Production green quality comprehensive evaluation method for distribution transformer

Abstract

A production green quality comprehensive evaluation method for a distribution transformer comprises the steps of collecting data, forming a quality detection index matrix, a production energy consumption index matrix and a carbon footprint characteristic index matrix, screening procedures which have great influence on energy consumption by using a K-means clustering algorithm, constructing a production energy consumption index set, updating the production energy consumption index matrix, and evaluating the production green quality of the distribution transformer. And performing normalization processing on various index matrixes to form a normalized index data matrix. On the basis, respectively calculating the weight of each index by adopting a CRITIC method and an entropy weight method to obtain an overall CRITIC weight vector and an overall entropy weight vector; according to a minimum identification information principle, solving an optimal combination weight through a Lagrange multiplier method; and calculating a comprehensive evaluation score and a corresponding grade. The multi-dimensional data fusion mode can reflect the green production quality of the distribution transformer more comprehensively and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological management of electric power materials, and specifically to a comprehensive evaluation method for the production green quality of distribution transformers. Background Art

[0002] Traditional industries still have problems with high energy consumption and high emissions, and the pressure for transformation has not yet been fundamentally alleviated.

[0003] Distribution transformers are a type of voltage conversion equipment widely used in power grids, and their quality level is related to the normal operation of the power grid. In the quality supervision of distribution transformers, there are currently mature supervision standards and procedures at home and abroad, but these quality supervision standards and procedures do not take into account the low-carbon nature of the production process. With the steady progress of the construction of a green modern digital supply chain, the transformation of various businesses has accelerated, and the results of green chain construction have continued to make breakthroughs, but the construction of a material quality index evaluation system from a low-carbon perspective is still unclear, and the existing power grid material quality supervision standards do not take into account the low-carbon nature of material production. The low-carbon index and the existing quality index complement each other, but they may also conflict. There is an urgent need to research and develop a set of quantitative green evaluation index systems for typical power grid materials that take into account both quality goals and low-carbon goals, laying a good foundation for subsequent green quality and efficiency evaluation and material low-carbonization. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned deficiencies in the prior art, thereby providing a comprehensive evaluation method for the production green quality of distribution transformers.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for comprehensively evaluating the production green quality of a distribution transformer, comprising the following steps:

[0006] Step 1: Data Collection

[0007] Obtain the quality inspection index data, production energy consumption index data and carbon footprint characteristic index data of the distribution transformer product to be evaluated, and form the quality inspection index matrix V a , production energy consumption index matrix V b , Carbon footprint characteristic indicator matrix V c ;

[0008] Step 2: Based on the K-means clustering algorithm, the production energy consumption indicators are screened to obtain the processes that have a greater impact on energy consumption, and the production energy consumption indicator set is constructed to obtain the updated production energy consumption indicator matrix V b1 ;

[0009] Step 3: Data integration and normalization

[0010] For the quality inspection index matrix V a , Updated production energy consumption index matrix Vb1 and carbon footprint characteristic indicator matrix V c Perform normalization processing to form the corresponding normalized indicator matrix H a , H b , H c , and establish the normalized overall data matrix X as follows:

[0011] X=[H a ,H b ,H c ];

[0012] Step 4: Calculation of indicator weights based on the CRITIC method

[0013] The normalized index matrix H is calculated using the CRITIC method. a , H b , H c The corresponding indicator CRITIC weight W C1 , W C2 , W C3 , combined into the overall CRITIC weight vector W C ;

[0014] Step 5: Weight calculation based on entropy weight method

[0015] The entropy weight method is used to calculate the normalized index matrix H a , H b , H c The corresponding index entropy weight W Q1 , W Q2 , W Q3 , combined into the overall entropy weight vector W Q ;

[0016] Step 6: Determine the optimal combination weight vector W

[0017] So that the combined weights and the overall CRITIC weight vector W C , the total entropy weight vector W Q With the goal of being close to each other, the objective function is constructed according to the minimum discriminant information principle, and the Lagrange multiplier method is used to solve the optimal combination weight;

[0018] Step 7: Calculate the comprehensive evaluation score and corresponding grade

[0019] The product score S is calculated based on the normalized overall data matrix X and the optimal combination weight vector W of the distribution transformer product to be evaluated, and the percentage score correction is performed to obtain the final score S h and the corresponding evaluation levels.

