Method for optimizing pressing rivet quality based on grey correlation and factor analysis method

Abstract

The invention discloses a method for optimizing pressing rivet quality based on grey correlation and a factor analysis method, and relates to the technical field of pressing rivet quality optimization. Firstly, a taguchi method is adopted, a press riveting quality test model with busbar design parameters and control parameters of an automatic squeeze riveter as design optimization variables and with push-out force and twisting force as responses is constructed, a grey correlation method is adopted, the correlation degree in all responses is analyzed, dimension reduction is conducted on multiple indexes in combination with a factor analysis method, weights of different indexes are obtained, and a press riveting quality test model is obtained. Reducing multiple indexes into a single index, then calculating a weighted grey correlation degree, carrying out comprehensive evaluation, finally obtaining an optimal parameter level combination, and analyzing a factor with the maximum influence degree on the indexes. According to the method, the factors which have the most influence on the pressing rivet quality can be quickly selected from the numerous parameters, and when the pressing rivet quality is evaluated by the push-out force and the twisting force in the pressing rivet quality evaluation indexes, the problems of information overlapping of multiple evaluation indexes and the like can be solved.

Description

Technical Field

[0001] The invention belongs to the technical field of press riveting quality optimization, and relates to a method for optimizing press riveting quality based on grey correlation and factor analysis. Background Art

[0002] The riveting nut and busbar connection are important electrical connectors in the high-voltage distribution system of electric vehicles. The working environment of the riveting nut and busbar puts higher requirements on the riveting quality. Common influencing factors in the riveting process are: busbar design parameters: hole diameter, material and thickness, automatic riveting machine control parameters: pressing loading force, pressing speed, pressing holding time and the distance between the center line of the hole and the edge, etc. The riveting quality evaluation indicators are: push-out force, twist-out force and limit installation torque, etc.

[0003] Riveting strength is an important indicator to measure the quality of riveting, which mainly includes push-out force, twist-out force and maximum installation torque. Optimizing the quality of riveting can ensure a stronger connection between riveted parts and reduce problems such as loosening and falling off caused by insufficient riveting strength. Summary of the invention

[0004] In order to make up for the deficiencies of the prior art, the present invention provides a method for optimizing the press riveting quality based on grey correlation and factor analysis, which can quickly select the most influential factors on the press riveting quality from a large number of parameters, and when the push-out force, twist-out force and limit installation torque are used to evaluate the press riveting quality in the press riveting quality evaluation indicators, the problem of information overlap of multiple evaluation indicators can be solved. The present invention is achieved through the following technical solutions: The present invention provides a method for optimizing the press riveting quality based on grey correlation and factor analysis, including adopting the Taguchi method, constructing a press riveting quality test model with busbar design parameters and automatic press riveting machine control parameters as design optimization variables and push-out force and twist-out force as responses;

[0005] Multi-objective optimization methods based on grey relational analysis include:

[0006] Step S1.1, determine the evaluation indicators according to the evaluation purpose and collect evaluation data; according to the number of indicators to be evaluated, set n data sequences to be processed to form the following matrix:

[0007]

[0008] Where m is the number of evaluation indicators,

[0009] X′ i =(x′ i (1), x′ i (2) x′ i (m)) T

[0010] A sequence of all data for a certain indicator;

[0011] i=1,2,…,n

[0012] is the number of different evaluation indicators;

[0013] Step S1.2, determining the reference sequence;

[0014] The optimal value of each evaluation index constitutes a reference sequence. The optimal value, that is, the value with the largest positive impact, is the optimal value, and the value with the smallest negative impact is the optimal value. The reference sequence is recorded as X′0=(x′0(1), x′0(2), …, x′0(m)) T ;

[0015] Step S1.3, in order to eliminate the influence of different dimensions of different evaluation indicators, the evaluation indicator data is normalized, and the evaluation indicator data is normalized using the mean method.

