A method for controlling the liquid level of a crystallizer of a slab continuous caster and an automatic casting start method

Abstract

The application discloses a slab continuous casting machine crystallizer liquid level control method and an automatic casting method, and belongs to the technical field of metallurgy. In the liquid level control stage one, the crystallizer liquid level is raised from the initial liquid level value to the first liquid level set value; in the liquid level control stage two, the crystallizer liquid level is raised from the first liquid level set value to the second liquid level set value, and the second liquid level set value is the liquid level when the casting blank starts to be pulled out. The control method determines whether the parameters are abnormal through parameter collection and parameter analysis, adjusts the stopper opening degree for the crystallizer liquid level control according to the determination result, carries out single variable statistical process control and multivariable statistical process control through single variable analysis and multivariable analysis, and then adjusts the control stopper opening degree value, so that the crystallizer liquid level control in the automatic casting operation is realized, unnecessary shutdown and adjustment time are avoided, the efficiency of the continuous casting production line is maximally improved, the production capacity of the production equipment is fully utilized, and the utilization rate of the production line is greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of metallurgical technology, and in particular relates to a method for controlling the liquid level in the crystallizer of a slab continuous casting machine and an automatic casting start method. Background Technology

[0002] Continuous casting is a commonly used casting process in the steel industry. It involves directly injecting molten metal into a crystallizer and controlling cooling conditions to allow the molten steel to solidify according to process requirements, achieving a continuous casting process and producing billets with specific shapes and dimensions. Continuous casting is characterized by high efficiency, speed, and automation. Currently, it accounts for over 90% of steel production, playing a crucial role in modern steel production. Automatic tundish start-up is a key technology ensuring automatic start-up within the required timeframe after important component replacements such as tundish changes and submerged entry nozzle replacements. Although some publicly available methods can achieve automatic tundish start-up, several critical factors still need to be considered in controlling the crystallizer level during automatic start-up, such as the timing of start-up, flow rate adjustment, and the stability of billet quality. The automatic start-up process relies on manual monitoring, lacking effective monitoring methods. Therefore, effectively controlling the crystallizer level significantly impacts the success rate of automatic tundish start-up and the quality of the billets during the start-up process. Summary of the Invention

[0003] In view of the above-mentioned problems, the present invention discloses a method for controlling the liquid level in the crystallizer of a slab continuous casting machine and an automatic casting start method.

[0004] The present invention adopts the following technical solution:

[0005] A method for controlling the liquid level in the crystallizer of a slab continuous casting machine, the method being used for automatic start-up of the casting process after tundish replacement in slab continuous casting, the method comprising the following steps:

[0006] S1, Level control stage one, stage one adjusts the liquid level of the crystallizer by controlling the opening value of the stopper rod, raising the liquid level of the crystallizer from the initial liquid level value to the first liquid level set value;

[0007] S2, Level control stage two, the second stage adjusts the crystallizer level by controlling the stopper opening value, raising the crystallizer level from the first level setting value to the second level setting value, the second level setting value being the level when the billet begins to be pulled out;

[0008] S3, Billet Pulling Stage: After the liquid level in the crystallizer reaches the second liquid level setting value, the fan-shaped section is activated, the billet is pulled out, and automatic casting is completed.

[0009] The adjustment and control of the stopper opening value includes: parameter acquisition and parameter analysis, determining whether the parameter is abnormal based on the result of the parameter analysis, and adjusting the stopper opening to control the liquid level of the crystallizer based on the determination result;

[0010] The parameters include: steel grade, cross-sectional width, tundish temperature, tundish weight, crystallizer liquid level, stopper opening, submerged nozzle depth, and casting speed.

[0011] The parameter analysis includes univariate analysis and multivariate analysis. The univariate analysis is used for univariate statistical process control, and the multivariate analysis is used for multivariate statistical process control. The univariate statistical process control and multivariate statistical process control are used for the adjustment control of the stopcock opening value.

[0012] The univariate statistical process control includes: collecting parameter data in chronological order during the production process, determining control limits based on the parameter data, and controlling the corresponding parameter data to be within the control limits during the production process.