[0020] The quality inspection index data include the high-voltage and ground insulation resistance, low-voltage and ground insulation resistance, no-load loss, load loss, no-load current, short-circuit impedance, top liquid temperature rise, top oil temperature, low-voltage no-load loss and low-voltage no-load current data of the transformer products to be evaluated; the production energy consumption indicators include oil tank production, clamp production, core production, insulation manufacturing, coil manufacturing, body drying and vacuum oil filling data; the carbon footprint characteristic indicators include the proportion of clean energy, steel transportation distance, copper transportation distance, transformer oil transportation distance, scrap copper price ratio and scrap steel price ratio data.

[0021] Based on the K-means clustering algorithm, the production energy consumption indicators are screened to obtain the processes that have a greater impact on energy consumption, and the production energy consumption indicator set is constructed, including the following steps:

[0022] (1) The standard deviation s of the energy consumption data of the i-th production link is calculated using the following formula i :

[0023]

[0024] Among them, x ij is the energy consumption data corresponding to the jth sample to be evaluated in the i-th production link. There are m samples and 7 production energy consumption indicators; It is the average value of the production energy consumption data of all samples to be evaluated in the i-th production link, which is calculated as follows:

[0025]

[0026] (2) The initial threshold T is defined by the following formula:

[0027]

[0028] (4) According to the following formula, the standard deviation s is calculated according to the T value. i Divide into A and B sets,

[0029] A={s i |s i >T}

[0030] B={s i |s i ≤T}

[0031] (4) Calculate the new threshold T' as follows: i Divide again,

[0032]

[0033] in, is the mean of the standard deviations of set A and set B;

[0034] (5) Repeat steps (3) to (4) until the threshold T reaches a constant value. At this time, the final sets A' and B' are obtained, and the process energy consumption index corresponding to the standard deviation data in set A' is selected to enter the production energy consumption index set.

[0035] The normalization process includes the following steps:

[0036] Let parameter a ij is the value of the i-th row and j-th column of the data matrix V. Assume that there are m products to be evaluated and n evaluation indicators. Then i=1,2,3,…,m. And j=1,2,…,n. Use b ij Indicates a ij The normalized result of . The algorithm is described as follows. The reverse index means that the larger the index data, the worse the transformer product characteristics, while the positive index is the opposite.

[0037] (1) If a ij is a reverse indicator, the normalization algorithm is:

[0038]

[0039] (2) If a ij is a positive indicator, then the normalization algorithm is:

[0040]

[0041] Step 4 is as follows:

[0042] (1) Calculation of indicator variability

[0043] Suppose there are m samples and n indicators in the tth (t=1,2,3) m×n normalized indicator matrix, and S j represents the standard deviation of the jth indicator, x ij Represents the jth index value of the i-th sample, and the mean of this index is

[0044]

[0045] (2) Calculate the Pearson correlation coefficient r between indicator j and other indicators k in the normalized indicator matrix in turn according to the following formula jk , where k = 1, 2, ..., j-1, j+1, ..., n, represents the mean of the index k,

[0046]

[0047] (3) Calculate the information content C of the jth indicator in the normalized indicator matrix j ,

[0048]

[0049] (4) Calculate the CRITIC weight w of the jth indicator in the normalized indicator matrix Cj ,

[0050]

[0051] (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Ct ,

[0052] W Ct =(w C1 ,w C2 ,...,w Cn )

[0053] Repeat (1) to (5) to obtain the CRITIC weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. C1 , W C2 , W C3 ;

[0054] (6) is combined into the overall CRITIC weight vector W as follows C ,

[0055] W C =(0.2W C1 ,0.5W C2 ,0.3W C3 ) T .

