A method for correcting human error / bias in measuring the thickness of decarburized layer of spring steel

Abstract

The invention discloses a method for correcting human error / deviation in the measurement of spring steel decarburization layer thickness, which relates to the technical field of steel production. After adopting a large amount of data fitting correction, the observation points or test points of the measurement error or deviation caused by human factors are continuously calibrated, and the boundary of the partial decarburization layer is gradually defined. When the judgment limit is accurate, the calibration of the experimental data is realized, and a multivariate linear regression equation that conforms to the experimental results is obtained. The method has guiding significance for controlling the thickness of the partial decarburization layer of the product and ensuring the reliable delivery of the rolled material with excellent performance.

Description

Technical Field

[0001] The invention relates to the technical field of steel production, in particular to a method for correcting human error / deviation in measuring the thickness of a decarburized layer of spring steel. Background Art

[0002] Before hot rolling or hot forging, spring steel billets generally need to be continuously heated to a high temperature in a non-vacuum or protective atmosphere heating furnace and kept warm for a certain period of time (depending on the type of steel, the high temperature is different, but generally it needs to be heated to about 1200℃). In the actual heating process, since the heating furnace is in a non-vacuum or protective atmosphere state, oxidation and decarburization will occur in the billet during the heating to high temperature process and the subsequent insulation process. The iron oxide scale formed by oxidation can be removed by subsequent high-pressure water descaling, and the decarburized layer formed by decarburization will remain on the surface of the billet. The decarburized layer retained on the surface of the billet will generally become thinner with the reduction of the rolled material size during the subsequent hot rolling or hot forging process, and finally a thin decarburized layer will be formed on the surface of the finished rolled material. The presence of a decarburized layer on the surface of spring steel rolled material will have an extremely harmful effect on its fatigue properties. Studies have shown that for spring steel bar products, in order to have high fatigue life and anti-elastic decay performance, the thickness of the decarburized layer on the surface of the spring steel rolled material is generally required to be less than 0.5% of the effective thickness of the rolled material. Therefore, accurate measurement of the decarburized layer is a prerequisite for ensuring reliable delivery of rolled products with excellent performance.

[0003] The measurement of decarburized layer is generally carried out by metallographic method. First, the sample containing decarburized layer is prepared manually, then the decarburized layer is observed under a metallographic microscope, and finally the thickness of the decarburized layer is measured offline manually by the metallographic photograph of the decarburized layer. The thickness of the decarburized layer can also be directly measured online by using the equipped software while observing the decarburized layer under a metallographic microscope. Regardless of which method is used to measure the thickness of the decarburized layer, the boundary between the decarburized layer and the non-decarburized layer needs to be finally determined by the measurer or observer. Therefore, the accurate definition of this boundary is the prerequisite for ensuring the accurate measurement of the thickness of the decarburized layer.

[0004] In most cases, there is a clear boundary between the decarburized layer and the non-decarburized layer, so the accurate definition of this boundary is generally less affected by human factors. However, sometimes the boundary between the decarburized layer and the non-decarburized layer is not clear, and the accurate definition of this boundary is greatly affected by human factors, resulting in human errors or even deviations in the measurement of the decarburized layer thickness.

[0005] Since the decarburized layer is formed during the continuous heating and insulation of the billet, existing studies have shown that the thickness of the decarburized layer is related to factors such as heating rate, heating temperature, insulation time and oxygen content (heating atmosphere). Since the formation of the decarburized layer is controlled by diffusion, there must be a certain intrinsic relationship between the thickness of the decarburized layer and the heating rate, heating temperature, insulation time and oxygen content. By establishing this intrinsic relationship, it is possible to calibrate the decarburized layer thickness measurement in cases where the boundary between the decarburized layer and the non-decarburized layer is greatly affected by human factors.