[0013] The multivariate statistical process control includes: for a production process in which multiple variables are highly correlated, simultaneously collecting parameter data of multiple variables, and using multivariate statistical process control to comprehensively analyze the correlation between multiple variables and the production process and product quality, thereby realizing multivariate production process monitoring and quality anomaly analysis.

[0014] Furthermore, the adjustment and control of the stopper opening value includes: using regression analysis to fit a formula relating the stopper opening to the steel flow rate, where the steel flow rate Q = f(K) = V*S, and K is the stopper opening, V is the molten steel flow rate, and S is the opening area.

[0015] Furthermore, the multivariate analysis includes: applying PCA-T to the parameters. 2 - The SPE analysis method is used to analyze and obtain control limits, and compare them with T. 2 And the SPE statistic and control limits, if T 2 If the SPE statistic is greater than the control limit, the observation data is considered faulty; if T 2 If the SPE statistic is less than or equal to the control limit, the observed data are considered normal.

[0016] Furthermore, the multivariate analysis includes:

[0017] A process monitoring model was established, collecting operational data during the initial casting process. Data from the automatic casting process when the tundish quality met requirements was selected as training data. This training data was first standardized, and then principal component analysis was performed to construct a Ti model. 2 Using the SPE statistic, determine the corresponding control limits, which are the criteria for the model to judge anomalies;

[0018] Process monitoring and anomaly detection are used to obtain new intermediate ladle pouring data and calculate T based on the new data. 2 The statistical quantity and SPE statistical quantity are compared with the control limit to determine whether the corresponding control limit is exceeded. This is used for online real-time detection of tundish opening anomalies and to realize automatic monitoring of tundish opening during continuous casting.

[0019] Parameter optimization for the automatic pouring process T 2 The main parameters that exceed the control limits for the statistics and SPE statistics are optimized. For the stopper opening, the range of opening K is selected as μ±2σ, where μ is the parameter mean and σ is the parameter standard deviation.

[0020] Furthermore, the PCA-T 2 -SPE analysis methods include T 2 The calculation of the statistic, namely T 2 The steps for calculating a statistic include:

[0021] S1.1 The original data of the parameters are centered by subtracting the mean of each variable, so that the mean of the data is zero;

[0022] S1.2 Calculate the covariance matrix of the data to reflect the correlation between the variables;

[0023] S1.3 Perform eigenvalue decomposition on the covariance matrix to obtain its eigenvalues ​​and eigenvectors;

[0024] S1.4 Multiply the original data by the eigenvector matrix to obtain the projected data, wherein the projected data is the principal component;

[0025] S1.5 Calculation T 2 The statistic, with the formula: Among them, z i It is the i-th principal component, λ i is the i-th eigenvalue, and k is the number of principal components;

[0026] S1.6 Calculate T 2 The control limits for the statistic are given by the following formula: Where n is the number of samples, k is the number of principal components, α is the significance level, and F0 is the number of samples. α (k, n―k) is the α quantile of an F distribution with degrees of freedom k and n―k.

[0027] Furthermore, the PCA-T 2 -SPE analysis methods include: The steps for implementing the SPE statistic include:

[0028] S2.1 Preprocess the raw data of the parameters, the preprocessing including centering, standardization or removal of outliers;

[0029] S2.2 Perform principal component analysis on the preprocessed data to obtain its principal component matrix and contribution rate;

[0030] S2.3 Determine the number of principal components based on the contribution rate, and select principal components with a cumulative contribution rate of 90% or higher;

[0031] S2.4 preprocesses the new observation data to make it have the same scale as the original data.

[0032] S2.5 Calculate the SPE statistic using the formula: SPE = |x 2 |=|(I―PP T )x| 2 Where x is the new observation data, P is the principal component matrix, and I is the identity matrix;

[0033] S2.6 The control limits for the SPE statistic are calculated using the following formula: In the formula, For X n×p The i-th power of the j-th eigenvalue of the covariance matrix, u α α is the critical value of a normal distribution at a significance level of α, g is the number of principal components retained in the model, and p is the number of variables.

