Molten Iron Carbon Regression Analysis Method Based on Component Correlation

Abstract

The present invention relates to a method for regression analysis of molten iron carbon based on component correlation, belonging to the field of molten iron carbon prediction, and includes S1: randomly extracting a plurality of molten iron production samples and taking tempered drill samples to prepare the samples; S2: using an infrared carbon-sulfur analyzer to analyze the carbon content in the molten iron, and analyzing the contents of other components in the molten iron by direct reading spectroscopy to obtain the analysis data of the molten iron samples; S3: according to the idea of multiple linear regression analysis, performing regression processing on the data analyzed from the molten iron samples to obtain the correlation between each component in the molten iron and the carbon content; S4: through variance analysis, calculating the significant differences between each component and the carbon content; S5: excluding the components that have no significant effect on the carbon content prediction and retaining the components that have a significant prediction effect on the carbon content; S6: calculating the regression coefficients for the components with a significant prediction effect to obtain a regression equation; S7: by giving the contents of other components, calculating the carbon content in the molten iron using the regression equation.

Description

Technical Field

[0001] The present invention belongs to the field of hot metal carbon prediction, and relates to a hot metal carbon regression analysis method based on component correlation. Background Art

[0002] Hot metal is the main raw material for converter steelmaking, generally accounting for 70%-100% of the charged amount. The chemical composition of hot metal directly affects metallurgical technical indexes such as the reduction of blowing loss and the consumption of steel materials, as well as economic benefits. The carbon oxidation reaction is an extremely important reaction in the steelmaking process, and the carbon content in hot metal directly affects the blowing loss.

[0003] With the shortening of the converter smelting cycle, higher requirements are placed on the detection time of hot metal components. At present, when measuring the carbon content in hot metal, most use an infrared carbon-sulfur analyzer. Although the measurement results are accurate, sample preparation is very difficult; another method is to use direct reading spectroscopy analysis, which will cause relatively large errors in the results. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to use SSPS software to analyze the correlation between the carbon content of hot metal and other components, so as to establish a prediction model for hot metal carbon.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A hot metal carbon regression analysis method based on component correlation, comprising the following steps:

[0007] S1: Randomly select multiple hot metal production samples, and take tempered drill samples to prepare samples;

[0008] S2: Use an infrared carbon-sulfur analyzer to analyze the carbon content in hot metal, and analyze the contents of other components in hot metal by direct reading spectroscopy to obtain hot metal sample analysis data;

[0009] S3: According to the idea of multiple linear regression analysis, perform regression processing on the data analyzed from hot metal samples to obtain the correlation between each component in hot metal and the carbon content;

[0010] S4: Through variance analysis, calculate the significant difference Sig (significant difference) between each component and the carbon content;

[0011] S5: Exclude the components that have no significant effect on carbon content prediction, and retain the components that have a significant prediction effect on carbon content;

[0012] S6: Calculate the regression coefficients for the components with significant prediction effects to obtain a regression equation;

[0013] S7: By giving the contents of other components, use the regression equation to calculate the carbon content in hot metal.

[0014] Furthermore, the regression equation is as follows:

[0015] C% = 4.056 - 4.653 * S% + 1.461 * Mn% + 2.537 * Ti% - 0.368 * Si%

[0016] Where C% represents the percentage content of carbon element in molten steel, S% represents the percentage content of sulfur element in molten steel, Mn% represents the percentage content of manganese element in molten steel, Ti% represents the percentage content of titanium element in molten steel, and Si% represents the percentage content of silicon element in molten steel.

[0017] Furthermore, in step S3, the data analyzed from the hot metal sample is subjected to regression processing to obtain the correlation between each component in the hot metal and the carbon content. The linear regression analysis includes the following parameters:

[0018] The Pearson Correlation Coefficient is used to measure whether two data sets are on a straight line and to measure the linear relationship between interval variables. The formula is as follows:

[0019]

[0020] Where X and Y are two continuous variables, Xi is the i-th value in X, i ranges from 1 to N, and Yi is the i-th value in Y. is the average of X. is the average of Y;

[0021] The multiple correlation coefficient R is an index indicating the degree of linear correlation between the independent variable and other dependent variables. The value range of the multiple correlation coefficient is between 0 and 1. The closer its value is to 1, the stronger the linear relationship; conversely, the weaker the linear relationship.