[0056] Step 5 is as follows:

[0057] (1) Assume that there are m samples and n indicators in the t-th (t=1, 2, 3) m×n normalized indicator matrix. Calculate the i-th sample value x under the j-th indicator in the normalized indicator matrix. ij The proportion of this indicator ij :

[0058]

[0059] (2) Calculate the entropy value e of the j-th indicator in the normalized indicator matrix j :

[0060]

[0061] (3) Calculate the information entropy redundancy d of the jth indicator in the normalized indicator matrix j :

[0062] d j =1-e jj=1,2,...,n

[0063] (4) Calculate the weight w of each indicator in the normalized indicator matrix Qj :

[0064]

[0065] (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Qt :

[0066] W Qt =(w Q1 ,w Q2 ,...,w Qn )

[0067] Repeat (1) to (5) to obtain the entropy weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. Q1 , W Q2 , W Q3 ;

[0068] (6) is combined into the overall entropy weight vector W according to the following formula Q :

[0069] W Q =(0.2W Q1 ,0.5W Q2 ,0.3W Q3 ) T .

[0070] Step 5 specifically includes:

[0071] The objective function is constructed as:

[0072]

[0073] Among them, w (j) is the jth component of the optimal combined weight vector W; w C(j) 、w Q(j) are the overall CRITIC weight vector W C and the total entropy weight vector W Q The j-th component of ;

[0074] Using the Lagrange multiplier method to solve the objective function, we can get:

[0075]

[0076] Step 7 is as follows:

[0077] (1) Calculate the product score vector S:

[0078] S=W·X

[0079] Where X is the normalized overall data matrix of the product to be evaluated; W is the optimal combination weight vector;

[0080] (2) Perform percentage-based scoring correction to obtain the final product score vector S h :

[0081]

[0082] Among them, s i is the i-th score component of the product score vector S, s k is the final score vector S h The kth fractional component of .

[0083] (3) Determine product grade:

[0084] The final score of the specified product is S h 60 to 70 points is level IV, 70 to 80 points is level III, 80 to 90 points is level II, and 90 to 100 points is level I.

[0085] The present invention has the following advantages:

[0086] 1. Evaluation system of multi-dimensional data fusion

[0087] This method not only considers the quality inspection indicators of distribution transformers, but also incorporates production energy consumption indicators and carbon footprint characteristic indicators into the evaluation system. This multi-dimensional data fusion method can more comprehensively and accurately reflect the green production quality of distribution transformers.

[0088] 2. Screening of production energy consumption indicators based on K-means clustering algorithm

[0089] The production energy consumption indicators are screened through the K-means clustering algorithm to effectively identify the processes that have a greater impact on energy consumption, thereby constructing a more streamlined and targeted production energy consumption indicator set.

[0090] 3. Use the combined weight calculation method

[0091] The present invention adopts two weight calculation methods, the CRITIC method and the entropy weight method, at the same time, which can make full use of the complementarity of the two methods, thereby enhancing the accuracy and rationality of weight allocation. The CRITIC method can objectively reflect the importance of each indicator in the evaluation system by considering the contrast intensity and conflict between indicators; and the entropy weight law is based on the principle of information entropy, and determines the weight according to the degree of variation of the data, avoiding the interference of subjective factors. Combining the two, the intrinsic characteristics of the indicators and the data distribution characteristics can be comprehensively considered, making the weight allocation more scientific and comprehensive, thereby improving the reliability and effectiveness of the evaluation results. This combined method is particularly suitable for complex evaluation problems, and can more accurately reflect the contribution of each indicator in the comprehensive evaluation, providing strong support for scientific decision-making.

[0092] The present invention constructs the objective function based on the minimum discriminant information principle and solves the optimal combination weights by the Lagrange multiplier method, which can greatly reduce the differences in weights corresponding to different methods and make the combination weights scientific and representative.

