Transformer Condition Assessment Method Based on Spherical Fuzzy Sets and Improved MEREC Combined Weighting

Abstract

The present invention discloses a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, which is applied to the technical field of transformer condition assessment. The method includes the following steps: establishing a transformer health condition assessment index system and defining a health condition interval based on a health index function; determining the subjective weight of evaluation indexes through spherical fuzzy sets; determining the objective weight of evaluation indexes based on the MEREC method of correlation coefficients; endowing the evaluation indexes with a comprehensive weight based on variable weight theory; and calculating the correlation degree of each evaluation index relative to the health level based on the optimized cloud model. The present invention can accurately assess the health condition of a transformer, providing a theoretical basis for the safe operation of a power system and the protection of power equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer condition assessment, and more particularly to a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting. Background Art

[0002] The health status of a transformer is the basis for ensuring safety. Therefore, the accurate assessment of the health status of a transformer is very important for maintaining the safety of the power system and extending the service life of the transformer. However, in the current transformer health status assessment methods, the weight allocation cannot accurately describe the importance of each status index, and the dynamic change of the index weight is ignored in the actual assessment process of the transformer status, which cannot timely reflect the impact on the transformer health status and is difficult to provide an accurate and reasonable basis and decision for the real-time risk assessment of the transformer. Therefore, how to provide a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0003] In view of this, the present invention provides a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, which solves the problems of unreasonable allocation of subjective and objective weights of evaluation indexes in the prior art and ignores the dynamic change of index weights.

[0004] In order to achieve the above object, the present invention provides the following technical solutions:

[0005] A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, comprising the following steps:

[0006] S1. Establish a transformer health status assessment index system, and define the health status interval of each health evaluation index based on the health index function;

[0007] S2. Determine the subjective weight of the health evaluation index through spherical fuzzy sets;

[0008] S3. Determine the objective weight of the health evaluation index based on the MEREC method of correlation coefficients;

[0009] S4. Assign a comprehensive weight to the health evaluation index based on the variable weight theory to obtain a comprehensive weight matrix;

[0010] S5. Optimize the cloud model, calculate the correlation matrix of each health evaluation index relative to the health level; obtain the health level of the transformer by calculating the product of the correlation matrix and the comprehensive weight matrix.

[0011] Optionally, the health evaluation indicators in the transformer health status evaluation index system include: polarization index A1, winding dielectric loss A2, winding leakage current A3, and insulation resistance absorption ratio A4; micro water in oil A5, oil dielectric loss A6, oil breakdown voltage A7, and furfural in oil A8; total hydrocarbon relative gas production rate A9, hydrogen content A10, total hydrocarbon content A11, and acetylene content A12; the health grades in the transformer health status evaluation index system are divided into normal state H1, degradation state H2, attention state H3, abnormal state H4, and severe state H5.

[0012] Optionally, the health index function is specifically:

[0013] ;

[0014] In the formula, f ( x ) ∈ [0, 10], a and b respectively represent the lower and upper limits of the test values of the health evaluation indicators. The health index interval is divided at intervals of 2. The test value is obtained by inverse calculation through the health index function, and then the health status interval is defined.

[0015] Optionally, S2 is specifically:

[0016] S21. Let U be a non-empty universe of discourse, and define a spherical fuzzy set P on U:

[0017] ;

[0018] ;

[0019] ;

[0020] In the formula, u p : U ∈ [0, 1], v p : U ∈ [0, 1], τ p : U ∈ [0, 1], u p is the membership degree, v p is the non-membership degree, τ p is the hesitation degree, and , λ p is the rejection degree, SC ( A ) is the score function;

[0021] S22. Define spherical fuzzy numbers A i =( u Ai , v Ai , τ Ai )( i = 1, 2, ..., n ), and the corresponding expert weight vector W m =( w m1 , w m2 , ..., w mc ), where the number of experts is c ; The spherical weighted arithmetic mean SWAM, spherical weighted geometric mean SWGM, spherical fuzzy weighted arithmetic mean SFNWAA, and spherical fuzzy weighted geometric mean SFNWGA are defined as follows:

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] S23. Combine the spherical fuzzy numbers of each evaluation index given by the experts and the expert weight vector, calculate the spherical weighted arithmetic mean SWAM of each evaluation index, calculate the SWAM value of each health evaluation index and SC ( A i ):

[0027] ;

[0028] ;

[0029] Subjective weight W s =( w s1 , w s2 , ..., w sn ).