[0022] R 2 The coefficient of determination, used to determine the goodness of fit of the linear regression line, R 2 The coefficient of determination is equal to the ratio of the regression sum of squares to the total sum of squares, reflecting the percentage of the variability of the dependent variable that can be explained by the regression model;

[0023]

[0024] RSS represents the residual sum of squares, TSS represents the total sum of squares, n - k - 1 is the degree of freedom of the residual sum of squares, and n - 1 is the degree of freedom of the total sum of squares, where n and k are both constant coefficients;

[0025] The Durbin-Watson test is used to test the independence of the error terms in a regression model. Its value ranges from 0 to 4. When the residuals and independent variables are independent of each other, DW≈2; when the residuals of adjacent points are positively correlated, DW<2; when the residuals of adjacent points are negatively correlated, DW>2.

[0026] Furthermore, step S4 specifically includes:

[0027] Calculate the standardized coefficient Beta of each variable. First, perform a Z-transformation on each independent variable, and then perform regression. The Z-transformation is: (data value - data average value) divided by the standard deviation;

[0028] Calculate the group variance value F of each variable; calculate the test result of F, that is, the significant difference Sig (significant difference);

[0029] Calculate the partial correlation coefficient of each variable. The partial correlation coefficient represents the degree of correlation between the independent variable and the dependent variable after excluding the influence of other variables.

[0030] Furthermore, in step S5, specifically: The components with Sig less than 0.05 are the components that have a significant predictive effect on the carbon content, including sulfur content, manganese content, titanium content, and silicon content; the components with Sig greater than 0.05 are the components that do not have a significant predictive effect on the carbon content, including phosphorus content, vanadium content, and arsenic content.

[0031] The beneficial effects of the present invention are as follows: The present invention provides a molten iron carbon regression analysis model, and directly obtains the carbon content in the molten iron through this model. Compared with the prior art, the analysis efficiency is improved and the calculation accuracy is improved.

[0032] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent description, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail preferably with reference to the accompanying drawings, where:

[0034] Figure 1 is the regression standardized residual plot of the present invention;

[0035] Figure 2 is the comparison chart of the calculation of the molten iron carbon regression analysis method based on SPSS and the actual detection value of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The following describes the implementation manners of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0037] Among them, the drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and should not be construed as a limitation on the present invention; in order to better illustrate the embodiments of the present invention, some components in the drawings will be omitted, enlarged or reduced, which does not represent the size of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0038] In the drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc. indicating the orientation or positional relationship, they are based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the drawings are only for illustrative purposes and should not be construed as a limitation on the present invention. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0039] The present invention provides a method for regression analysis of the carbon content in hot metal based on component correlation. First, 35 hot metal production samples are randomly selected, and samples are taken by tempering and drilling. The C content is analyzed using an infrared carbon-sulfur analyzer, and other components are analyzed by direct reading spectroscopy. The analyzed data is shown in Table 1

[0040] Table 1

[0041] Serial number C％ Si% Mn% P％ V％ Ti% As% S％ Serial number C％ Si% Mn% P％ V％ Ti% As% S％ 1 4.28 0.28 0.22 0.115 0.098 0.099 0.001 0.051 19 4.57 0.56 0.23 0.11 0.102 0.167 0.001 0.028 2 4.61 0.39 0.29 0.124 0.117 0.168 0.004 0.029 20 4.24 0.61 0.23 0.11 0.098 0.127 0.001 0.032 3 4.24 0.34 0.26 0.122 0.091 0.082 0.001 0.042 21 4.36 0.51 0.24 0.116 0.112 0.152 0.003 0.036 4 4.58 0.57 0.25 0.115 0.097 0.209 0.001 0.024 22 4.14 0.26 0.21 0.108 0.073 0.06 0.005 0.053 5 4.41 0.41 0.24 0.118 0.096 0.14 0.001 0.036 23 4.36 0.41 0.23 0.116 0.086 0.102 0.003 0.032 6 4.73 0.46 0.26 0.119 0.103 0.215 0.001 0.028 24 4.58 0.62 0.37 0.138 0.03 0.12 0.001 0.032 7 4.38 0.38 0.24 0.113 0.107 0.129 0.001 0.031 25 4.28 0.45 0.32 0.13 0.027 0.096 0.001 0.05 8 4.74 0.43 0.25 0.118 0.102 0.205 0.005 0.021 26 4.52 0.59 0.39 0.14 0.032 0.115 0.001 0.032 9 4.48 0.32 0.29 0.128 0.101 0.128 0.002 0.02 27 4.66 0.5 0.39 0.134 0.03 0.107 0.001 0.033 10 4.54 0.52 0.27 0.124 0.093 0.135 0.001 0.031 28 4.58 0.67 0.4 0.142 0.032 0.132 0.001 0.034 11 4.61 0.5 0.27 0.118 0.087 0.179 0.001 0.029 29 4.26 0.38 0.34 0.138 0.029 0.077 0.001 0.068 12 4.35 0.43 0.25 0.118 0.097 0.121 0.002 0.032 30 4.61 0.56 0.41 0.131 0.028 0.136 0.003 0.026 13 4.46 0.43 0.22 0.112 0.101 0.122 0.001 0.034 31 4.66 0.71 0.42 0.139 0.028 0.152 0.001 0.02 14 4.38 0.58 0.24 0.111 0.108 0.168 0.001 0.027 32 4.69 0.79 0.43 0.137 0.028 0.16 0.001 0.019 15 4.65 0.51 0.25 0.112 0.092 0.15 0.002 0.023 33 4.6 0.7 0.42 0.133 0.028 0.15 0.001 0.022 16 4.55 0.43 0.25 0.115 0.087 0.14 0.001 0.026 34 4.71 0.69 0.39 0.126 0.026 0.14 0.001 0.026 17 4.41 0.35 0.26 0.123 0.097 0.096 0.001 0.038 35 4.57 0.4 0.36 0.123 0.024 0.092 0.002 0.048 18 4.35 0.42 0.25 0.12 0.094 0.073 0.001 0.045