[0016]

[0017] Where: x i (k) is the normalized index data;

[0018] x′ i (k) is the kth data of the i-th evaluation index;

[0019] Step S1.4, calculate the evaluation index sequence x for each evaluated object i The absolute difference between the corresponding elements of (k) and the reference sequence x0(k), that is:

[0020] |x0(k)-x i (k)|

[0021] Where: x i (k) is the kth data of the i-th evaluation index;

[0022] x0(k) is the kth data of the reference sequence;

[0023] Step S1.5, determine the minimum difference between the two levels The maximum difference between the two levels

[0024] Step S1.6, calculating the correlation coefficient; using the following formula, respectively calculate the correlation coefficient of the comparison sequence of each evaluation index and the corresponding element of the reference sequence;

[0025]

[0026] Where ρ is the resolution factor, which is 0.5;

[0027] Step S1.7, calculating the correlation degree;

[0028] Calculate the correlation coefficients between the corresponding elements of the evaluation index sequences and the reference sequences for each evaluation object to reflect the relationship between each evaluation object and the reference sequence;

[0029] The steps of factor analysis include,

[0030] Step 2.1, judge the prerequisite conditions for factor analysis; conduct Bartlett's sphericity test and KMO test on the original variable data, or calculate the correlation coefficients between the original variables;

[0031] Step 2.2, process the sample indicators;

[0032] Suppose there are n samples and p original variables, which are represented by x1, x2,....x p respectively, and the mean of each variable is 0 and the standard deviation is 1. Now, each original variable is represented by a linear combination of k (k < p) factors f1, f2,....f k , that is:

[0033] Matrix form: X = AF + ε

[0034] Where: F is the common factor; A is the factor loading matrix, and a pk is the factor loading, which is the loading of the pth original variable on the kth factor; ε is the special factor, representing the part of the original variable that cannot be explained by the factors;

[0035] In the solution of the factor loading matrix A, the principal component analysis method is used. Through coordinate transformation, the original p variables are standardized and then linearly combined to convert them into another set of uncorrelated variables y, that is:

[0036]

[0037] Among them,

[0038] Where: u ip is the factor loading after coordinate transformation, which is the factor loading of the extracted ith factor on the pth original variable;

[0039] Step 2.3, calculate the correlation coefficient matrix of the variables and find the eigenvalues and corresponding eigenvectors of the correlation coefficient matrix;

[0040] Let λ1, λ2,..., λ p and u1, u2,..., u p be the eigenvalues and eigenvectors respectively. Factor analysis uses principal component analysis to obtain p eigenvalues and eigenvectors, and on this basis, calculate the factor loading matrix:

[0041]

[0042] When calculating the factor loading matrix, not all eigenvalues ​​are selected, only the first k eigenvalues ​​and eigenvectors are selected, and the following loading matrix containing k factors is obtained:

[0043]

[0044] Step 2.4, determine the number of factors;

[0045] Determine the number of factors based on the cumulative variance contribution rate of the factors:

[0046] First, the correlation matrix is ​​constructed and the characteristic equation is solved to calculate the eigenvalues;

[0047] For a k×k correlation matrix R, the characteristic equation is:

[0048] |R-λi|=0

[0049] Where i is the k×k identity matrix and λ is the eigenvalue;

[0050] After solving, k eigenvalues ​​can be obtained;

[0051] Then the variance contribution of each factor is calculated.

[0052]

[0053] In the formula, h i is the variance contribution rate of the i-th factor, λ i is the eigenvalue of the i-th factor;

[0054]

[0055] The number of characteristic roots with cumulative variance contribution greater than 85% is selected as the number of factors;

[0056] Where: C k is the cumulative variance contribution rate of the first k common factors;

[0057] Step 2.5, calculate the factor score coefficient;

[0058] The factor score coefficient is calculated using the following formula;

[0059]

[0060] Where: h i is the variance contribution rate of the i-th factor, α i is the factor score coefficient;

[0061] Step 2.6, calculate the weighted coefficient correlation;

[0062] The factor score coefficient calculated above is used as the weight to obtain the grey relational degree;

[0063] Right now:

[0064] Where r' 0i It is the weighted coefficient correlation, and is sorted according to the correlation of each indicator; the correlation range is between 0 and 1. The larger the coefficient, the stronger the correlation between the evaluation indicator and the reference sequence, that is, the better the effect.

[0065] The beneficial effects of the present invention are:

[0066] This method optimizes the press riveting quality based on the grey correlation analysis method and the factor analysis method. It can quickly select the most influential factors on the press riveting quality from many parameters, and when the push-out force and twist-out force are used to evaluate the press riveting quality in the press riveting quality evaluation indicators, it can solve the problem of information overlap of multiple evaluation indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] The present invention will be further described below in conjunction with the accompanying drawings;

[0068] Figure 1 is the mean main effect diagram of the present invention;

[0069] Figure 2 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0070] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0071] The accompanying drawings are a specific embodiment of the method for optimizing the press riveting quality based on the grey correlation and factor analysis method of the present invention. The embodiment includes using the Taguchi method to construct a press riveting quality test model with busbar design parameters and automatic press riveting machine control parameters as design optimization variables and push-out force and twist-out force as responses.