[0006] Patent CN105387824 B discloses a method for measuring the depth of decarburized layer in quenched steel. The invention uses a wavelength dispersive spectrometer to measure the carbon element characteristic X-ray Kα line on the cross-sectional metallographic sample, and the measurement point extends from the surface of the cross-sectional metallographic sample to the matrix. The distance between the edge of the sample corresponding to the transition node where the background signal and the signal intensity gradually increase is the thickness of the full decarburized layer, the distance between the edge corresponding to the transition node between the gradual increase in signal intensity and the gradual stabilization of signal intensity is the total decarburized layer depth, and the difference between the total decarburized layer and the full decarburized layer is the partial decarburized layer depth. It is suitable for measuring the depth of decarburized layer in quenched steel that cannot be measured by metallographic method, and has higher accuracy than the hardness method, chemical method and spectral analysis method for measuring carbon content. Although this scheme adopts some methods to reduce the influence of sample preparation pollution and pollution during the measurement process on the accuracy and measurement efficiency of the measured carbon element characteristic X-ray, this scheme lacks correction of experimental data and fails to minimize this error, so the accuracy is relatively poor.

[0007] Patent CN110648421 B invention provides a method for calculating the thickness of the decarburized layer on the surface of decarburized bearing steel, which belongs to the field of high-speed wire rolling technology. The method first calculates the theoretical equilibrium carbon concentration on the surface of the steel part under different decarburization atmosphere conditions; then calculates the carbon content in the decarburized layer and the diffusion coefficient of carbon in ferrite under different heating temperatures and heating times; finally, according to the decarburization model calculation formula and combined with Gaussian error function analysis, the thickness of the decarburized layer on the surface of the steel part is calculated. Through the furnace atmosphere parameters, temperature parameters and time parameters in the real-time database of the hot rolling production line, the online prediction of the thickness of the decarburized layer during the hot rolling process can be carried out in real time, and the hot rolling process parameters can be adjusted accordingly to achieve the purpose of reducing the thickness of the decarburized layer and improving the surface quality of the bearing steel. Although this scheme theoretically calculates the thickness of the decarburized layer, it does not take into account the influence of the oxide layer on the thickness of the decarburized layer, and the influence of the oxide layer on the thickness model of the decarburized layer is still unknown. Summary of the invention

[0008] The present invention aims at solving the above technical problems, overcomes the shortcomings of the prior art, and provides a method for correcting the human error / deviation in the measurement of the thickness of the decarburized layer of spring steel.

[0009] Firstly, the four-factor three-level orthogonal experimental design method and metallographic method were used to study the effects of heating temperature, heating rate, holding time and oxygen content on the thickness of the decarburized layer of the steel, and a preliminarily multivariate linear regression fitting relationship between the thickness of the decarburized layer of the steel and the heating temperature, heating rate, holding time and oxygen content was established.

[0010] Then, according to the goodness of fit Adj-R 2 The size of the observation points or test points that may be caused by human factors, the measurement error or deviation, and the decarburization layer thickness values ​​corresponding to these observation points or test points are adjusted according to the fitting relationship, and then the multivariate linear regression fitting is performed again until Adj-R 2 Reaching above 0.90 and the qualitative relationship between the decarburized layer thickness and the heating rate, heating temperature, holding time and oxygen content does not violate the basic diffusion law;

[0011] Finally, the thickness of the decarburized layer at the adjusted observation point or test point is compared with the actual decarburized layer, and the boundary between the decarburized layer and the non-decarburized layer is redefined to determine the accuracy of the redefined boundary. If it is accurate, the error or deviation caused by human factors can be corrected. If it is inaccurate, it means that the fitting regression relationship established above is inappropriate and needs to be recalibrated using other relationships.

[0012] The technical solution further defined in the present invention is:

[0013] The above-mentioned method for correcting the human error / bias in the measurement of the thickness of the decarburized layer of spring steel is specifically as follows:

[0014] (1) Using the four-factor three-level orthogonal experimental design method, an orthogonal table including heating temperature, heating rate, insulation time and oxygen content was designed, and the samples were tested;

[0015] (2) The samples after the experiment were processed by metallographic method, and the thickness of the partial decarburized layer of each sample was measured. Combined with the orthogonal table in step (1), a multivariate linear regression fitting relationship between the thickness of the decarburized layer of the steel and the heating temperature, heating rate, holding time and oxygen content was preliminarily constructed, and the multivariate linear regression fitting equation was obtained:

