Distribution transformer new energy efficiency performance analysis method based on principal component analysis algorithm

Abstract

The invention discloses a distribution transformer new energy efficiency performance analysis method based on a principal component analysis algorithm. The method comprises the following steps: collecting index data about a transformer in data; calling a correlation analysis model, extracting index data of a plurality of groups of transformers with different capacities, performing correlation analysis, and determining correlation among the index data; according to a principal component analysis algorithm, analyzing the index data with correlation of different types of transformers, and performing dimensionality reduction on the number of indexes; analyzing the effectiveness of the indexes obtained through the principal component analysis model through a factor analysis method; and performing energy efficiency analysis and screening on the transformer by using indexes capable of effectively evaluating the new energy efficiency performance of the distribution transformer. The new energy efficiency performance analysis method for the distribution transformer based on the principal component analysis algorithm has the characteristics of being capable of automatically screening and preliminarily reviewing bidding documents and improving the reviewing efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy efficiency analysis of electric power equipment, and in particular relates to a new energy efficiency performance analysis method for distribution transformers based on a principal component analysis algorithm. Background Art

[0002] The new power system has the characteristics of clean, low-carbon, safe and controllable. The safe and reliable operation of the power grid is related to thousands of households, so the importance of power materials is self-evident. Selecting the best power materials is not only a basic requirement for material management, but also a strong guarantee for the safe operation of the power grid. However, the bidding materials for power materials have problems such as complex information and data standardization. Even the differences in the evaluation angles and experience of the evaluation experts will affect the selection of power materials.

[0003] Transformers are basic equipment for power transmission and distribution, and are widely used in industries, agriculture, transportation, urban communities, etc. Transformers come in many types and each type includes many parameters, such as magnetic flux density, total mass, number of silicon steel sheets, number of flat copper wires, transformer oil quality, steel quantity, loss, capacity, etc. Therefore, a method is urgently needed to analyze the performance of many types of transformers, so as to facilitate the intelligent evaluation of transformers and improve the quality and efficiency of the evaluation. Summary of the invention

[0004] The purpose of the present invention is to provide a new energy efficiency performance analysis method for distribution transformers based on a principal component analysis algorithm, which has the characteristics of being able to automatically screen and preliminarily review bidding documents and improve review efficiency.

[0005] The technical solution adopted by the present invention is a new energy efficiency performance analysis method for distribution transformers based on a principal component analysis algorithm, which specifically includes the following steps:

[0006] Step 1: Collect the indicator data about transformers in the information to form a data set;

[0007] Step 2: Call the correlation analysis model, extract several groups of indicator data of transformers with different capacities, perform correlation analysis, and determine the correlation between the indicator data;

[0008] Step 3: According to the principal component analysis algorithm, the relevant index data of different types of transformers are analyzed respectively, and the number of indicators is reduced;

[0009] Step 4: Analyze the effectiveness of the indicators obtained by the principal component analysis algorithm through factor analysis;

[0010] Step 5: Use indicators that can effectively evaluate the new energy efficiency performance of distribution transformers to analyze and screen transformers for energy efficiency.

[0011] The present invention is also characterized in that:

[0012] The indicator data of the transformer in step 1 include: magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity; according to the different transformer capacities, the indicator data are labeled and divided to form different data sets.

[0013] Step 2 is specifically as follows: build a virtual environment through anaconda, import the correlation analysis model in pycharm, extract several groups of transformer index data with different capacities, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and no-load loss, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and load loss; use the linear analysis model to use a dot matrix analysis diagram to reflect the relationship between the indicators, and use the dot matrix analysis diagram to find the correlation curve between the indicators to determine the correlation between the indicator data.

[0014] Step 3 specifically includes:

[0015] Step 3.1: Select n indicators from the transformer's magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity indicators, and organize the original data of the n indicators into an n-row and m-column matrix by column, where m represents the amount of data contained in each indicator;

[0016] Step 3.2: Zero-mean each row of the matrix to obtain a new matrix X;

[0017] Step 3.3: Find the covariance matrix P of matrix X. The formula is as follows:

[0018]

[0019] Step 3.4: Find the eigenvalues ​​and corresponding eigenvectors of the matrix P;

[0020] Step 3.5: Organize the eigenvectors corresponding to the eigenvalues ​​from large to small into a matrix Y from top to bottom, and take the first k rows to form a matrix Q;

[0021] Step 3.6: Construct the matrix Z=kQ, which is the data matrix after dimension reduction to k dimension, and obtain the contribution of each indicator; according to the contribution, select 2 to 3 indicators for evaluating transformer efficiency.