[0093] 4. Correction of the percentage scoring system and determination of the final evaluation grade

[0094] After constructing the comprehensive evaluation model and calculating the score, the percentage system was used to correct the score and obtain the final score and corresponding evaluation level. This processing method makes the evaluation results more intuitive, easy to understand, and highly practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 Flowchart of the K-means clustering algorithm described in this method;

[0096] Figure 2 Schematic diagram of the comprehensive evaluation index system of this method;

[0097] Figure 3 Flowchart of this method. DETAILED DESCRIPTION

[0098] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0099] After collecting data, the K-means algorithm is used to construct an energy consumption index set. Combined with the other two types of indicators, the matching index matrix is ​​constructed. After the index matrix is ​​normalized, the weights are calculated using the CRITIC method and the entropy weight method to obtain the overall weight vector, and the optimal combination weight is calculated based on the minimum discriminant information principle. Finally, the evaluation model is constructed by combining the index system and the optimal weight, and the product score is calculated and corrected to obtain the final score and evaluation level.

[0100] Figure 3The schematic diagram of the process of the green quality comprehensive evaluation method for power transformer production provided by the present invention is divided into seven steps. Steps 1 to 3 are data collection and analysis; Steps 4 and 5 use two methods to calculate the weights of each indicator; Step 6 uses the minimum discriminant information principle to combine the two types of weights into the optimal weight. Finally, Step 7 uses the evaluation system to calculate the product score and the corresponding classification result.

[0101] The specific steps are detailed below.

[0102] Step 1: Collect data. Conduct quality inspection and production data collection on the transformer products to be evaluated. It is divided into the following three categories:

[0103] (1) Quality inspection index data: including 10 items of transformer products: high voltage / low voltage and ground insulation resistance, no-load loss, no-load current, short-circuit impedance, load loss, top liquid temperature rise, top oil temperature, no-load loss, and no-load current. The quality inspection index matrix V is composed of quality inspection index data. a .

[0104] (2) Production energy consumption index data: including the electrical energy consumption of the seven core production links, namely, oil tank production, clamp production, iron core production, insulation production, coil production, body drying and vacuum oil filling. The production energy consumption index matrix V is composed of the production energy consumption index data. b .

[0105] (3) Carbon footprint characteristic indicator data: including the proportion of clean energy, transportation distance of steel, copper and transformer oil, and price ratio of scrap copper and scrap steel. The carbon footprint characteristic indicator matrix V is constructed from the carbon footprint characteristic indicator data. c .

[0106] Step 2: Based on the K-means clustering algorithm, the production energy consumption index screening is carried out to obtain the production links with greater impact on energy consumption, and the production energy consumption index set is constructed to obtain the updated production energy consumption index matrix V b1 :

[0107] (1) The standard deviation s of the energy consumption data of the i-th production link is calculated using the following formula i :

[0108]

[0109] Among them, x ij is the energy consumption data corresponding to the jth sample to be evaluated in the i-th production link. There are m samples and 7 production energy consumption indicators; is the average value of the production energy consumption data of all samples to be evaluated in the i-th process, which is calculated as follows:

[0110]

[0111] (2) The initial threshold T is defined by the following formula:

[0112]

[0113] (3) According to the following formula, the standard deviation s is calculated according to the T value. i Divide into A and B sets,

[0114] A={s i |s i >T}

[0115] B={s i |s i ≤T}

[0116] (4) Calculate the new threshold T' as follows: i Divide again,

[0117]

[0118] in, is the mean of the standard deviations in set A and set B;

[0119] (5) Repeat steps (3) to (4) until the threshold T reaches a constant value. At this time, the final sets A' and B' are obtained, and the link energy consumption index corresponding to the standard deviation in set A' is selected to enter the production energy consumption index set.

[0120] Step 3: Data integration and normalization. Based on steps 1 and 2, determine the quality inspection index data matrix V a , production energy consumption index data matrix V b , Carbon footprint characteristic indicator data matrix V c The above matrices are normalized to form the corresponding normalized index data matrix H a , H b , H c . Details are as follows:

[0121] Let parameter a ij is the value of the i-th row and j-th column of the data matrix V. Assume that there are m products to be evaluated and n evaluation indicators. Then i=1,2,3,…,m. And j=1,2,…,n. Use b ij Indicates a ij The normalized result of . The algorithm is described as follows. The reverse index means that the larger the index data, the worse the transformer product characteristics, while the positive index is the opposite.