[0072] Multi-objective optimization methods based on grey relational analysis include:

[0073] Step S1.1, determine the evaluation indicators according to the evaluation purpose and collect evaluation data; according to the number of indicators to be evaluated, set n data sequences to be processed to form the following matrix:

[0074]

[0075] Where m is the number of data groups in one evaluation indicator,

[0076] X′ i =(x′ i (1), x′ i (2), …, x′ i (m)) T

[0077] A sequence of all data for a certain indicator;

[0078] i=1,2,…,n

[0079] is the number of different evaluation indicators. In this embodiment, the number of evaluation indicators is 2;

[0080] Step S1.2, determining the reference sequence;

[0081] The optimal value of each evaluation index constitutes a reference sequence. The optimal value, that is, the value with the largest positive impact, is the optimal value, and the value with the smallest negative impact is the optimal value. The reference sequence is recorded as X′0=(x′0(1), x′0(2), …, x′0(m)) T .

[0082] Step S1.3, in order to eliminate the influence of different dimensions of different evaluation indicators, the evaluation indicator data is normalized, and the evaluation indicator data is normalized using the mean method.

[0083]

[0084] Where: x i (k) is the normalized index data;

[0085] x′ i (k) is the kth data of the ith evaluation indicator.

[0086] Step S1.4, calculate the evaluation index sequence x for each evaluated object i The absolute difference between the corresponding elements of (k) and the reference sequence x0(k), that is:

[0087] |x0(k)-x i (k)|

[0088] Where: x i (k) is the kth data of the i-th evaluation index;

[0089] x0(k) is the kth data of the reference sequence.

[0090] Step S1.5, determine the minimum difference between the two levels The maximum difference between the two levels

[0091] Step S1.6, calculating the correlation coefficient; using the following formula, respectively calculate the correlation coefficient of the comparison sequence of each evaluation index and the corresponding element of the reference sequence;

[0092]

[0093] Where ρ is the resolution factor, which is 0.5.

[0094] Step S1.7, calculating the correlation degree;

[0095] The correlation coefficient between the evaluation index sequence of each evaluation object and the corresponding element of the reference sequence is calculated to reflect the relationship between each evaluation object and the reference sequence.

[0096] The steps of factor analysis include:

[0097] Step 2.1, determine the prerequisites for factor analysis; perform Bartlett's sphericity test and KMO test on the original variable data, or calculate the correlation coefficients between the original variables.

[0098] In step 2.1, the original variable data, namely the push-out force and the twist-out force, are subjected to Bartlett's sphericity test. If the p-value is less than 0.05, it indicates that there is a certain correlation between the indicators, indicating that the original variables are suitable for factor analysis. The KMO test is performed on the original indicator data. Generally, the KMO value is between 1 and 0.5. The closer the KMO value is to 1, the more suitable the original variables are for factor analysis.

[0099] In step 2.1,

[0100] Calculate the correlation coefficient between the original variables. The purpose of obtaining the correlation coefficient is to check the correlation between the two variables, which can reflect whether the two evaluation indicators are suitable for factor analysis.

[0101]

[0102] Where: r is the correlation coefficient,

[0103] x i is the i-th data in the push force,

[0104] is the average push-out force,

[0105] yi is the i-th data in the torque output force,

[0106] is the average torque output,

[0107] When 0<|r|≤0.3, the two evaluation indicators are weakly correlated;

[0108] When 0.3 < |r| ≤ 0.5, there is a low correlation between the two evaluation indicators;

[0109] When 0.5 < |r| ≤ 0.8, there is a significant correlation between the two evaluation indicators;

[0110] When 0.8 < |r| ≤ 1, there is a high correlation between the two evaluation indicators.

[0111] Step 2.2, process the sample indicators;

[0112] Suppose there are n samples and p original variables, which are represented by x1, x2,....x p respectively, and the mean of each variable is 0 and the standard deviation is 1. Now each original variable is represented by a linear combination of k (k < p) factors f1, f2,....f k , that is:

[0113] Matrix form: X = AF + ε

[0114] In the formula: F is the common factor; A is the factor loading matrix, and a pk is the factor loading, which is the loading of the i-th original variable on the k-th factor; ε is the special factor, representing the part of the original variable that cannot be explained by the factor;

[0115] In the solution of the factor loading matrix A, the principal component analysis method is used. Through coordinate transformation, the original p variables are standardized and then linearly combined to convert them into another set of uncorrelated variables y, that is:

[0116]

[0117] Among them,

[0118] In the formula: u ip is the factor loading after coordinate transformation, which is the factor loading of the i-th factor extracted on the p-th original variable.