[0030] Optionally, S3 is specifically as follows:

[0031] S31. Define m health evaluation indicators, and each health evaluation indicator includes nEvaluation matrix of data:

[0032] ;

[0033] S32. Calculate the health index of each group of data H =( h 1, h 2,..., h n ), and use it as a reference sequence reflecting the system performance. The new evaluation matrix is X n×(m+1) ;

[0034] S33. Normalize the new evaluation matrix:

[0035] ;

[0036] In the formula, is a cost-type index, is a benefit-type index;

[0037] S34. Calculate the correlation coefficient between the data and the reference sequence:

[0038] ;

[0039] S35. The overall performance of each group of data in the correlation coefficient matrix is:

[0040] ;

[0041] S36. Calculate the overall performance by deleting each health evaluation index from the data:

[0042] ;

[0043] S37. Calculate the sum of absolute deviations of each health evaluation index:

[0044] ;

[0045] S38. Obtain the objective weights of each health evaluation index:

[0046] .

[0047] Optionally, S4 is specifically:

[0048] S4 is specifically:

[0049] S41. Calculate the constant weight comprehensive weight W c :

[0050] W c=δ·W s +λW o

[0051] W o =( w o1 , w o2 ,..., w on );

[0052] In the formula, W s is the subjective weight vector, W o is the objective weight vector, δ and λ are the subjective and objective weight coefficients respectively;

[0053] S42. Define the index status vector X =( x 1,..., x n )), the variable weight w j =( w 1,..., w n ), and the penalty term n with dimension S ( X ):

[0054] ;

[0055] ;

[0056] In the formula: j = 1, 2,..., n ; α ≥ 0; 0 < β ≤ 1, β is the negative factor of the penalty term S ( X ). If the index factor x ij ≤ β value, it means that the estimated value of this index is in the marginal state, and its weight value is increased by the variable weight method, α represents the penalty factor of the penalty term S ( X ), that is, when α value increases, the degree of penalty also increases;

[0057] S43. Calculate the variable weight comprehensive weight matrix WV ( X )=( W V 1( X ),..., W V n ( X )):

[0058] ;

[0059] Where, W c · S ( X )=( w c 1 S 1( X ),..., w c n S n ( X )).

[0060] Optionally, the cloud model optimization in S5 is as follows:

[0061] Determine the digital characteristics of the cloud model:

[0062] ;

[0063] ;

[0064] H e =k ;

[0065] Where, X max Indicates that each evaluation indicator is at the upper limit of the five health levels. X min Indicates that each evaluation index is at the lower limit of the five health levels, ( E x ，E n , H e ) are the three digital features of the cloud model, E x For expectations, E n is entropy, H e is super entropy;

[0066] Optimizing cloud model expectations E x :

[0067] ;

[0068] wherein, and represent the expected value and entropy of the p th level, λ is a constant determined by the above formula λ The range is a < λ < b , take , the expected value of the optimized cloud model , in order to be applicable to practical engineering applications, the expected values of health level 1 and health level 5 are set to 0 and m respectively, where m is the maximum value of the index value, and at the same time, its entropy value is updated to , where is the length of the level interval;

[0069] Taking E n as the expectation, H e 2 generate normal random numbers with as the standard deviation, E x as the expectation generate normal random numbers with x as the standard deviation, generate cloud droplets ([[]] x,μ ):

[0070] ;

[0071] Repeat the above steps until the requirement for the number of cloud droplets is met.