[0022] Step 4 specifically includes:

[0023] The transformer's magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel mass, transformer oil mass, no-load loss and load loss indicators are subjected to KMO inspection. The KMO calculation formula is as follows:

[0024]

[0025] In the formula, r ij represents the correlation coefficient, i represents the number of rows, j represents the number of columns, and k is 10;

[0026] The KMO value of the indicator is compared with the indicator after dimensionality reduction by the principal component analysis algorithm to determine the effectiveness of the indicator obtained by the principal component analysis algorithm.

[0027] The beneficial effects of the present invention are:

[0028] The present invention proposes a new energy efficiency performance analysis method for distribution transformers based on the principal component analysis algorithm. Through intelligent means, big data analysis and artificial intelligence technology are used to automatically screen and preliminarily review bidding documents, thereby reducing the workload of experts and improving review efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of a new energy efficiency performance analysis method of a distribution transformer based on a principal component analysis algorithm of the present invention;

[0030] Figure 2 It is a relationship diagram between silicon steel sheet quality and no-load loss in the new energy efficiency performance analysis method of the distribution transformer based on the principal component analysis algorithm of the present invention;

[0031] Figure 3 It is a relationship diagram between silicon steel sheet quality and load loss in the new energy efficiency performance analysis method of the distribution transformer based on the principal component analysis algorithm of the present invention;

[0032] Figure 4 It is a scree map of a silicon steel sheet oil-immersed transformer in a new energy efficiency performance analysis method for a distribution transformer based on a principal component analysis algorithm of the present invention;

[0033] Figure 5 It is a thermal diagram of a silicon steel sheet oil-immersed transformer in a new energy efficiency performance analysis method for distribution transformers based on a principal component analysis algorithm of the present invention. DETAILED DESCRIPTION

[0034] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] The present invention provides a new energy efficiency performance analysis method for distribution transformers based on principal component analysis algorithm, such as Figure 1 As shown, the specific steps include:

[0036] Step 1: Collect the indicator data about the transformer in the data and convert it into array form;

[0037] The data on magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, steel quality, transformer oil quality, no-load loss, and load loss of silicon steel sheet oil-immersed transformers, silicon steel sheet dry-type transformers, and amorphous alloy oil-immersed transformers are collected. According to the different transformer capacities, the indicator data are labeled and divided to form different data sets.

[0038] Step 2: Call the correlation analysis model, extract several groups of transformer index data with different capacities, perform correlation analysis, and determine the correlation between the index data;

[0039] The correlation analysis model of the present invention adopts a linear regression method. Linear regression is one of the most basic regression analysis methods, which aims to model the relationship between one or more independent variables and a continuous dependent variable as a linear equation. This method can be used to predict the value of the dependent variable, and can also be used to describe the relationship between the independent variable and the dependent variable. There are three types of linear correlation, namely: positive correlation, negative correlation, and no correlation (random). Therefore, the parameters describing the correlation must have two functions: the direction of correlation and the degree of correlation. For example, when the linear correlation is positive, the parameter is greater than 0; when the linear correlation is negative, the parameter is less than 0; and the larger the parameter value, the stronger the linear correlation. The correlation parameters can be obtained through the covariance:

[0040] cov(X,Y)=E[(X-μX)(Y-μY)]

[0041] If the covariance is positive, it means that X and Y change in the same direction. The larger the covariance value, the higher the degree of the same direction; vice versa. The covariance will be affected by the change range of X and Y themselves. The correlation coefficient r is defined as follows:

[0042]

[0043] The correlation coefficient r is the ratio of the covariance of X and Y to the standard deviation of X and Y, which is used to eliminate the fluctuation of their own amplitude. Therefore, the correlation coefficient ρ can more intuitively express the correlation between variables. The correlation is defined in the range [-1,1], 1 means completely linear positive correlation, -1 means completely linear negative correlation, and 0 means completely uncorrelated (random);

[0044] In Python, the correlation coefficient ρ of two data sets can be calculated using the corr() function.