[0034] Furthermore, the liquid level in the crystallizer rises rapidly at first and then slowly. In the automatic casting model, the liquid level height in the crystallizer during the casting process is:

[0035]

[0036] Where h is the liquid level height in the crystallizer, T is the casting start time, L1 is the total length of the crystallizer, L2 is the length of the dummy bar, v1 is the actual casting speed, and v2 is the target casting speed.

[0037] Furthermore, the crystallizer level control includes: during the casting process, after the billet is pulled out, controlling the crystallizer level to fluctuate periodically, with the fluctuation range being 800±3mm.

[0038] Furthermore, the liquid level control stage one includes four time periods, namely the first time period, the second time period, the third time period and the fourth time period, all of which are within the range of 1-10 seconds. Through the four time periods, the liquid level of the crystallizer rises from the initial liquid level value to the first liquid level set value.

[0039] At the beginning of the first time period, the stopper rod opening is adjusted to the first opening value and maintained at this value throughout the first time period. At the end of the first time period, the first opening value is reduced by 3-3.5 mm to reach the initial value of the second time period. During the second time period, the opening value slowly increases, and at the end of the second time period, the opening value increases by 0.5-1 mm relative to the initial value of the second time period, reaching the tail value of the second time period. During the third time period, the opening value continues to increase based on the tail value of the second time period. The stopper rod opening value increases by 1-1.5 mm within the segment, reaching the end value of the third time period; in the fourth time period, the stopper rod opening value continues to increase based on the end value of the third time period, increasing by 0.5-1 mm; the initial liquid level value is 680±5 mm; the first liquid level setting value is 752 mm; the second liquid level setting value is 775 mm; the first stopper rod opening value is k×10 mm, where k is the ratio of the current minimum cross-sectional width of the billet to the standard cross-sectional width, and the standard cross-sectional width is 1300 mm.

[0040] An automatic start-up method for tundish in slab continuous casting, wherein the start-up method uses the above-mentioned control method to control the liquid level in the crystallizer.

[0041] Beneficial effects:

[0042] This invention focuses on the dynamic and stable control of the crystallizer liquid level to meet the process requirements of tundish start-up in continuous casting production. Combining automatic control principles, by effectively controlling the dynamic stability of the crystallizer liquid level, it ensures that the continuous casting machine produces high-quality steel without leakage or overflow during operation. The application of automatic tundish start-up technology in slab continuous casting can effectively improve the production efficiency and quality stability of slab continuous casting production lines. By monitoring the status parameters of the tundish and crystallizer in real time and automatically starting the casting process according to preset parameters, unnecessary downtime and adjustment time can be avoided, maximizing the efficiency of the continuous casting production line, fully utilizing the capacity of production equipment, and thus greatly improving the utilization rate of the production line.

[0043] The crystallizer liquid level control method disclosed in this invention reduces human intervention, lowers quality fluctuations and production risks caused by differences in operator skill levels and fatigue, and ensures the stability and consistency of billet quality.

[0044] In the technical solution of this invention, the crystallizer liquid level control is based on the liquid level setting value and the set time in the casting procedure, and the stopper rod executes the corresponding opening degree according to the crystallizer liquid level setting value; after the crystallizer liquid level reaches the set value, the crystallizer liquid level enters the automatic control mode, and the stopper rod opening degree determines the flow rate of molten steel from the tundish into the crystallizer. During the automatic control of the crystallizer liquid level, the stopper rod opening degree control is automatically realized by the crystallizer liquid level control system; the crystallizer liquid level control method disclosed in this invention not only realizes automatic casting and ensures the automatic casting rate, but also ensures the safety of tundish casting. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of the crystallizer liquid level during the automatic opening of the tundish in the continuous casting of slabs according to the present invention;

[0047] Figure 2 This is a schematic diagram of the stopper rod position during the automatic opening of the tundish in the continuous casting of slabs according to the present invention;

[0048] Figure 3 This is a schematic diagram of the casting speed during the automatic opening of the tundish in the continuous casting of slabs according to the present invention;