[0072] Optionally, S5 is specifically: input the operation status data of each health evaluation index of the transformer into the cloud model, obtain the correlation degree between each health evaluation index and the health level, and calculate the correlation matrix of the health evaluation index D , obtain the correlation matrix of the health level A = D · W , where W is the weight matrix, and finally obtain the health level of the transformer.

[0073] As can be seen from the above technical solutions, compared with the prior art, the present invention provides a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, which has the following beneficial effects: The present invention reasonably distributes the subjective and objective weights of evaluation indicators through spherical fuzzy sets and the improved MEREC method, providing an accurate theoretical basis for the transformer health condition assessment process; The present invention endows the dynamic change of the index weight, timely reflects the influence of other factors on the transformer health condition, and thus provides an accurate and reasonable basis and decision for the real-time risk assessment of the transformer; The present invention optimizes the expected value of the cloud model, accurately describes the correlation degree between the index value and each level, and fits the actual engineering application scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0075] Figure 1 It is a flowchart of the transformer condition assessment method of the present invention;

[0076] Figure 2 It is a health level diagram before cloud model optimization in the embodiment of the present invention;

[0077] Figure 3 It is a health level diagram after cloud model optimization in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0078] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0079] The embodiment of the present invention discloses a transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, as Figure 1 shown, including the following steps:

[0080] S1. Establish a transformer health condition assessment index system, and define the health condition interval of each health evaluation index based on the health index function;

[0081] S2. Determine the subjective weight of the health evaluation index through spherical fuzzy sets;

[0082] S3. Determine the objective weights of the health evaluation indicators using the MEREC method based on the correlation coefficient;

[0083] S4. Assign comprehensive weights to the health evaluation indicators based on the variable weight theory to obtain the comprehensive weight matrix;

[0084] S5. Optimize the cloud model, calculate the correlation matrix of each health evaluation indicator relative to the health level; obtain the health level of the transformer by calculating the product of the correlation matrix and the comprehensive weight matrix.

[0085] Furthermore, the health evaluation indicators in the transformer health status evaluation index system include: polarization index A1, winding dielectric loss A2, winding leakage current A3, and insulation resistance absorption ratio A4; micro water in oil A5, oil dielectric loss A6, oil breakdown voltage A7, and furfural in oil A8; total hydrocarbon relative gas generation rate A9, hydrogen content A10, total hydrocarbon content A11, and acetylene content A12; the health levels in the transformer health status evaluation index system are divided into normal state H1, degradation state H2, attention state H3, abnormal state H4, and severe state H5.

[0086] In the embodiment of the present invention, the polarization index A1, winding dielectric loss A2, winding leakage current A3, and insulation resistance absorption ratio A3 are obtained through electrical tests; the micro water in oil A5, oil dielectric loss A6, oil breakdown voltage A7, and furfural in oil A8 are obtained through oil chemical tests; the total hydrocarbon relative gas generation rate A9, hydrogen content A10, total hydrocarbon content A11, and acetylene content A12 are obtained through dissolved gas analysis in oil. The health levels are divided into normal state H1, degradation state H2, attention state H3, abnormal state H4, and severe state H5.

[0087] Furthermore, the health index function is specifically:

[0088] ;

[0089] In the formula, f ( x ) ∈ [0, 10], a and b respectively represent the lower and upper limits of the test values of the health evaluation indicators. Divide the health index into health index intervals at intervals of 2, inversely calculate the test values through the health index function, and then define the health status intervals.