[0016] y=A0+A1*x1+A2*x2+A3*x3+A4*x4

[0017] Where: y is the predicted decarburized layer thickness, µm; A0, A1, A2, A3, A4 are the coefficients of the fitting equation; X1, X2, X3, X4 are heating temperature, heating rate, holding time and oxygen content respectively;

[0018] (3) Check the corrected goodness of fit Adj-R 2Whether it reaches 0.9 and the qualitative relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law,

[0019] If Adj-R 2 If the decarburization thickness reaches 0.9 and the qualitative relationship between the decarburization layer thickness and the four factors does not violate the basic diffusion law, it means that the above multivariate linear fitting equation can be used to measure the partial decarburization thickness accurately; if the Adj-R 2 It did not reach 0.9. The thickness of the decarburized layer in all parts was adjusted several times to observe Adj-R. 2 The size of the observation points or test points that may be caused by human factors is determined. According to the metallographic diagram of the sample, the thickness of the decarburized layer corresponding to these observation points or test points is remeasured and brought into the orthogonal table for fitting to obtain the multivariate linear regression equation after quadratic fitting.

[0020] (4) Check Adj-R after quadratic fitting 2 Whether it reaches 0.9 and whether the qualitative relationship between the decarburized layer thickness and the four factors violates the basic diffusion law, repeat step (3) many times until Adj-R 2 It reaches above 0.90 and the qualitative relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law;

[0021] (5) Finally, the adjusted decarburized layer thickness at the observation point or test point is compared with the actual decarburized layer, and the boundary between the decarburized layer and the non-decarburized layer is redefined to determine the accuracy of the redefined boundary. If it is accurate, the error or deviation caused by human factors can be corrected. If it is inaccurate, it means that the fitting regression relationship established above is inappropriate and needs to be recalibrated using other relationships.

[0022] The beneficial effects of the present invention are:

[0023] (1) The present invention can predict and control the thickness of the decarburized layer of the rolled material through the linear regression fitting relationship equation between the thickness of the decarburized layer of silicon-manganese spring steel and four factors, provide guidance for process optimization, and reduce the loss of rework caused by excessive decarburized layer;

[0024] (2) The present invention eliminates the problem that it is difficult to accurately define the boundary of the transition zone due to subjective judgment during the measurement of the partial decarburization layer, which leads to measurement errors / deviations. After adopting a large amount of data fitting correction, the observation points or test points of the measurement errors or deviations caused by human factors are continuously calibrated, and the boundary of the partial decarburization layer is gradually defined. When the judgment limit is accurate, the calibration of the experimental data is realized, and a multivariate linear regression equation that conforms to the experimental results is obtained. This has guiding significance for controlling the thickness of the partial decarburization layer of the product and ensuring the reliable delivery of rolled materials with excellent performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a data processing flow chart of the present invention;

[0026] Figure 2-4 This is the metallographic diagram of the embodiment. DETAILED DESCRIPTION Example

[0027] This embodiment provides a method for correcting human error / bias in measuring the thickness of a decarburized layer of spring steel, such as Figure 1 As shown, the experimental steel is 60Si2Mn spring steel with a size of 10×10×12mm. A circular hole with a diameter of 5mm is taken in the middle of the sample, and a nickel-chromium wire is passed through it. The sample is suspended in a tubular heating furnace, and the sample is heated from room temperature using different heating speeds, holding temperatures, holding times, and decarburization atmospheres. The surface of the experimental steel after heat treatment is polished, and the thickness of the partial decarburization layer is measured using metallographic method. The initial experimental data are shown in Table 1:

[0028] Table 1 Decarburized layer thickness

[0029] X1 X2 X3 X4 Y Experimental Group Heating temperature (℃) Heating speed (℃ / min) Insulation time (min) Oxygen concentration (%) Partial decarburized layer µm 1 1080 5 75 2 355.8 2 1080 6.5 85 5 315.3 3 1080 8 95 7 224.5 4 1100 5 85 7 285.9 5 1100 6.5 95 2 256.7 6 1100 8 75 5 242.2 7 1120 5 95 5 350.7 8 1120 6.5 75 7 290.6 9 1120 8 85 2 295.3

[0030] The multivariate linear regression fitting relationship between the thickness of the decarburized layer of the steel and the heating temperature, heating rate, holding time and oxygen content was initially established, and the goodness of fit (Adj-R 2 )=0.37939<0.9, and the qualitative relationship between the decarburized layer thickness and the four factors violates the basic diffusion law.