[0049] Figure 4 This is a schematic diagram of the weight of the tundish during the automatic opening and pouring process of the tundish in the continuous casting of slabs according to the present invention;

[0050] Figure 5 This is a schematic diagram of the cumulative contribution rate of PCA principal component variance during the automatic opening of the tundish in slab continuous casting of the present invention;

[0051] Figure 6 The present invention relates to parameter monitoring during the automatic start-up process of tundish in slab continuous casting. 2 Schematic diagram of SPE results;

[0052] Figure 7 This is a schematic diagram illustrating the contribution values ​​of the main variables in the abnormal data sample of the replacement of the No. 35 tundish during the automatic start-up process of slab continuous casting tundish in this invention. Detailed Implementation

[0053] To better understand the technical solution of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0055] Example 1

[0056] A method for controlling the liquid level in the crystallizer of a slab continuous casting machine, the method being used for automatic start-up of the casting process after tundish replacement in slab continuous casting, the method comprising the following steps:

[0057] S1 Liquid level control stage one, wherein the liquid level of the crystallizer is adjusted by controlling the opening value of the stopper rod, and the liquid level of the crystallizer is raised from the initial liquid level value to the first liquid level set value;

[0058] S2 Liquid level control stage two, the stage two adjusts the liquid level of the crystallizer by controlling the opening value of the stopper rod, raising the liquid level of the crystallizer from the first liquid level setting value to the second liquid level setting value, the second liquid level setting value being the liquid level when the billet begins to be pulled out;

[0059] During the S3 billet pulling-out stage, after the liquid level in the crystallizer reaches the second liquid level setting value, the fan-shaped section operation is started, the billet is pulled out, and the automatic casting is completed.

[0060] The adjustment and control of the stopper opening value includes: parameter acquisition and parameter analysis, determining whether the parameter is abnormal based on the result of the parameter analysis, and adjusting the stopper opening to control the liquid level of the crystallizer based on the determination result;

[0061] The parameters include: steel grade, cross-sectional width, tundish temperature, tundish weight, crystallizer liquid level, stopper opening, submerged nozzle depth, and casting speed.

[0062] The parameter analysis includes univariate analysis and multivariate analysis. The univariate analysis is used for univariate statistical process control, and the multivariate analysis is used for multivariate statistical process control. The univariate statistical process control and multivariate statistical process control are used for the adjustment control of the stopcock opening value.

[0063] The univariate statistical process control includes: collecting parameter data in chronological order during the production process, determining control limits based on the parameter data, and controlling the corresponding parameter data to be within the control limits during the production process.

[0064] The multivariate statistical process control includes: for a production process in which multiple variables are highly correlated, simultaneously collecting parameter data of multiple variables, and using multivariate statistical process control to comprehensively analyze the correlation between multiple variables and the production process and product quality, thereby realizing multivariate production process monitoring and quality anomaly analysis.

[0065] Furthermore, the adjustment and control of the stopper opening value includes: using regression analysis to fit a formula relating the stopper opening to the steel flow rate, where the steel flow rate Q = f(K) = V*S, and K is the stopper opening, V is the molten steel flow rate, and S is the opening area.

[0066] Furthermore, the multivariate analysis includes: applying PCA-T to the parameters. 2 - The SPE analysis method is used to analyze and obtain control limits, and compare them with T. 2 And the SPE statistic and control limits, if T 2 If the SPE statistic is greater than the control limit, the observation data is considered faulty; if T 2 If the SPE statistic is less than or equal to the control limit, the observed data are considered normal.

[0067] Furthermore, the multivariate analysis includes:

[0068] A process monitoring model was established, collecting operational data during the initial casting process. Data from the automatic casting process when the tundish quality met requirements was selected as training data. This training data was first standardized, and then principal component analysis was performed to construct a Ti model. 2 Using the SPE statistic, determine the corresponding control limits, which are the criteria for the model to judge anomalies;

[0069] Process monitoring and anomaly detection are used to obtain new intermediate ladle pouring data and calculate T based on the new data. 2 The statistical quantity and SPE statistical quantity are compared with the control limit to determine whether the corresponding control limit is exceeded. This is used for online real-time detection of tundish opening anomalies and to realize automatic monitoring of tundish opening during continuous casting.