[0090] In the embodiment of the present invention, the health index is divided into health index intervals at intervals of 2, the test values are inversely calculated through the health index function, and then the health status intervals are defined, as shown in Table 1:

[0091] Table 1

[0092]

[0093] Furthermore, S2 is specifically as follows:

[0094] S21. Let U be a non-empty universe of discourse, and define a spherical fuzzy set P on U:

[0095] ;

[0096] ;

[0097] ;

[0098] In the formula, u p : U ∈ [0, 1], v p : U ∈ [0, 1], τ p : U ∈ [0, 1], u p is the membership degree, v p is the non-membership degree, τ p is the hesitation degree, and , λ p is the rejection degree, SC ( A ) is the score function;

[0099] S22. Define the spherical fuzzy number A i =( u Ai , v Ai , τ Ai )( i = 1, 2,..., n ), and the corresponding expert weight vector W m =( w m1 , w m2 ,..., w mc ), where the number of experts is c ; The spherical weighted arithmetic mean SWAM, spherical weighted geometric mean SWGM, spherical fuzzy weighted arithmetic mean SFNWAA, and spherical fuzzy weighted geometric mean SFNWGA are defined as follows:

[0100] ;

[0101] ;

[0102] ;

[0103] ;

[0104] In the embodiments of the present invention, the spherical fuzzy numbers corresponding to different language terms are shown in Table 2:

[0105] Table 2

[0106]

[0107] S23. Combine the spherical fuzzy numbers of each evaluation index given by the experts and the expert weight vector, calculate the spherical weighted arithmetic mean SWAM of each evaluation index, and calculate the SWAM values of each health evaluation index and SC ( A i ):

[0108] ;

[0109] ;

[0110] Subjective weight W s =( w s1 , w s2 ,..., w sn ).

[0111] Furthermore, S3 is specifically as follows:

[0112] S31. Define m health evaluation indexes, and each health evaluation index includes an evaluation matrix of n data:

[0113] ;

[0114] S32. Calculate the health index of each group of data based on the health index function H =( h 1, h 2,..., h n ), and use it as the reference sequence reflecting the system performance. The new evaluation matrix is X n×(m+1) ;

[0115] S33. Normalize the new evaluation matrix:

[0116] ;

[0117] In the formula, is a cost-type index, is a benefit-type index;

[0118] S34. Calculate the correlation coefficient between the calculation data and the reference sequence:

[0119] ;

[0120] S35. The overall performance of each group of data in the correlation coefficient matrix is:

[0121] ;

[0122] S36. Obtain the overall performance by deleting the calculation data of each health evaluation index:

[0123] ;

[0124] S37. Calculate the sum of absolute deviations of each health evaluation index:

[0125] ;

[0126] S38. Obtain the objective weights of each health evaluation index:

[0127] .

[0128] Furthermore, S4 is specifically as follows:

[0129] S4 is specifically as follows:

[0130] S41. Calculate the constant-weight comprehensive weight W c :

[0131] W c =δ·W s +λW o

[0132] W o =( w o1 , w o2 ,..., w on );

[0133] In the formula, W s is the subjective weight vector, W o is the objective weight vector, δ andλ are the subjective and objective weight coefficients respectively;

[0134] S42. Define the index status vector X =( x 1,..., x n ), the variable weight w j =( w 1,..., w n ), and the dimension is n of the penalty term S ( X ):

[0135] ;

[0136] ;

[0137] In the formula: j = 1, 2,..., n ; α ≥ 0; 0 < β ≤ 1, β is the negative factor of the penalty term S ( X ). If the index factor x ij ≤ β value, it means that the estimated value of this index is in the marginal state, and its weight value is increased by the variable weight method, α represents the penalty factor of the penalty term S ( X ), that is, when α value increases, the degree of penalty also increases;

[0138] S43. Calculate the variable weight comprehensive weight matrix W V ( X ) = ( W V 1( X ),..., W V n ( X )):

[0139] ;

[0140] In the formula, W c · S ( X ) = ( w c 1 S 1(X ),... w c n S n ( X )).

[0141] Furthermore, the cloud model optimization in S5 is specifically as follows:

[0142] Determine the digital characteristics of the cloud model:

[0143] ;

[0144] ;

[0145] H e =k ;

[0146] Wherein, X max represents the upper limit of the interval where each evaluation index is in five health levels, X min represents the lower limit of the interval where each evaluation index is in five health levels, ([[]] E x ，E n , H e ) are the three digital characteristics of the cloud model, E x is the expectation, E n is the entropy, H e is the hyperentropy; in the embodiments of the present invention, the value of K is determined according to the actual situation.