[0045] The linear regression algorithm aims to find a straight line equation, that is, the linear regression equation, expressed as y=a+bx, to simulate the correlation between two data sets {x}, {y}, where a is the intercept and b is the regression coefficient. The goal of linear regression is to find the best fit line, which fits as many data points in the data set as possible. The best fit is judged by the coefficient of determination R; the closer the coefficient of determination is to 1, the more accurate the fit. The coefficient of determination R is expressed by the following formula:

[0046]

[0047] In the formula, SSR (Sum of Squares Residual) is the residual sum of squares, which represents the difference between the fitted curve and the actual observed values; SST (Sum of Squares Total) is the total sum of squares, which represents the difference between the actual observed values ​​and their means.

[0048] The specific calculation process is as follows:

[0049] First, we need to calculate the mean of the actual observations, denoted as y mean .

[0050] Calculate the residual sum of squares SSR. Process each observation, calculate the difference between the fitted curve and the actual observation, and add the squares of the differences, expressed as follows:

[0051]

[0052] In the formula, y i is the i-th observation, y predi is the predicted value corresponding to the fitting curve.

[0053] Calculate the total sum of squares SST. Process each observation, calculate the difference between the actual observation and the mean, and add the squares of the differences, expressed as follows:

[0054] SST=∑((y i -y mean ) 2 )

[0055] The determination coefficient R was calculated from the calculated SSR and SST.

[0056] The interpretation of the coefficient of determination is as follows:

[0057] When R 2 When it is close to 1, it means that the fitting curve can explain most of the variation in the data and the fitting effect is good;

[0058] When R 2 When it is close to 0, it means that the fitting curve cannot explain the variation of the data and the fitting effect is poor;

[0059] When R 2 =0, it means that the fitting curve cannot explain the variation of the target variable;

[0060] When R 2 When =1, it means that the fitting curve completely explains the variation of the target variable and the fitting effect is very good.

[0061] 10 groups of transformer indicator data of different capacities were extracted, and the training set and test set were randomly split from the selected indicator data in a ratio of 8:2. The LinearRegression() function was called in python to establish a linear regression model, and the fit() function was used to calculate the best fit line. The relationship between magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality and no-load loss was analyzed respectively, and the relationship between magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality and load loss was analyzed respectively; a dot matrix diagram was obtained Figure 2 and Figure 3 , the origin in the figure is the training set, and the triangular points are the training data set. Figures 2-3 It can be seen that the quality of silicon steel sheets has a linear correlation with no-load loss and load loss.

[0062] Through correlation analysis, it was determined that the collected indicators were correlated.

[0063] Step 3: According to the principal component analysis algorithm, the relevant indicator data of different types of transformers are analyzed respectively, and the number of indicators is reduced.

[0064] Import the transformer index data in step 1 into pycharm, and list the correlation matrix as shown in Table 1 based on the correlation obtained in step 2. Then, through the eigenvalue and eigenvector calculation modules in the PCA principal component analysis algorithm, obtain the eigenvalues ​​and contributions of the magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, steel quality, transformer oil quality, no-load loss, and load loss of 9 indicators of silicon steel sheet dry-type transformer, silicon steel sheet oil-immersed transformer, and amorphous alloy oil-immersed transformer, respectively. Use the idea of ​​dimensionality reduction to remove noise and unnecessary features, thereby realizing the speed processing of data. Reduce 9 indicators to 2-3 indicators, highlight the main contradictions, and thus reveal the regularity of internal variables.

[0065] The specific operation of the principal component analysis algorithm is as follows: there are m 9-dimensional data:

[0066] Step 3.1: Select 9 indicators from the transformer's magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss and load loss indicators, and organize the original data of the 9 indicators into a 9-row m-column matrix by column;

[0067] Step 3.2: Zero-mean each row of the matrix to obtain a new matrix X;

[0068] Step 3.3: Find the covariance matrix P of matrix X. The formula is as follows:

[0069]

[0070] Step 3.4: Find the eigenvalues ​​and corresponding eigenvectors of the matrix P;

[0071] Step 3.5: Organize the eigenvectors corresponding to the eigenvalues ​​from large to small into a matrix Y from top to bottom, and take the first k rows to form a matrix Q;

[0072] Step 3.6: Construct the matrix Z=kQ, which is the data matrix after dimension reduction to k dimension, and obtain the contribution of each indicator; according to the contribution, select 2 to 3 indicators for evaluating transformer efficiency.