[0070] Parameter optimization for the automatic pouring process T 2 The main parameters that exceed the control limits for the statistics and SPE statistics are optimized. For the stopper opening, the range of opening K is selected as μ±2σ, where μ is the parameter mean and σ is the parameter standard deviation.

[0071] Furthermore, the PCA-T 2 -SPE analysis methods include T 2 The calculation of the statistic, namely T 2 The steps for calculating a statistic include:

[0072] S1.1 The original data of the parameters are centered by subtracting the mean of each variable, so that the mean of the data is zero;

[0073] S1.2 Calculate the covariance matrix of the data to reflect the correlation between the variables;

[0074] S1.3 Perform eigenvalue decomposition on the covariance matrix to obtain its eigenvalues ​​and eigenvectors;

[0075] S1.4 Multiply the original data by the eigenvector matrix to obtain the projected data, wherein the projected data is the principal component;

[0076] S1.5 Calculation T 2 The statistic, with the formula: Among them, z i It is the i-th principal component, λ i is the i-th eigenvalue, and k is the number of principal components;

[0077] S1.6 Calculate T 2 The control limits for the statistic are given by the following formula: Where n is the number of samples, k is the number of principal components, α is the significance level, and F0 is the number of samples. α (k, n―k) is the α quantile of an F distribution with degrees of freedom k and n―k.

[0078] Furthermore, the PCA-T 2 -SPE analysis methods include: The steps for implementing the SPE statistic include:

[0079] S2.1 Preprocess the raw data of the parameters, the preprocessing including centering, standardization or removal of outliers;

[0080] S2.2 Perform principal component analysis on the preprocessed data to obtain its principal component matrix and contribution rate;

[0081] S2.3 Determine the number of principal components based on the contribution rate, and select principal components with a cumulative contribution rate of 90% or higher;

[0082] S2.4 preprocesses the new observation data to make it have the same scale as the original data.

[0083] S2.5 Calculate the SPE statistic using the formula: SPE = |x 2 |=|(I―PP T )x| 2 Where x is the new observation data, P is the principal component matrix, and I is the identity matrix;

[0084] S2.6 The control limits for the SPE statistic are calculated using the following formula: In the formula, For X n×P The i-th power of the j-th eigenvalue of the covariance matrix, u α α is the critical value of a normal distribution at a significance level of α, h is the number of principal components retained in the model, and p is the number of variables.

[0085] Furthermore, the liquid level in the crystallizer rises rapidly at first and then slowly. In the automatic casting model, the liquid level height in the crystallizer during the casting process is:

[0086]

[0087] Where: h: crystallizer liquid level height (mm); T: casting start time (s); L1: total length of crystallizer (mm); L2: dummy bar length (mm); v1: actual casting speed (m / min); v2: target casting speed (m / min). Figure 1 As shown, Figure 1 This is a schematic diagram of the crystallizer liquid level during the automatic start-up process of the tundish in slab continuous casting. The diagram shows the liquid level change curves from t1 to t6, where t1 to t5 is the first stage of crystallizer liquid level control and t6 is the second stage of crystallizer liquid level control.

[0088] Furthermore, the crystallizer level control includes: during the casting process, after the billet is pulled out, controlling the crystallizer level to fluctuate periodically, with the fluctuation range being 800±3mm.

[0089] Furthermore, the liquid level control stage one includes four time periods, namely the first time period, the second time period, the third time period and the fourth time period, all of which are within the range of 1-10 seconds. Through the four time periods, the liquid level of the crystallizer rises from the initial liquid level value to the first liquid level set value.

