis according to claim 1, is characterized in that: describedly set up transformer fault state, five kinds of crucial hydrocarbon gas H in being tested by oil chromatography ₂, CH ₄, C ₂h ₂, C ₂h ₄, C ₂h ₆as characteristic gas, transformer fault is divided into the double overheated eight kinds of malfunctions of non-fault, shelf depreciation, low-yield electric discharge, high-energy discharge, cryogenic overheating t ﹤ 300 DEG C, middle temperature overheated t ﹤ 700 DEG C, hyperthermia and superheating t ﹥ 700 DEG C and electric discharge.

3. a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis according to claim 1, it is characterized in that: described sample data standardization, nondimensionalization process is carried out to sample data, the mode of process be in each fault sample every gas content all divided by the maximal value of gas content in this fault sample.

4. a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis according to claim 1, it is characterized in that: described entropy assessment agriculture products weight, if transformer has m kind fault mode, often kind of fault mode has n index, formed m × n exponent number according to matrix A '=[a ' _ij] _{m × n}, wherein a ' _ijrepresent the numerical value of a jth index in i-th kind of fault, carry out standardization to fault data, the mode of process is that under each malfunction, every gas content, all divided by the maximal value of gas content under this malfunction, obtains A=[a after standardization _ij] _{m × n}; The entropy of a jth index is:

$E_j=\mathrm{\&alpha;}\&CenterDot;\underset{i=1}{\overset{m}{\&Sigma;}}{b}_{ij}{\mathrm{lnb}}_{ij},j=1,2,\mathrm{...}n$

Wherein, work as b _ijwhen=0, make b _ijlnb _ij=0, the entropy power of a jth index is:

${\mathrm{\&omega;}}_j=(1-{\stackrel{\&OverBar;}{E}}^{35}){\mathrm{\&omega;}}_j^{\&prime;}+{\stackrel{\&OverBar;}{E}}^{35}{\mathrm{\&omega;}}_j^{\&prime;\&prime;},E_j<1$

ω _j＝0E _j＝1

Wherein ${\mathrm{\&omega;}}_j^{\&prime;}=\frac{1-E_j}{\underset{j=1}{\overset{n}{\&Sigma;}}(1-E_j)},{\mathrm{\&omega;}}_j^{\&prime;\&prime;}=\frac{1+\stackrel{\&OverBar;}{E}-E_j}{\underset{j=1,E_j\&NotEqual;1}{\overset{n}{\&Sigma;}}(1+\stackrel{\&OverBar;}{E}-E_j)},$ for all be not the mean value of the entropy of 1.

5. a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis according to claim 1, it is characterized in that: described calculating grey Euclid weighted association degree, if transformer has m kind fault mode, often kind of fault mode has n index, then its reference sequence is designated as x ₁, x ₂..., x _m, compare ordered series of numbers and be designated as y ₁, y ₂..., y _t, wherein

x _i＝{x _i(1),x _i(2),…,x _i(n)}i＝1,2,…,m

y _j＝{y _j(1),y _j(2),…,y _j(n)}j＝1,2,…,t

Wherein x _i(k) and y _ik () represents x respectively _iand y _ia kth index;

Correlation coefficient is designated as:

${\mathrm{\&xi;}}_{ij}\left(k\right)=\frac{{\mathrm{\&Delta;}}_{\min }+{\mathrm{\&rho;\&Delta;}}_{max}}{{\mathrm{\&Delta;}}_{ij}\left(k\right)+{\mathrm{\&rho;\&Delta;}}_{max}}$

Wherein: ${\mathrm{\&Delta;}}_{\min }=\underset{j}{\min }\underset{k}{\min }|x_i\left(k\right)-y_j\left(k\right)|,{\mathrm{\&Delta;}}_{max}=\underset{j}{\max }\underset{k}{\max }|x_i\left(k\right)-y_j\left(k\right)|,$

Δ _ij(k)=| x _i(k)-y _j(k) |, get ρ=0.5;

If the correlation coefficient ξ of a kth index _ij(k) and weighted association degree r _ijdifference be:

ε _ij(k)＝ξ _ij(k)-r _ij

Then

ξ _ij(k)＝ε _ij(k)+r _ij

Wherein $r_{ij}=\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&xi;}}_{ij}\left(k\right){\mathrm{\&omega;}}_j\left(k\right),$ Because $\underset{k=1}{\overset{n}{\&Sigma;}}\mathrm{\&omega;}\left(k\right)=1,$ So $\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}\mathrm{\&omega;}\left(k\right)=0;$

Grey Euclid weighted association degree: $R_{ij}=1-{\&lsqb;{(r_{ij}-1)}^2+\underset{k=1}{\overset{n}{\&Sigma;}}{\mathrm{\&epsiv;}}_{ij}^2\left(k\right)\mathrm{\&omega;}\left(k\right)\&rsqb;}^{1/2}.$

6. a kind of Diagnosis Method of Transformer Faults based on entropy assessment and grey correlation analysis according to claim 1, it is characterized in that: described transformer fault type analysis, to the grey Euclid weighted association degree descending sort calculated, the maximum fault type of the degree of association is the fault type of follow-up transformer, and the sorting representationb follow-up transformer of the degree of association belongs to the arranging situation of various fault type possibility size.