[0147] Optimize the cloud model expectation value E x :

[0148] ;

[0149] Wherein, and represent the expectation value and entropy of the p th level, λ is a constant, determined by the above formula λ The range is a < λ < b , take , the optimized cloud model expectation value , for practical engineering applications, the expected values of health level 1 and health level 5 are set to 0 and m respectively, where m is the maximum value of the index value. At the same time, its entropy value is updated to , where is the length of the grade interval;

[0150] Taking E n as the expectation, H e 2 generate normal random numbers with , taking E x as the expectation generate normal random numbers with x , generate cloud droplets ( x,μ ):

[0151] ;

[0152] Repeat the above steps until the requirement of the number of cloud droplets is met. The cloud model grades before and after optimization are as shown in Figure 2 and Figure 3 respectively.

[0153] Furthermore, S5 is specifically: input the operation status data of each health evaluation index of the transformer into the cloud model, obtain the correlation degree between each health evaluation index and the health level, and calculate the correlation matrix of health evaluation indexes D , obtain the correlation matrix of health levels A = D · W , where W is the weight matrix, and finally obtain the health level of the transformer.

[0154] The embodiments in this specification are described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting, characterized in that It includes the following steps: S1. Establish a transformer health status evaluation index system, and define the health status interval of each health evaluation index based on the health index function; S2. Determine the subjective weight of the health evaluation index through spherical fuzzy sets; S3. Determine the objective weight of the health evaluation index based on the MEREC method of correlation coefficient; S4. Assign a comprehensive weight to the health evaluation index based on the variable weight theory to obtain a comprehensive weight matrix; S5. Optimize the cloud model, calculate the correlation matrix of each health evaluation index relative to the health level; obtain the health level of the transformer by calculating the product of the correlation matrix and the comprehensive weight matrix; S3 specifically is: S31. Define m health evaluation indexes, and each health evaluation index contains an evaluation matrix of n data: S32. Calculate the health index H=(h1, h2,..., h n ) of each group of data based on the health index function, and use it as the reference sequence reflecting the system performance. The new evaluation matrix is X n×(m+1) ; S33. Normalize the new evaluation matrix: In the formula, ψ is a cost-type index, and Ω is a benefit-type index; S34. Calculate the correlation coefficient between the data and the reference sequence: The overall performance of each group of data in the correlation coefficient matrix is: S36. Obtain the overall performance by deleting the data calculated for each health evaluation index: S37. Calculate the sum of the absolute deviations of each health evaluation index: S38. Obtain the objective weight of each health evaluation index:

2. A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting according to claim 1, characterized in that The health evaluation indexes in the transformer health status evaluation index system include: polarization index A1, winding dielectric loss A2, winding leakage current A3, and insulation resistance absorption ratio A4; micro water in oil A5, oil dielectric loss A6, oil breakdown voltage A7, and furfural in oil A8; total hydrocarbon relative gas production rate A9, hydrogen content A10, total hydrocarbon content A11, and acetylene content A12; the health levels in the transformer health status evaluation index system are divided into normal state H1, degradation state H2, attention state H3, abnormal state H4, and severe state H5.

3. According to a transformer state evaluation method based on combined weighting of spherical fuzzy sets and improved MEREC described in claim 1, the health index function specifically is: In the formula, f(x) ∈ [0, 10], a and b respectively represent the lower limit and the upper limit of the test value of the health evaluation index. Divide the health index into health index intervals at intervals of 2. Inversely calculate the test value through the health index function, and then define the health status interval.