[0073] The principal component coefficient matrix is ​​a matrix composed of the eigenvectors corresponding to each variable when performing principal component analysis. The principal component coefficient matrix is ​​calculated from the covariance matrix of the data. Each element in the principal component coefficient matrix represents the importance of the variable in calculating the principal component. The principal component coefficient matrix is ​​an orthogonal matrix, usually with a value between -1 and 1, which is used to measure the similarity and difference between variables. Table 1 is the correlation coefficient matrix of silicon steel sheet oil-immersed transformer.

[0074] Table 1 Correlation coefficient matrix of main components of silicon steel sheet oil-immersed transformer

[0075]

[0076]

[0077] The contributing rate of principal component refers to the proportion of the variance of the principal component in the total variance of the random variable under investigation. It is mainly used to measure the explanatory power of the principal component for the variability of the original variable. The cumulative contribution rate of the principal component is an important basis for selecting effective principal components. Table 2 shows the characteristic values, contribution rates, and cumulative contribution rates of silicon steel sheet oil-immersed transformers;

[0078] Table 2 Characteristic values, contribution rates, and cumulative contribution rates of each main component of silicon steel sheet oil-immersed transformer

[0079] principal component Eigenvalue Contribution % Cumulative contribution % Total mass 5.26026056 58.44 58.44 Silicon steel sheet 1.26750182 14.08 72.53 Transformer oil 0.91940091 10.21 82.75 Steel 0.64449782 7.16 89.91 Flat copper wire 0.44217972 4.91 94.82 Copper Foil 0.21529397 2.39 97.21 No-load loss 0.19552676 2.17 99.39 Load loss 0.03915222 0.44 99.82 capacity 0.01618521 0.18 100

[0080] Scree Plot is a visualization tool for principal component analysis (PCA) that helps us decide how many principal components (PCs) to keep in a data set. In PCA, each principal component represents a source of variance in the data. The first principal component explains the largest part of the variance in the data, the second principal component explains the largest part of the remaining variance, and so on. Each subsequent principal component explains the largest part of the variance not explained by the previous principal component, which is represented by Figure 4 It can be seen that there are two principal components with eigenvalues ​​greater than 1. By analyzing the mutation of characteristic root attenuation, it can be determined that two principal components should be extracted.

[0081] The results of PCA can be displayed through heat maps. In the PCA heat map, the horizontal axis represents the samples or observation points of the data, and the vertical axis represents the number of the principal component (PC). The color of each grid represents the projection value of the corresponding sample on the principal component. The darker the color, the larger the projection value, and the lighter the color, the smaller the projection value or close to 0. Heat maps can help us intuitively understand the distribution of data on the principal components. If a principal component can well explain the changes in the original data, it has a higher importance. On the contrary, if the projection value of a principal component is close to zero in most samples, it means that the principal component has a weaker ability to explain the data, such as Figure 5 As shown, the total mass and silicon steel sheet mass are darker in color and have higher importance.

[0082] Step 4: Analyze the effectiveness of the indicators obtained by the principal component analysis algorithm through factor analysis;

[0083] Factor analysis method, identifies and extracts a set of hidden potential factors (also called factors), finds hidden representative factors from a large number of variables, these factors represent a basic structure, can reflect the main information of the original variables. The present invention adopts KMO test. KMO (Kaiser-Meyer-Olkin) test statistic is an indicator for comparing simple correlation coefficients and partial correlation coefficients between variables. It is mainly used in factor analysis of multivariate statistics. KMO statistic takes values ​​between 0 and 1.

[0084] When the sum of squares of simple correlation coefficients among all variables is much larger than the sum of squares of partial correlation coefficients, the closer the KMO value is to 1, the stronger the correlation among variables is, and the more suitable the original variables are for factor analysis; when the sum of squares of simple correlation coefficients among all variables is close to 0, the closer the KMO value is to 0, the weaker the correlation among variables is, and the less suitable the original variables are for factor analysis. The calculation formula is as follows:

[0085]

[0086] In the formula, r ij represents the correlation coefficient, which is calculated in step 2; i represents the number of rows, j represents the number of columns, and k is 10;

[0087] Principle of the test: If there are common factors in the original data, the partial correlation coefficients between the variables should be very small. At this time, the value of KMO is close to 1. Therefore, the original data is suitable for factor analysis.