[0119] Step 2.3, calculate the correlation coefficient matrix of the variables and find the eigenvalues and corresponding eigenvectors of the correlation coefficient matrix;

[0120] Let λ1, λ2,..., λ p and u1, u2,..., u p be the eigenvalues and eigenvectors respectively. Factor analysis uses principal component analysis to obtain p eigenvalues and eigenvectors. On this basis, the factor loading matrix is calculated:

[0121]

[0122] When calculating the factor loading matrix, not all eigenvalues ​​are selected, only the first k eigenvalues ​​and eigenvectors are selected, and the following loading matrix containing k factors is obtained:

[0123]

[0124] Step 2.4, determine the number of factors;

[0125] Determine the number of factors based on the cumulative variance contribution rate of the factors:

[0126] First, the correlation matrix is ​​constructed and the characteristic equation is solved to calculate the eigenvalues.

[0127] For a k×k correlation matrix R, the characteristic equation is:

[0128] |R-λi|=0

[0129] Where i is the k×k identity matrix and λ is the eigenvalue;

[0130] After solving, k eigenvalues ​​λ can be obtained.

[0131] The variance contribution of each factor was then calculated;

[0132]

[0133] In the formula, h i is the variance contribution rate of the i-th factor, λ i is the eigenvalue of the i-th factor.

[0134]

[0135] Usually, the number of characteristic roots with cumulative variance contribution greater than 85% is selected as the number of factors;

[0136] Where: C k is the cumulative variance contribution rate of the first k common factors.

[0137] Step 2.5, calculate the factor score coefficient;

[0138] The factor score coefficient is calculated using the following formula;

[0139]

[0140] Where: h i is the variance contribution rate of the i-th factor, α i is the factor score coefficient.

[0141] Step 2.6, calculate the weighted coefficient correlation;

[0142] The factor score coefficient calculated above is used as the weight to obtain the grey relational degree;

[0143] Right now:

[0144] Where r' 0i is the weighted coefficient correlation. And according to the correlation of each indicator, it is sorted; the correlation range is between 0 and 1. The larger the coefficient, the stronger the correlation between the evaluation indicator and the reference sequence, that is, the better the effect.

[0145] Specific embodiment one.

[0146] The Taguchi method is used to construct a riveting quality test model with busbar design parameters and automatic riveting machine control parameters as design optimization variables, and push-out force and twist-out force as responses, totaling 12 groups, as shown in the following table.

[0147] Table 1 Taguchi orthogonal design table and experimental results

[0148]

[0149] Multi-objective optimization method based on grey relational analysis.

[0150] The test results of the push-out force and the twist-out force are sorted by columns to form an original matrix Q, which has 12 rows and 2 columns.

[0151] Normalize the original matrix to obtain the normalized matrix S.

[0152]

[0153] Because the expected value of the push-out force and the twist-out force after normalization is 1. Therefore, the reference sequence is set to X′0=(1, 1), and the normalized data matrix S is used as the comparison sequence. The deviation data sequence is calculated based on the gray correlation coefficient and the maximum difference Δ is obtained. max Minimum difference Δ min .

[0154]

[0155] Then substitute the difference into the following formula:

[0156]

[0157] The resolution coefficient ρ is taken as 0.5, and the matrix R composed of the grey relational coefficients is obtained.

[0158]

[0159] The factor analysis method is used to calculate the objective weight of each output to achieve the purpose of dimensionality reduction of multiple responses. Because the two indicators in this paper have different meanings, in order to make the indicators more concentrated, the data are standardized. The original data are subjected to Bartlett's sphericity test and KMO test. The KMO value is 0.5, which is less than 0.6, and is not suitable for factor analysis, but the P value of Bartlett's sphericity test is 0.003, and the significance level of Bartlett's sphericity test is less than 0.05. The null hypothesis of the sphericity test is rejected. Through the test, it is shown that there is a correlation between the predictor variables, and principal component analysis can be used. As shown in Table 2.