4. According to a transformer state evaluation method based on combined weighting of spherical fuzzy sets and improved MEREC described in claim 1, S2 specifically is: S21. Let U be a non-empty universe of discourse, and define a spherical fuzzy set P on U: P = {x, u p (x), v p (x), τ p (x) | x ∈ U}; where \(u\) p : \(U\in[0,1]\), \(v\) p : \(U\in[0,1]\), \(\tau\) p : \(U\in[0,1]\), \(u\) p is the membership degree, \(v\) p is the non - membership degree, \(\tau\) p is the hesitation degree, and \(\lambda\) p is the rejection degree, \(SC(A)\) is the scoring function; S22. Define the spherical fuzzy number A i =(u Ai , v Ai , τ Ai ), i = 1, 2,..., n, and the corresponding expert weight vector W m =(w m1 , w m2 ,..., w mc ), where the number of experts is c; the spherical weighted arithmetic mean SWAM, the spherical weighted geometric mean SWGM, the spherical fuzzy weighted arithmetic mean SFNWAA, and the spherical fuzzy weighted geometric mean SFNWGA are defined as follows: S23. Combine the spherical fuzzy numbers of each evaluation index given by experts and the expert weight vector to calculate the spherical weighted arithmetic mean (SWAM) of each evaluation index, and calculate the SWAM value and SC(A i ): A i，SWAM =(u SWAM , v SWAM , τ SWAM ) Subjective weight \(W\) s =(w s1 , w s2 ,..., w sn ).

5. A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting according to claim 1, characterized in that S4 specifically is: S41. Calculate the constant weight comprehensive weight W c : W c = δ·W s + λW o W o = (w o1 , w o2 ,..., w on ); where, W s is the subjective weight vector, W o is the objective weight vector, and δ and λ are the subjective and objective weight coefficients respectively; S42. Define the index status vector X = (x1,..., x n ), the variable weight w j = (w1,..., w n ), and the penalty term S(X) with dimension n: X → S(X) = (S1(X),..., S n (X)); where: j = 1, 2,..., n; α ≥ 0; 0 < β ≤ 1, and β is the negative factor of the penalty term S(X). If the value of the index factor x ij ≤ β, it means that the estimated value of this index is in a marginal state, and its weight value is increased by the variable weight method. α represents the penalty factor of the penalty term S(X), that is, when the value of α increases, the degree of penalty also increases; S43. Calculate the variable weight comprehensive weight matrix W V (X) = (W V 1(X),..., W V n (X)): Where, W c ·S(X) = (w c 1S1(X),..., w c n S n (X)).

6. A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting according to claim 1, characterized in that The cloud model optimization in S5 specifically is: Determine the digital characteristics of the cloud model: H e = k; where X max represents the upper limit of the interval in which each evaluation index is in five health levels, and X min represents the lower limit of the interval in which each evaluation index is in five health levels. (E x , E n , H e ) are the three digital characteristics of the cloud model, where E x is the expectation, E n is the entropy, and H e is the hyperentropy; Optimize the expected value E of the cloud model x : In the formula, and represent the expected value and entropy of the p-th level. λ is a constant, and the range of λ is determined by the above formula as a < λ < b. Take The expected value of the optimized cloud model To be applicable to practical engineering applications, the expected values of health level 1 and health level 5 are set to 0 and m respectively, where m is the maximum value of the index value. At the same time, its entropy value is updated to where β is the length of the level interval; With E n as the expectation, H e 2 generates a normal random number E′ with the standard deviation n , and with E x as the expectation generates a normal random number x with the standard deviation, and generates cloud droplets (x, μ): Repeat the above steps until the requirement for the number of cloud droplets is met.

7. A transformer condition assessment method based on spherical fuzzy sets and improved MEREC combined weighting according to claim 1, characterized in that, S5 specifically is: Input the operation state data of each health evaluation index of the transformer into the cloud model, obtain the correlation degree between each health evaluation index and the health level, calculate the health evaluation index correlation matrix D, obtain the health level correlation matrix A = D·W, where W is the weight matrix, and finally obtain the health level of the transformer.

Images