[0088] Kaiser gave the commonly used KMO metric standard: 0.9 and above means very suitable; 0.8 means suitable; 0.7 means average; 0.6 means not very suitable; and below 0.5 means extremely unsuitable. The KMO statistic takes values ​​between 0 and 1. When the sum of squares of simple correlation coefficients between all variables is much larger than the sum of squares of partial correlation coefficients, the KMO value is close to 1. The closer the KMO value is to 1, the stronger the correlation between variables, and the more suitable the original variables are for factor analysis; when the sum of squares of simple correlation coefficients between all variables is close to 0, the KMO value is close to 0. The closer the KMO value is to 0, the weaker the correlation between variables, and the less suitable the original variables are for factor analysis.

[0089] Based on the above, transformers are divided into three categories, and the transformer flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, capacity, no-load loss and load loss indicators are tested by KMO. The KMO values ​​of the indicators are compared with the indicators after dimensionality reduction by the principal component analysis algorithm to determine the effectiveness of the indicators obtained by the principal component analysis algorithm. Table 3 shows the KMO values ​​of silicon steel sheet oil-immersed transformers.

[0090] Table 3 KMO value of silicon steel sheet oil immersed transformer

[0091]

[0092] It can be seen from the data in Table 2 and Table 3 that the three indicators of total mass, silicon steel sheet quality and transformer oil quality obtained by the method of the present invention using PCA analysis and dimensionality reduction are consistent with the results obtained by KMO detection.

[0093] Step 5: Use indicators that can effectively evaluate the new energy efficiency performance of distribution transformers to analyze and screen transformers for energy efficiency.

[0094] Example 1

[0095] This embodiment provides a new energy efficiency performance analysis method for distribution transformers based on the principal component analysis algorithm. Figure 1 As shown, the specific steps include:

[0096] Step 1: Collect the indicator data about transformers in the information to form a data set;

[0097] Step 2: Call the correlation analysis model, extract several groups of indicator data of transformers with different capacities, perform correlation analysis, and determine the correlation between the indicator data;

[0098] Step 3: According to the principal component analysis algorithm, the relevant index data of different types of transformers are analyzed respectively, and the number of indicators is reduced;

[0099] Step 4: Analyze the effectiveness of the indicators obtained by the principal component analysis algorithm through factor analysis;

[0100] Step 5: Use indicators that can effectively evaluate the new energy efficiency performance of distribution transformers to analyze and screen transformers for energy efficiency.

[0101] Example 2

[0102] Based on Example 1, the indicator data of the transformer described in step 1 include: magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss and load loss; according to the different transformer capacities, the indicator data are labeled and divided to form different data sets.

[0103] Step 2 is specifically as follows: build a virtual environment through anaconda, import the correlation analysis model in pycharm, extract several groups of transformer index data with different capacities, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and no-load loss, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and load loss; use the linear analysis model to use a dot matrix analysis diagram to reflect the relationship between the indicators, and use the dot matrix analysis diagram to find the correlation curve between the indicators to determine the correlation between the indicator data.

[0104] Example 3

[0105] Based on Example 2, step 3 specifically includes:

[0106] Step 3.1: Select 9 indicators from the magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity indicators of the transformer. The 9 indicators are specifically magnetic flux density, total mass, silicon steel sheet mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity, and organize the original data of the 9 indicators into a 9-row m-column matrix by column, where m represents the amount of data contained in each indicator;

[0107] Step 3.2: Zero-mean each row of the matrix to obtain a new matrix X;

[0108] Step 3.3: Find the covariance matrix P of matrix X. The formula is as follows:

[0109]

[0110] Step 3.4: Find the eigenvalues ​​and corresponding eigenvectors of the matrix P;

[0111] Step 3.5: Organize the eigenvectors corresponding to the eigenvalues ​​from large to small into a matrix Y from top to bottom, and take the first k rows to form a matrix Q;

[0112] Step 3.6: Construct the matrix Z=kQ, which is the data matrix after dimension reduction to k dimension, and obtain the contribution of each indicator; according to the contribution, select 2 to 3 indicators for evaluating transformer efficiency.