[0160] The correlation coefficient matrix is ​​obtained from the following formula, and the results are shown in Table 3. There is a good correlation between the two indicators. When the correlation coefficient is above 0.7, factor analysis can be used.

[0161]

[0162] Table 2 KMO and Bartlett's sphericity test

[0163]

[0164] Table 3 Correlation matrix

[0165]

[0166] Determine the number of factors based on the cumulative variance contribution rate of the factors: usually the number of characteristic roots with a cumulative variance contribution rate greater than 85% is selected as the number of factors.

[0167] Obtain the cumulative variance contribution rate.

[0168] Table 4 Total variance explained

[0169]

[0170] Extraction method: principal component analysis.

[0171] It can be seen from Table 4 that the cumulative variance contribution rate of the first factor can reach 88.716%.

[0172] To calculate the factor score, the factor score coefficient is calculated using the following formula.

[0173]

[0174] Where: λ1 is the variance contribution rate, α i is the factor score coefficient.

[0175] The results are shown in Table 5 below.

[0176] Table 5 Factor score coefficient matrix

[0177]

[0178] Calculate the grey weighted correlation

[0179] The factor score coefficient calculated above is used as the weight to obtain the grey relational degree.

[0180] Right now:

[0181] Where r' oi , and sort them according to the degree of correlation.

[0182] The indicators are sorted according to their relevance.

[0183] Table 6 Correlation coefficient results and rankings of each sample

[0184]

[0185] The correlation range is between 0 and 1. The larger the coefficient, the stronger the correlation with the reference sequence, that is, the better the effect. Among them, the index of sample 10 is the closest, and it is inferred that the pressure riveting effect of this sample is the best, followed by sample 5 and sample 8; among them, sample 6 has the smallest correlation and its pressure riveting effect is the worst.

[0186] According to the obtained grey correlation degree, with aperture, maximum pressure value and pressing speed as three factors and grey correlation degree as response, an orthogonal experimental table was constructed to carry out range analysis.

[0187] Table 7 Orthogonal experiment table

[0188]

[0189] Calculated by the following formula

[0190]

[0191] Where: s is the mean;

[0192] y is the number of experiments with the same factor and parameters;

[0193] x is the sum of the grey relational degrees under the same factor and the same parameter influence;

[0194] The mean response table 8 and mean main effect plot are obtained.

[0195] Table 8 Mean response table

[0196]

[0197] Combined with the mean response table in Table 8, it can be concluded that the hole size has the greatest impact on the riveting quality, followed by the maximum pressure value, and the pressing speed is the smallest.

[0198] As shown in the mean main effect diagram in the attached figure, the optimal aperture is 8.78, the optimal maximum pressure value is 20.0, and the optimal downward pressure speed is 1.

[0199] Based on the above analysis, we can know that the optimal factor is the aperture, and the best combination is "aperture 8.78, maximum pressure value 20.0, and downward pressure speed 1".