[0122] 1. If a ij is a reverse indicator, the normalization algorithm is:

[0123]

[0124] 2. If a ij is a positive indicator, then the normalization algorithm is:

[0125]

[0126] At the same time, the normalized overall data matrix X is established.

[0127] X=[H a ,H b ,H c ]

[0128] Step 4: Calculate the indicator weights based on the CRITIC method. Calculate H a , H b , H c The indicator weight W C1 , W C2 , W C3 . And finally combined into the overall CRITIC weight vector W C The steps are detailed below:

[0129] (1) Calculate the indicator variability according to the following formula.

[0130] Suppose there are m samples and n indicators in the tth (t=1,2,3) m×n normalized indicator matrix, and S j represents the standard deviation of the jth indicator, with a total of n indicators and m samples. ij Represents the jth index value of the i-th sample, and the mean of this index is

[0131]

[0132] (2) Calculate the Pearson correlation coefficient r between indicator j and all other indicators k in the normalized indicator matrix in turn according to the following formula jk . Where k = 1, 2, ..., j-1, j+1, ..., n, Represents the mean of the indicator k.

[0133]

[0134] (3) Calculate the indicator information C of the j-th indicator in the normalized indicator matrix j .

[0135]

[0136] (4) Calculate the CRITIC weight w of the jth indicator in the normalized indicator matrix Cj .

[0137]

[0138] (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Ct ,

[0139] W Ct =(w C1 ,w C2 ,...,w Cn )

[0140] Repeat (1) to (5) to obtain the CRITIC weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. C1 , W C2 , W C3 ;

[0141] (6) is combined into the overall CRITIC weight vector W as follows C .

[0142] W C =(0.2W C1 ,0.5W C2 ,0.3W C3 ) T .

[0143] Step 6: Weight calculation based on entropy weight method.

[0144] Calculate H a , H b , H c The indicator weight W Q1 , W Q2 , W Q3 . And finally combined into the overall entropy weight vector W Q The steps are detailed below:

[0145] (1) Assume that there are m samples and n indicators in the t-th (t=1, 2, 3) m×n normalized indicator matrix, and calculate the proportion p of the i-th sample value under the j-th indicator to the indicator. ij :

[0146]

[0147] (2) Calculate the entropy value e of the j-th indicator in the normalized indicator matrix j :

[0148]

[0149] (3) Calculate the information entropy redundancy d of the jth indicator in the normalized indicator matrix j :

[0150] d j =1-e j ,j=1,...,n

[0151] (4) Calculate the weight w of each indicator in the normalized indicator matrix Qj :

[0152]

[0153] (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Qt :

[0154] W Qt =(w Q1 ,w Q2 ,...,w Qn )

[0155] Repeat (1) to (5) to obtain the entropy weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. Q1 , W Q2 , W Q3 ;

[0156] (6) is combined into the overall entropy weight vector W according to the following formula Q .

[0157] W Q =(0.2W Q1 ,0.5W Q2 ,0.3W Q3 ) T

[0158] Step 6: Let the optimal combination weight vector be W. If there are n indicators in the population, then the vector has n indicator weight components. To make this weight consistent with the overall CRITIC weight vector W C , the total entropy weight vector W Q Based on the minimum discriminant information principle, the following objective function is constructed. (j) is the jth component of the optimal combination weight vector W. C(j) 、w Q(j) are the overall CRITIC weight vector W C and the total entropy weight vector W Q The j-th component of .

[0159]

[0160] Using the Lagrange multiplier method to solve the above problem, we can get:

[0161]

[0162] The optimal combination weight vector W is calculated according to this formula, where W = (w(j)), j = 1, 2, ..., n.