[0031] By adjusting the decarburized layer thickness data of all parts multiple times, the Adj-R 2 The size of the observation points or test points 1, 2, 3, 4, 5, 6, and 9 may be caused by human factors. According to the metallographic diagram of the sample, the thickness of the decarburized layer corresponding to these observation points or test points is re-measured, and the orthogonal table is used for fitting to obtain the multivariate linear regression equation after quadratic fitting. Check Adj-R again 2 Still less than 0.9, repeat the above steps several times until Adj-R 2 The relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law. Finally, the following fitting formula is obtained:

[0032] y=-115.55+0.59167*x1-35.6*x2+0.055*x3-5.2386*x4

[0033] The corrected data are shown in Table 2:

[0034] Table 2 Decarburized layer thickness and four factors

[0035] X1 X2 X3 X4 Y Experimental Group Heating temperature (℃) Heating speed (℃ / min) Insulation time (min) Oxygen concentration (%) Partial decarburized layer µm 1 1080 5 75 2 345.8 2 1080 6.5 85 5 275.3 3 1080 8 95 7 214.5 4 1100 5 85 7 315.9 5 1100 6.5 95 2 286.7 6 1100 8 75 5 212.2 7 1120 5 95 5 350.7 8 1120 6.5 75 7 290.6 9 1120 8 85 2 265.3

[0036] ①The fourth group of samples

[0037] The heat treatment was carried out at a temperature of 1100°C, a heating rate of 5°C / min, a holding time of 85min, and an oxygen concentration of 7%.

[0038] The initial experimental data is 285.9µm, and the predicted value of the first fitting formula is 313.8842µm, which is 27.98421µm higher than the initial experimental data. The actual decarburized layer thickness is compared with the metallographic image, and the initial data is increased by 10µm to correct the data. At the same time, other data are adjusted. The fitting formula still cannot meet the expected value. Choose to reduce or continue to increase the correction data. After repeating the above steps many times until Adj-R 2 The relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law. Finally, the following fitting formula is obtained:

[0039] y=-115.55+0.59167*x1-35.6*x2+0.055*x3-5.2386*x4

[0040] The goodness of fit (Adj-R 2 ) = 0.92372>0.9, and the qualitative relationship between the decarburization layer thickness and the four factors does not violate the basic diffusion law. The thickness of the oxidation decarburization layer in the fourth group of experiments after correction is 30µm higher than that before adjustment, which is close to the actual partial decarburization layer thickness in the metallographic image.

[0041] Successfully calibrated experimental data, such as Figure 2 shown.

[0042] ② The sixth group of samples

[0043] The heat treatment was carried out at a temperature of 1100°C, a heating rate of 8°C / min, a holding time of 75min, and an oxygen concentration of 5%.

[0044] The initial experimental data is 242.2µm, and the predicted value of the first fitting formula is 259.64298µm, which is 17.44298µm higher than the initial experimental data. The actual decarburized layer thickness of the metallographic image is compared, and the initial data is increased by 10µm to correct the data. At the same time, other data are adjusted. The fitting formula still cannot meet the expected value. Choose to reduce or continue to increase the correction data. After repeating the above steps many times until Adj-R 2 The relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law. Finally, the following fitting formula is obtained:

[0045] y=-115.55+0.59167*x1-35.6*x2+0.055*x3-5.2386*x4

[0046] The goodness of fit (Adj-R 2 ) = 0.92372>0.9, and the qualitative relationship between the decarburization layer thickness and the four factors does not violate the basic diffusion law. The thickness of the oxidation decarburization layer in the fourth group of experiments after correction is reduced by 30µm compared with that before adjustment, which is close to the actual partial decarburization layer thickness in the metallographic image.