[0090] At the beginning of the first time period, the stopper rod opening is adjusted to the first opening value and maintained at this value throughout the first time period. At the end of the first time period, the first opening value is reduced by 3-3.5 mm to reach the initial value of the second time period. During the second time period, the opening value slowly increases, and at the end of the second time period, the opening value increases by 0.5-1 mm relative to the initial value of the second time period, reaching the tail value of the second time period. During the third time period, the opening value continues to increase based on the tail value of the second time period. Within the first stage, the stopper rod opening value increases by 1-1.5 mm, reaching the end value of the third stage. In the fourth stage, the stopper rod opening value continues to increase based on the end value of the third stage, increasing by 0.5-1 mm. The initial liquid level is 680±5 mm; the first liquid level setting is 752 mm; the second liquid level setting is 775 mm; the first stopper rod opening value is k×10 mm, where k is the ratio of the current minimum cross-sectional width of the cast billet to the standard cross-sectional width, and the standard cross-sectional width is 1300 mm. Figure 2 As shown, Figure 2This diagram illustrates the stopper rod position during the automatic opening of the tundish in slab continuous casting. The diagram shows the stopper rod opening curve from t1 to t6. By controlling the stopper rod opening, the flow rate of molten steel entering the crystallizer from the tundish can be controlled, i.e., the steel throughput, thus achieving the goal of controlling the crystallizer level. (The diagram is then compared to...) Figure 1 and Figure 2 When the stopper rod opening is large, the liquid level in the crystallizer increases significantly. When the liquid level in the crystallizer fluctuates significantly, the stopper rod opening will also fluctuate significantly. Figure 3 This diagram illustrates the casting speed during the automatic start-up process of the tundish in slab continuous casting. As shown, starting from t6, the casting speed increases slowly, gradually reaching the target speed. The speed increases synchronously with the liquid level in the crystallizer, ultimately stabilizing at the target speed, with the crystallizer level stabilizing at 800±3mm. (Comparison) Figure 2 and Figure 3 As the pulling speed increases, the stopper opening also increases to meet the crystallizer's requirements for throughput. For example... Figure 4 The diagram shows the weight of the tundish during the automatic start-up process of slab continuous casting. Throughout the process, the tundish flow rate steadily increases and eventually stabilizes at around 70t, at which point the tundish enters the stable casting stage. Automatic start-up is then complete, and the tundish maintains a stable weight until the ladle or tundish needs to be replaced. If the tundish reaches the end of its lifespan, the automatic start-up process for tundish replacement will begin again.

[0091] An automatic start-up method for tundish in slab continuous casting, wherein the start-up method uses the above-mentioned control method to control the liquid level in the crystallizer.

[0092] Example 2

[0093] A method for controlling the liquid level in the crystallizer of a slab continuous casting machine, the method being used for automatic start-up of the tundish in slab continuous casting, the method comprising:

[0094] The liquid level control stage one adjusts the liquid level of the crystallizer by controlling the opening value of the stopper rod, raising the liquid level of the crystallizer from the initial liquid level value to the first liquid level set value;

[0095] The second stage of liquid level control adjusts the liquid level in the crystallizer by controlling the opening value of the stopper rod, raising the liquid level in the crystallizer from the first liquid level setting value to the second liquid level setting value, which is the liquid level when the billet begins to be pulled out.

[0096] The adjustment and control of the stopper opening value includes: parameter acquisition and parameter analysis, determining whether the parameter is abnormal based on the result of the parameter analysis, and adjusting the stopper opening to control the liquid level of the crystallizer based on the determination result;

[0097] During automatic casting, the increase in the crystallizer level is directly affected by the stopper rod opening. Regression analysis is used to fit the relationship between the stopper rod opening and the molten steel flow rate: molten steel flow rate Q = f(K) = V*S, where K is the stopper rod opening, V is the molten steel flow rate, and S is the opening area. The stopper rod opening K determines the opening area S, thus becoming a function of the molten steel flow rate and K. In the first stage of level control, the flow rate into the crystallizer is controlled by adjusting the stopper rod opening and opening time, thereby controlling the crystallizer level increase. In the second stage of level control, the crystallizer level is stabilized at 800±3mm when the casting speed changes by controlling the stopper rod opening K.