[0042] According to the idea of multiple linear regression analysis, data regression processing is performed using SPSS software, and the analysis results are shown in Table 2 and Table 3, where the dependent variable is C%; the predictive variables (constants) are S%, Mn%, Ti%, Si%.

[0043] Table 2

[0044]

[0045] Table 3

[0046]

[0047] Table 3 shows that the model correlation reaches 88.5%, and the effect of fitting the data is better. In addition, the DW test value of the model is 1.897, which is close to 2, indicating no autocorrelation among the independent variables.

[0048] Table 4

[0049]

[0050] In Table 4, "Regression" refers to the regression method, "Residual" refers to the difference between the measured value and the predicted value, "df" refers to the degrees of freedom, which is the number of variables that can take values freely, "Mean Square" is the variance divided by the degrees of freedom, "F" is the variance F-test statistic, used to test whether the regression equation is meaningful, "Sig" refers to the significant difference. When the value corresponding to Sig is less than 0.05, it indicates that the established regression equation has statistical significance, that is, there is a linear relationship between the independent variable and the dependent variable.

[0051] "Sum of Squares due to Regression" represents the part that can be explained by the independent variables included in the regression model in the variation of the response variable. "Sum of Squares of Residuals" represents the part of the variation of the response variable that is not explained by the variables included in the regression model. These two values are related to the sample size and the number of independent variables in the model. The larger the sample size, the greater the corresponding variation.

[0052] Table 5

[0053]

[0054] In Table 5, "Beta" is the standardized regression coefficient, which is used to compare the absolute effects or contributions of each coefficient. First, each independent variable is subjected to a Z transformation, and then regression is performed. The Z transformation is: data value - data mean, and then divided by the standard deviation. The Z transformation can standardize the scales and dimensions of each independent variable, and thus the effect sizes of each independent variable in the regression equation can be compared according to the magnitudes of the standardized regression coefficients.

[0055] "t" represents the t-test, which is used to test the significance of each parameter in the regression equation.

[0056] Sig (significant difference) in Tables 4 and 5 is the significant difference. Among them, the sig in Table 4 is 0, which is less than 0.05, indicating that the four independent variables S%, Mn%, Ti%, and Si% have a significant predictive effect on the dependent variable. While for the three variables P%, V%, and As% in Table 5, the sig values are all greater than 0.05, which have no predictive effect and need to be excluded.

[0057] Table 6

[0058]

[0059] From Table 6, the regression equation C% = 4.056 - 4.653*S% + 1.461*Mn% + 2.537*Ti% - 0.368*Si% can be obtained. It can be seen from the t-test that the regression coefficients are significant.

[0060] Table 7

[0061]

[0062]

[0063] From Table 7 and the regression standardized residuals Figure 1 it can be obtained that the standard deviation of the predicted value in the regression analysis is 0.1433, the standard deviation of the residual (the difference between the actual observed value and the fitted value) is 0.075, and the residuals follow a normal distribution.

[0064] The maximum error between the predicted value and the actual value obtained by this model is 0.143% (see Table 8 and Figure 2 ), which is less than the analysis error. By directly reading the spectrum analysis to quickly detect other components of the hot metal, the model is used to automatically calculate the carbon content of the hot metal, quickly guiding the production of the converter.