[0113] Step 4 specifically includes:

[0114] The transformer's magnetic flux density, total mass, silicon steel sheet mass, flat copper wire mass, steel material mass, transformer oil mass, no-load loss, load loss and capacity index are subjected to KMO inspection. The KMO calculation formula is as follows:

[0115]

[0116] In the formula, r ij represents the correlation coefficient, i represents the number of rows, j represents the number of columns, and k is 10;

[0117] The KMO value of the indicator is compared with the indicator after dimensionality reduction by the principal component analysis algorithm to determine the effectiveness of the indicator obtained by the principal component analysis algorithm.

Claims

1. A new energy efficiency performance analysis method for distribution transformers based on principal component analysis algorithm, characterized in that: The specific steps include: Step 1: Collect the indicator data about transformers in the information to form a data set; Step 2: Call the correlation analysis model, extract several groups of indicator data of transformers with different capacities, perform correlation analysis, and determine the correlation between the indicator data; Step 3: According to the principal component analysis algorithm, the relevant index data of different types of transformers are analyzed respectively, and the number of indicators is reduced; Step 4: Analyze the effectiveness of the indicators obtained by the principal component analysis algorithm through factor analysis; Step 5: Use indicators that can effectively evaluate the new energy efficiency performance of distribution transformers to analyze and screen transformers for energy efficiency.

2. The new energy efficiency performance analysis method for distribution transformers based on principal component analysis algorithm according to claim 1 is characterized in that: The indicator data of the transformer described in step 1 include: magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity; according to the different transformer capacities, the indicator data are labeled and divided to form different data sets.

3. The new energy efficiency performance analysis method for distribution transformer based on principal component analysis algorithm according to claim 1 is characterized in that: Step 2 is specifically as follows: build a virtual environment through anaconda, import the correlation analysis model in pycharm, extract several groups of transformer index data with different capacities, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and no-load loss, and analyze the relationship between magnetic flux density, total mass, silicon steel sheet quality, amorphous alloy quality, copper foil quality, copper wire quality, flat copper wire quality, steel quality, transformer oil quality and load loss; With the help of linear analysis model, the relationship between indicators is reflected by dot matrix analysis diagram. At the same time, the correlation curve between indicators is obtained with the help of dot matrix analysis diagram to determine the correlation between indicator data.

4. The new energy efficiency performance analysis method for distribution transformer based on principal component analysis algorithm according to claim 1 is characterized in that: Step 3 specifically includes: Step 3.1: Select n indicators from the transformer's magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel quality, transformer oil quality, no-load loss, load loss and capacity indicators, and organize the original data of the n indicators into an n-row and m-column matrix by column, where m represents the amount of data contained in each indicator; Step 3.2: Zero-mean each row of the matrix to obtain a new matrix X; Step 3.3: Find the covariance matrix P of matrix X. The formula is as follows: Step 3.4: Find the eigenvalues ​​and corresponding eigenvectors of the matrix P; Step 3.5: Organize the eigenvectors corresponding to the eigenvalues ​​from large to small into a matrix Y from top to bottom, and take the first k rows to form a matrix Q; Step 3.6: Construct the matrix Z=kQ, which is the data matrix after dimension reduction to k dimension, and obtain the contribution of each indicator; according to the contribution, select 2 to 3 indicators for evaluating transformer efficiency.

5. The new energy efficiency performance analysis method for distribution transformer based on principal component analysis algorithm according to claim 1 is characterized in that: Step 4 specifically includes: The transformer's magnetic flux density, total mass, silicon steel sheet mass, amorphous alloy mass, copper foil mass, copper wire mass, flat copper wire mass, steel mass, transformer oil mass, no-load loss and load loss indicators are subjected to KMO inspection. The KMO calculation formula is as follows: In the formula, r ij represents the correlation coefficient, i represents the number of rows, j represents the number of columns, and k is 10; The KMO value of the indicator is compared with the indicator after dimensionality reduction by the principal component analysis algorithm to determine the effectiveness of the indicator obtained by the principal component analysis algorithm.

Images