[0163] Step 7: Calculate the product score vector S based on the normalized overall data matrix X and the optimal combination weight vector W of the distribution transformer product to be evaluated, perform percentage scoring correction, and obtain the final score S h and the corresponding evaluation levels.

[0164] (1) Calculate the product score vector S as follows: Where X is the normalized overall data matrix of the product to be evaluated. W is the optimal combination weight vector in step 6.

[0165] S=X·W

[0166] (2) Perform the percentage scoring correction as follows to obtain the final product score vector S h This algorithm ensures that the converted scores are between 60 and 100, while also making their distribution more even.

[0167]

[0168] Among them, s i is the i-th score component of the product score vector S, s k is the final score vector S h The kth fractional component of .

[0169] (3) Determine the product grade as follows.

[0170] It is stipulated that products with a final score of 60 to 70 points are classified as Level IV, 70 to 80 points are classified as Level III, 80 to 90 points are classified as Level II, and 90 to 100 points are classified as Level I.

[0171] Table 1 Sample classification

[0172]

[0173] The present invention is further described below by taking the sampling inspection data of power material procurement of a local power grid company as an example.

[0174] Step 1: There are 8 samples to be tested from 8 manufacturers in this power transformer material inspection. The sample model is S20-400kVA three-dimensional coil oil-immersed transformer. The data collection and arrangement are shown in Table 2, Table 3, and Table 4. Determine the quality inspection index matrix V a , Carbon footprint characteristic indicator matrix V c As shown below.

[0175] Table 2 Sample production energy consumption index data table

[0176]

[0177] Table 3 Sample quality test index data table

[0178]

[0179] Table 4 Sample carbon footprint characteristic index data table

[0180]

[0181] Step 2: Based on the K-means clustering algorithm, the production energy consumption indicators are screened to obtain the three production links of coil manufacturing, body drying, and vacuum oiling, and the updated production energy consumption indicator matrix V is determined. b1 .

[0182] Step 3: Test the indicator matrix V for the above quantities a , Updated production energy consumption index matrix V b1 And the carbon footprint characteristic indicator matrix V c The matrices are normalized to form the corresponding normalized indicator data matrix H a , H b , H c .

[0183]

[0184]

[0185] Step 4: Calculate H based on the CRITIC method of the present invention a , H b , H c The indicator weight W C1 , W C2 , W C3 . And finally combined into the overall CRITIC weight vector W C The calculation results are shown below.

[0186] W C1 =(0.337,0.405,0.258)

[0187] W C2 =(0.095,0.112,0.097,0.102,0.108,0.098,0.094,0.091,0.103,0.101)

[0188] W C3 =(0.165,0.148,0.185,0.144,0.161,0.199)

[0189]

[0190] Step 5: Calculate H based on the entropy weight method of the present invention a , H b , H c The indicator weight W Q1 , W Q2 , W Q3 . And finally combined into the overall entropy weight vector W Q The calculation results are shown below.

[0191] W Q1 =(0.336,0.496,0.168)

[0192] W Q2 =(0.117,0.111,0.09,0.101,0.103,0.094,0.091,0.103,0.094,0.096)

[0193] W Q3 =(0.237,0.159,0.187,0.182,0.106,0.131)

[0194]

[0195] Step 6: The optimal combination weight W is calculated according to the method of the present invention as shown below.

[0196]

[0197] Step 7: Calculate the scores and grade the results according to the method of the present invention.

[0198] S=(0.60,0.56,0.38,0.64,0.51,0.45,0.55,0.47) T

[0199] S h =(94.00,87.14,60.00,100.00,79.55,71.04,85.89,74.02) T

[0200] Table 5 Sample evaluation scores and corresponding grades

[0201]