Claims

1. A method for optimizing the quality of press riveting based on grey correlation and factor analysis, characterized in that: This includes adopting the Taguchi method to construct a riveting quality test model with busbar design parameters and automatic riveting machine control parameters as design optimization variables and push-out force and twist-out force as responses; Multi-objective optimization methods based on grey relational analysis include: Step S1.1, determine the evaluation indicators according to the evaluation purpose and collect evaluation data; according to the number of indicators to be evaluated, set n data sequences to be processed to form the following matrix: Where m is the number of evaluation indicators, X′ i =(x′ i (1),x′ i (2),…,x′ i (m)) T A sequence of all data for a certain indicator; i=1,2,…,n is the number of different evaluation indicators; Step S1.2, determining the reference sequence; The optimal value of each evaluation index constitutes a reference sequence. The optimal value, that is, the value with the largest positive impact, is the optimal value, and the value with the smallest negative impact is the optimal value. The reference sequence is recorded as X′0=(x′0(1), x′0(2),..., x′0(m)) T ; Step S1.3, in order to eliminate the influence of different dimensions of different evaluation indicators, the evaluation indicator data is normalized, and the evaluation indicator data is normalized using the mean method. Where: x i (k) is the normalized index data; x′ i 〔k〕is the kth data of the ith evaluation index; Step S1.4, calculate the evaluation index sequence x for each evaluated object i The absolute difference between the corresponding elements of (k) and the reference sequence x0(k), that is: |x0(k)-x i (k)| Where: x i (k) is the kth data of the i-th evaluation index; x0(k) is the kth data of the reference sequence; Step S1.5, determine the minimum difference between the two levels The maximum difference between the two levels Step S1.6, calculating the correlation coefficient; using the following formula, respectively calculate the correlation coefficient of the comparison sequence of each evaluation index and the corresponding element of the reference sequence; Where ρ is the resolution factor, which is 0.5; Step S1.7, calculating the correlation degree; For each evaluation object, the correlation coefficient between its evaluation index sequence and the corresponding element of the reference sequence is calculated to reflect the relationship between each evaluation object and the reference sequence; The steps of factor analysis include: Step 2.1, determine the prerequisites for factor analysis; perform Bartlett's sphericity test and KMO test on the original variable data, or calculate the correlation coefficient between the original variables; Step 2.2, process sample indicators; Suppose there are n existing samples and p original variables, which are represented by x1, x2,.... x p respectively, and the mean of each variable is 0 and the standard deviation is 1. Now each original variable is represented by a linear combination of k (k < p) factors f1, f2,.... f k That is: Matrix form: X = AF + ε Where: F is the common factor; A is the factor loading matrix, a pk is the factor loading, which is the loading of the pth original variable on the kth factor; ε is a special factor, which indicates the part of the original variable that cannot be explained by the factor; The principal component analysis method is used to solve the factor loading matrix A. Through coordinate transformation, the original p variables are standardized and linearly combined to transform them into another set of unrelated variables y, namely: in, Where: u ip is the factor loading after coordinate transformation, which is the factor loading of the extracted i-th factor on the p-th original variable; Step 2.3, calculate the correlation coefficient matrix of the variables and find the characteristic roots and corresponding eigenvectors of the correlation coefficient matrix; Let λ1,λ2,…,λ p and u1, u2,…,u p are characteristic roots and eigenvectors respectively. Factor analysis uses principal component analysis to obtain p characteristic roots and eigenvectors, and on this basis calculates the factor loading matrix: When calculating the factor loading matrix, not all eigenvalues ​​are selected, only the first k eigenvalues ​​and eigenvectors are selected, and the following loading matrix containing k factors is obtained: Step 2.4, determine the number of factors; Determine the number of factors based on the cumulative variance contribution rate of the factors: First, construct the correlation matrix and solve the characteristic equation to calculate the eigenvalues. For a k×k correlation matrix R, the characteristic equation is: |R-λi|=0 Where i is the k×k identity matrix and λ is the eigenvalue; After solving, k eigenvalues ​​λ can be obtained; The variance contribution of each factor was then calculated; In the formula, h i is the variance contribution rate of the i-th factor, λ i is the eigenvalue of the i-th factor; The number of characteristic roots with cumulative variance contribution greater than 85% is selected as the number of factors; Where: C k is the cumulative variance contribution rate of the first k common factors; Step 2.5, calculate the factor score coefficient; The factor score coefficient is calculated using the following formula; Where: h i is the variance contribution rate of the i-th factor, α i is the factor score coefficient; Step 2.6, calculate the weighted coefficient correlation; The factor score coefficient calculated above is used as the weight to obtain the grey relational degree; Right now: Where r0' i is the weighted coefficient correlation; and the indicators are ranked according to their correlation; the correlation range is between 0 and 1, and the larger the coefficient, the stronger the correlation between the evaluation indicator and the reference sequence, that is, the better the effect.

2. The method for optimizing the pressure riveting quality based on grey correlation and factor analysis according to claim 1 is characterized in that: In step 2.1, the original variable data, namely the push-out force and the twist-out force, are subjected to Bartlett's sphericity test. If the p-value is less than 0.05, it indicates that there is a certain correlation between the indicators, indicating that the original variables are suitable for factor analysis. The KMO test is performed on the original indicator data. Generally, the KMO value is between 1 and 0.

5. The closer the KMO value is to 1, the more suitable the original variables are for factor analysis.

3. The method for optimizing the riveting quality based on grey correlation and factor analysis according to claim 1 is characterized in that: In step 2.1, Calculate the correlation coefficient between the original variables; Where: r is the correlation coefficient, x i is the i-th data in the push force, is the average push-out force, yi is the i-th data in the torque output force, is the average torque output, When 0<|r|≤0.3, the two evaluation indicators are weakly correlated; When 0.3<|r|≤0.5, the two evaluation indicators are poorly correlated; When 0.5<|r|≤0.8, the two evaluation indicators are significantly correlated; When 0.8<|r|≤1, the two evaluation indicators are highly correlated.

Images