[0047] Successfully corrected data, such as Figure 3 shown.

[0048] ③The seventh group of samples were heat treated at a temperature of 1120°C, a heating rate of 5°C / min, a holding time of 95min, and an oxygen concentration of 5%.

[0049] The initial experimental data is 350.7µm, and the predicted value of the first fitting formula is 324.37632µm, which is 26.3237µm lower than the initial experimental data. Compared with the actual decarburized layer thickness of the metallographic image, the initial data is increased by 10µm to correct the data. At the same time, other data are adjusted. The fitting formula still cannot meet the expected value. Choose to reduce or continue to increase the correction data. After repeating the above steps many times until Adj-R 2 The relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law. Finally, the following fitting formula is obtained:

[0050] y=-115.55+0.59167*x1-35.6*x2+0.055*x3-5.2386*x4

[0051] The goodness of fit (Adj-R 2 ) = 0.92372>0.9, and the qualitative relationship between the decarburization layer thickness and the four factors does not violate the basic diffusion law. The thickness of the oxidation decarburization layer in the fourth group of experiments after correction has not changed compared with that before adjustment, which is close to the actual partial decarburization layer thickness in the metallographic image.

[0052] Successfully corrected data, such as Figure 4 shown.

[0053] In addition to the above embodiments, the present invention may also have other implementation modes. Any technical solution formed by equivalent replacement or equivalent transformation falls within the protection scope required by the present invention.

Claims

1. A method for correcting human error / bias in measuring the thickness of a decarburized layer of spring steel, characterized in that: Specifically: (1) Using the four-factor three-level orthogonal experimental design method, an orthogonal table including heating temperature, heating rate, insulation time and oxygen content was designed, and the samples were tested; (2) The samples after the experiment were processed by metallographic method, and the thickness of the partial decarburized layer of each sample was measured. Combined with the orthogonal table in step (1), a multivariate linear regression fitting relationship between the thickness of the decarburized layer of the steel and the heating temperature, heating rate, holding time and oxygen content was preliminarily constructed, and the multivariate linear regression fitting equation was obtained: y=A0+A1*x1+A2*x2+A3*x3+A4*x4 Where: y is the predicted decarburized layer thickness, in µm; A0, A1, A2, A3, A4 are the coefficients of the fitting equation; X1, X2, X3, X4 are the heating temperature, heating rate, holding time and oxygen content, respectively; (3) Check the corrected goodness of fit Adj-R 2 Whether it reaches 0.9 and the qualitative relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law, If Adj-R 2 If the decarburization thickness reaches 0.9 and the qualitative relationship between the decarburization layer thickness and the four factors does not violate the basic diffusion law, the partial decarburization thickness measured by the above multivariate linear fitting equation is accurate; if Adj-R 2 It did not reach 0.

9. The thickness of the decarburized layer in all parts was adjusted several times to observe Adj-R. 2 The size of the observation points or test points that may be caused by human factors is determined. According to the metallographic diagram of the sample, the thickness of the decarburized layer corresponding to these observation points or test points is remeasured and brought into the orthogonal table for fitting to obtain the multivariate linear regression equation after quadratic fitting. (4) Check Adj-R after quadratic fitting 2 Whether it reaches 0.9 and whether the qualitative relationship between the decarburized layer thickness and the four factors violates the basic diffusion law, repeat step (3) many times until Adj-R 2 It reaches above 0.90 and the qualitative relationship between the decarburized layer thickness and the four factors does not violate the basic diffusion law; (5) Finally, the adjusted decarburized layer thickness at the observation point or test point is compared with the actual decarburized layer, and the boundary between the decarburized layer and the non-decarburized layer is redefined to determine the accuracy of the redefined boundary. If it is accurate, the error or deviation caused by human factors is corrected. If it is inaccurate, it means that the fitting regression relationship established above is inappropriate and needs to be recalibrated.

Images