Claims

1. A comprehensive evaluation method for the production green quality of distribution transformers, characterized in that: The following steps are involved: Step 1: Data Collection Obtain the quality inspection index data, production energy consumption index data and carbon footprint characteristic index data of the distribution transformer product to be evaluated, and form the quality inspection index matrix V a , production energy consumption index matrix V b , Carbon footprint characteristic indicator matrix V c ; Step 2: Based on the K-means clustering algorithm, the production energy consumption indicators are screened to obtain the processes that have a greater impact on energy consumption, and the production energy consumption indicator set is constructed to obtain the updated production energy consumption indicator matrix V b1 ; Step 3: Data integration and normalization For the quality inspection index matrix V a , Updated production energy consumption index matrix V b1 and carbon footprint characteristic indicator matrix V c Perform normalization processing to form the corresponding normalized indicator matrix H a , H b , H c , and establish the normalized overall data matrix X as follows: X=[H a ,H b ,H c ]; Step 4: Calculation of indicator weights based on the CRITIC method The normalized index matrix H is calculated using the CRITIC method. a , H b , H c The corresponding indicator CRITIC weight W C1 , W C2 , W C3 , combined into the overall CRITIC weight vector W C ; Step 5: Weight calculation based on entropy weight method The entropy weight method is used to calculate the normalized index matrix H a , H b , H c The corresponding index entropy weight W Q1 , W Q2 , W Q3 , combined into the overall entropy weight vector W Q ; Step 6: Determine the optimal combination weight vector W So that the combined weights and the overall CRITIC weight vector W C , the total entropy weight vector W Q With the goal of being close to each other, the objective function is constructed according to the minimum discriminant information principle, and the Lagrange multiplier method is used to solve the optimal combination weight; Step 7: Calculate the comprehensive evaluation score and corresponding grade The product score S is calculated based on the normalized overall data matrix X and the optimal combination weight vector W of the distribution transformer product to be evaluated, and the percentage score correction is performed to obtain the final score S h and the corresponding evaluation levels.

2. The method for comprehensive evaluation of green quality of distribution transformer production according to claim 1, characterized in that: The quality inspection index data include the high-voltage and ground insulation resistance, low-voltage and ground insulation resistance, no-load loss, load loss, no-load current, short-circuit impedance, top liquid temperature rise, top oil temperature, low-voltage no-load loss and low-voltage no-load current data of the transformer products to be evaluated; the production energy consumption indicators include oil tank production, clamp production, core production, insulation manufacturing, coil manufacturing, body drying and vacuum oil filling data; the carbon footprint characteristic indicators include the proportion of clean energy, steel transportation distance, copper transportation distance, transformer oil transportation distance, scrap copper price ratio and scrap steel price ratio data.

3. The method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1 is characterized in that: Based on the K-means clustering algorithm, the production energy consumption indicators are screened to obtain the processes that have a greater impact on energy consumption, and the production energy consumption indicator set is constructed, including the following steps: (1) The standard deviation s of the energy consumption data of the i-th production link is calculated using the following formula i : Among them, x ij is the energy consumption data corresponding to the jth sample to be evaluated in the i-th production link. There are m samples and 7 production energy consumption indicators; It is the average value of the production energy consumption data of all samples to be evaluated in the i-th production link, which is calculated as follows: (2) The initial threshold T is defined by the following formula: (3) According to the following formula, the standard deviation s is calculated according to the T value. i Divide into A and B sets, A={s i |s i >T} B={s i |s i ≤T} (4) Calculate the new threshold T' as follows: i Divide again, in, is the mean of the standard deviations of set A and set B; (5) Repeat steps (3) to (4) until the threshold T reaches a constant value. At this time, the final sets A' and B' are obtained, and the process energy consumption index corresponding to the standard deviation data in set A' is selected to enter the production energy consumption index set.