[0065] Table 8

[0066] Number of cases C Predicted value Residual Number of cases C Predicted value Residual 1 4.28 4.2882 -.00821 19 4.57 4.4792 .09083 2 4.61 4.6273 -.01732 20 4.24 4.3407 -.10067 3 4.24 4.3233 -.08328 21 4.36 4.4369 -.07692 4 4.58 4.6298 -.04985 22 4.14 4.1727 -.03274 5 4.41 4.4433 -.03332 23 4.36 4.3509 .00907 6 4.73 4.6816 .04841 24 4.58 4.5237 .05628 7 4.38 4.4497 -.06974 25 4.28 4.3687 -.08869 8 4.74 4.6852 .05476 26 4.52 4.5513 -.03130 9 4.48 4.5935 -.11351 27 4.66 4.5595 .10049 10 4.54 4.4572 .08280 28 4.58 4.5703 .00974 11 4.61 4.5855 .02451 29 4.26 4.2918 -.03175 12 4.35 4.4210 -.07098 30 4.61 4.6728 -.06276 13 4.46 4.3704 .08961 31 4.66 4.7006 -.04060 14 4.38 4.4936 -.11360 32 4.69 4.7107 -.02068 15 4.65 4.5069 .14306 33 4.60 4.6899 -.08991 16 4.55 4.4971 .05291 34 4.71 4.6058 .10420 17 4.41 4.3737 .03628 35 4.57 4.4447 .12530 18 4.35 4.2424 .10758

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the present technical solution, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for regression analysis of molten iron carbon based on component correlation, characterized in that: It includes the following steps: S1: Randomly select multiple molten iron production samples, and take tempered drill samples to prepare the samples; S2: Use an infrared carbon-sulfur analyzer to analyze the carbon content in the molten iron, and analyze the contents of other components in the molten iron by direct-reading spectroscopy to obtain the analysis data of the molten iron samples; S3: According to the idea of multiple linear regression analysis, perform regression processing on the data analyzed from the molten iron samples to obtain the correlation between each component in the molten iron and the carbon content; S4: Through analysis of variance, calculate the significance difference Sig between each component and the carbon content; S5: Exclude the components that have no significant effect on carbon content prediction, and retain the components that have a significant predictive effect on carbon content; S6: Calculate the regression coefficients for the components with significant predictive effects to obtain the regression equation; S7: By giving the contents of other components, use the regression equation to calculate the carbon content in the molten iron; The regression equation is as follows: C% = 4.056 - 4.653*S% + 1.461*Mn% + 2.537*Ti% - 0.368*Si% Where C% represents the percentage content of carbon element in the molten steel, S% represents the percentage content of sulfur element in the molten steel, Mn% represents the percentage content of manganese element in the molten steel, Ti% represents the percentage content of titanium element in the molten steel, and Si% represents the percentage content of silicon element in the molten steel; In step S3, when performing regression processing on the data analyzed from the molten iron samples to obtain the correlation between each component in the molten iron and the carbon content, the following parameters are included in the linear regression analysis: The Pearson correlation coefficient is used to measure whether two data sets are on the same line and to measure the linear relationship between interval variables. The formula is as follows: Where X and Y are two continuous variables, Xi is the i-th value in X, where i ranges from 1 to N, and Yi is the i-th value in Y. is the average value of X. is the average value of Y. The multiple correlation coefficient R is an index representing the degree of linear correlation between the independent variable and other dependent variables. The value range of the multiple correlation coefficient is between 0 and 1. The closer its value is to 1, the stronger its linear relationship; conversely, the weaker the linear relationship; R 2 Coefficient of determination, used to determine the goodness of fit of the linear regression line, R 2 The coefficient of determination is equal to the ratio of the regression sum of squares to the total sum of squares, reflecting the percentage of the variability of the dependent variable that can be explained by the regression model; RSS represents the residual sum of squares, TSS represents the total sum of squares, n - k - 1 is the degree of freedom of the residual sum of squares, and n - 1 is the degree of freedom of the total sum of squares, where n and k are both constant coefficients; The Durbin-Watson test is used to test the independence of the error terms of the regression model. Its value range is 0 to 4. When the residuals and the independent variables are independent of each other, DW ≈ 2; when the residuals of adjacent two points are positively correlated, DW < 2; when the residuals of adjacent two points are negatively correlated, DW > 2; Specifically included in step S4: Calculate the standardized coefficient Beta of each variable. First, perform Z transformation on each independent variable, and then perform regression. The Z transformation is: data value - data average value, and then divide by the standard deviation; Calculate the group variance value F of each variable; calculate the test result of F, that is, the significance difference Sig; Calculate the partial correlation coefficient of each variable. The partial correlation coefficient represents the degree of correlation between the independent variable and the dependent variable after excluding the influence of other variables; In step S5, specifically, the components with Sig less than 0.05 are the components that have a significant predictive effect on the carbon content, including sulfur content, manganese content, titanium content, and silicon content; the components with Sig greater than 0.05 are the components that do not have a significant predictive effect on the carbon content, including phosphorus content, vanadium content, and arsenic content.

Images