[0098] Statistical process monitoring model establishment and anomaly monitoring process: Model establishment: By collecting key parameters during the billet casting process, and selecting automatic casting operation data from tundish replacement when the billet quality meets the requirements as training data, the training data is first standardized, and then principal component analysis is performed to construct the T... 2 And the SPE statistic, determine the corresponding control limits, which are the criteria for the model to judge anomalies; process monitoring and anomaly detection: acquire new intermediate ladle pouring data, and calculate T from the new data. 2 The SPE statistic is compared with the control limits to determine whether the corresponding control limits are exceeded, thereby achieving online real-time detection of tundish opening anomalies and realizing automatic monitoring of tundish opening during continuous casting; parameter optimization process: for the automatic opening process T 2 The main reasons for the anomalies caused by the training sample analysis of the statistical measures and SPE measures exceeding the limit are to identify which main variables caused the quality anomalies of the billet in the automatic casting process. The variables that contribute the most to the process anomalies are optimized. Specifically, the data statistics method is used. Taking the stopper opening degree as an example, the range of μ±2σ of the opening degree K is selected as the limit for selecting the stopper opening degree. μ is the mean of the parameter and σ is the standard deviation of the parameter.

[0099] The parameters include: steel grade, cross-sectional width, tundish temperature, tundish weight, crystallizer liquid level, stopper opening, submerged nozzle depth, and casting speed.

[0100] The above parameters play a major role in the growth and stabilization of the crystallizer level. The steel grade determines the basic flow characteristics of the molten steel. Considering the significant differences between low-carbon steel, medium-carbon steel, and peritectic steel, parameters are set for different steel grades. The cross-sectional width and casting speed together determine the throughput of the crystallizer. The stopper opening affects the area of ​​molten steel flowing from the tundish into the crystallizer, thus affecting the throughput. During the initial casting process, the weight of the tundish determines the depth of the molten steel; the deeper the molten steel, the faster the flow rate. The tundish temperature determines the superheat, i.e., the temperature of the molten steel exceeding the liquidus line; higher temperatures result in stronger fluidity. The insertion depth of the submerged entry nozzle determines the flow rate of the molten steel after it enters the crystallizer. The depth must ensure that the molten steel flowing into the crystallizer from the side holes is below the protective slag to avoid slag entrapment. The casting speed also determines the throughput. After the billet is pulled out, in the second stage of level growth, the opening increases with the increase of the casting speed until the casting speed stabilizes at the target value, and the crystallizer level is controlled at 800±3mm.

[0101] The parameter analysis includes univariate analysis and multivariate analysis. The univariate analysis is used for univariate statistical process control, and the multivariate analysis is used for multivariate statistical process control. The univariate statistical process control and multivariate statistical process control are used for the adjustment control of the stopcock opening value.

[0102] Statistical process control (SPC) methods are applied to monitor the automatic start-up process of the tundish. Conventional monitoring methods monitor a single variable, selecting key variables from the parameters based on metallurgical experience for individual monitoring. SPC charts and process capability analysis are used to observe the fluctuations of this variable over a period of time and determine its upper and lower limits. However, with increasing process complexity and the number of variables, it becomes difficult to simultaneously monitor the impact of all variables on the overall production quality. This can lead to situations where monitored variables are within normal ranges, yet production quality is abnormal. To address the high correlation between multiple variables and the complexity of accurately describing the production process, multivariate statistical process control (MSC) is used to comprehensively consider the correlations of various variables, achieving multivariate production process monitoring and quality anomaly analysis. Univariate SPC monitors the crystallizer level, and the collected data is plotted on a graph in chronological order, along with the upper and lower control limits. During normal start-up, if the process changes remain within the control limits, the changes are considered normal and the process is under statistical control. If data points exceed the control limits, it indicates that an abnormal factor has significantly impacted the process.

[0103] Once the billet is pulled out and the liquid level in the crystallizer reaches the second set value, the sector section operation is activated, the billet is pulled out, and automatic casting is completed.

[0104] Furthermore, the criteria for identifying anomalies in univariate analysis include the following eight situations: one point is more than three standard deviations away from the center line; nine consecutive points are on the same side; six consecutive points rise or fall continuously; fourteen consecutive points fluctuate in rise and fall; two out of three consecutive points are more than two standard deviations away from the center line; four out of five consecutive points are more than one standard deviation away from the center line; fifteen consecutive points are within one standard deviation of the center line; and eight consecutive points are more than one standard deviation away from the center line.