4. The method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1, characterized in that: The normalization process includes the following steps: Let parameter a ij is the value of the i-th row and j-th column of the data matrix V. Assume that there are m products to be evaluated and n evaluation indicators. Then i=1,2,3,…,m. And j=1,2,…,n. Use b ij Indicates a ij The normalized result of . The algorithm is described as follows. The reverse index means that the larger the index data, the worse the transformer product characteristics, while the positive index is the opposite. (1) If a ij is a reverse indicator, the normalization algorithm is: (2) If a ij is a positive indicator, then the normalization algorithm is:

5. The method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1, characterized in that: Step 4 is as follows: (1) Calculation of indicator variability Suppose there are m samples and n indicators in the tth (t=1,2,3) m×n normalized indicator matrix, and S j represents the standard deviation of the jth indicator, x ij Represents the jth index value of the i-th sample, and the mean of this index is (2) Calculate the Pearson correlation coefficient r between indicator j and other indicators k in the normalized indicator matrix in turn according to the following formula jk , where k = 1, 2, ..., j-1, j+1, ..., n, represents the mean of the index k, (3) Calculate the information content C of the jth indicator in the normalized indicator matrix j , (4) Calculate the CRITIC weight w of the jth indicator in the normalized indicator matrix Cj , (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Ct , IN Ct =(in C1 ,In C2 ,...,In Cn ) Repeat (1) to (5) to obtain the CRITIC weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. C1 , W C2 , W C3 Finally, the overall CRITIC weight vector W is obtained by combining the following formula: C , <h2 style=";text-align:left;direction:ltr">W<h2 style=";text-align:left;direction:ltr"> C <h2 style=";text-align:left;direction:ltr"> =(0.2W<h2 style=";text-align:left;direction:ltr"> C1 <h2 style=";text-align:left;direction:ltr"> 0.5W<h2 style=";text-align:left;direction:ltr"> C2 <h2 style=";text-align:left;direction:ltr"> 0.3W<h2 style=";text-align:left;direction:ltr"> C3 <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> 。 6. A method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1, characterized in that: Step 5 is as follows: (1) Assume that there are m samples and n indicators in the t-th (t=1, 2, 3) m×n normalized indicator matrix. Calculate the i-th sample value x under the j-th indicator in the normalized indicator matrix. ij The proportion of this indicator ij : (2) Calculate the entropy value e of the j-th indicator in the normalized indicator matrix j : (3) Calculate the information entropy redundancy d of the jth indicator in the normalized indicator matrix j : d j =1-e j j=1,2,...,n (4) Calculate the weight w of each indicator in the normalized indicator matrix Qj : (5) Calculate all n indicators in the m×n normalized indicator matrix to obtain the weight vector W corresponding to the normalized indicator matrix Qt : IN Qt =(in Q1 ,In Q2 ,...,In Qn ) Repeat (1) to (5) to obtain the entropy weight vector W corresponding to the three normalized indicator matrices Ha, Hb, and Hc. Q1 , W Q2 , W Q3 ; (6) is combined into the overall entropy weight vector W according to the following formula Q : <h2 style=";text-align:left;direction:ltr">W<h2 style=";text-align:left;direction:ltr"> Q <h2 style=";text-align:left;direction:ltr"> =(0.2W<h2 style=";text-align:left;direction:ltr"> Q1 <h2 style=";text-align:left;direction:ltr"> 0.5W<h2 style=";text-align:left;direction:ltr"> Q2 <h2 style=";text-align:left;direction:ltr"> 0.3W<h2 style=";text-align:left;direction:ltr"> Q3 <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> 。 7. A method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1, characterized in that: Step 5 specifically includes: The objective function is constructed as: Among them, w (j) is the jth component of the optimal combined weight vector W; w C(j) 、w Q(j) are the overall CRITIC weight vector W C and the total entropy weight vector W Q The j-th component of ; Using the Lagrange multiplier method to solve the objective function, we can get:

8. The method for comprehensive evaluation of green quality of production of distribution transformers according to claim 1, characterized in that: Step 7 is as follows: (1) Calculate the product score vector S: S=W·X Among them, X is the normalized overall data matrix of the product to be evaluated; W is the optimal combination weight vector; (2) Perform percentage-based scoring correction to obtain the final product score vector S h : Among them, s i is the i-th score component of the product score vector S, s k is the final score vector S h The kth fractional component of . (3) Determine product grade: The final score of the specified product is S h 60 to 70 points is level IV, 70 to 80 points is level III, 80 to 90 points is level II, and 90 to 100 points is level I.

Images