[0105] Furthermore, the multivariate analysis includes: applying PCA-T to the parameters. 2 - The SPE analysis method is used to analyze and obtain control limits. If the control limits are exceeded, an anomaly is predicted.

[0106] Furthermore, the PCA-T 2 -SPE analysis methods include: T 2 The steps for calculating a statistic include:

[0107] S1.1 The original data of the parameters are centered by subtracting the mean of each variable, so that the mean of the data is zero;

[0108] S1.2. Calculate the covariance matrix of the data to reflect the correlation between the variables;

[0109] S1.3 Perform eigenvalue decomposition on the covariance matrix to obtain its eigenvalues ​​and eigenvectors;

[0110] S1.4 Multiply the original data by the eigenvector matrix to obtain the projected data, which is the principal component;

[0111] S1.5 Calculation T 2 The statistic, with the formula: Among them, z i It is the i-th principal component, λ i is the i-th eigenvalue, and k is the number of principal components;

[0112] S1.6 Calculate T 2 The control limits for the statistic are given by the following formula: Where n is the number of samples, k is the number of principal components, α is the significance level, and F0 is the number of samples. α (k, n―k) is the α quantile of an F distribution with degrees of freedom k and n―k.

[0113] Furthermore, the PCA-T 2 -SPE analysis methods include: The steps for implementing the SPE statistic include:

[0114] S2.1 Preprocess the raw data of the parameters, the preprocessing including centering, standardization or removal of outliers;

[0115] S2.2 Perform principal component analysis on the preprocessed data to obtain its principal component matrix and contribution rate;

[0116] S2.3 Determine the number of principal components based on the contribution rate, and select principal components with a cumulative contribution rate of 90% or higher;

[0117] S2.4 preprocesses the new observation data to make it have the same scale as the original data.

[0118] S2.5 Calculate the SPE statistic using the formula: SPE = |x 2 |=|(I―PP T )x| 2 Where x is the new observation data, P is the principal component matrix, and I is the identity matrix;

[0119] S2.6 The control limits for the SPE statistic are calculated using the following formula: In the formula, For X n×p The i-th power of the j-th eigenvalue of the covariance matrix, u α α is the critical value of a normal distribution at a significance level of α, h is the number of principal components retained in the model, and p is the number of variables.

[0120] Furthermore, the PCA-T 2 -SPE analysis methods include: comparing T 2 And the SPE statistic and control limits, if T 2 If the SPE statistic is greater than the control limit, the observation data is considered faulty; if T 2 If the SPE statistic is less than or equal to the control limit, the observed data are considered normal.

[0121] Furthermore, the liquid level in the crystallizer rises rapidly at first and then slowly. In the automatic casting model, the liquid level height in the crystallizer during the casting process is:

[0122]

[0123] Where: h: liquid level in the crystallizer (mm); T: casting start time (s); L1: total length of the crystallizer (mm); L2: length of the dummy bar (mm); v1: actual casting speed (m / min); v2: target casting speed (m / min).

[0124] Furthermore, the crystallizer level control includes: during the casting process, after the billet is pulled out, controlling the crystallizer level to fluctuate periodically, with the fluctuation range being 800±3mm.

[0125] Furthermore, the first stage of liquid level control includes four time periods: a first time period, a second time period, a third time period, and a fourth time period, each lasting 1-10 seconds. During these four time periods, the liquid level in the crystallizer rises from its initial value to a first set liquid level value. At the beginning of the first time period, the stopper opening is adjusted to a first opening value and maintained within the first time period. At the end of the first time period, the first opening value of the stopper is instantaneously reduced by 3-3.5 mm to obtain the initial value for the second time period, during which the stopper opening value slowly increases. The stopper opening value at the end of the second time period is 0.5-1 mm higher than the initial stopper opening value of the second time period; during the third time period, the stopper opening value continues to increase, increasing by 1-1.5 mm; during the fourth time period, the stopper opening value continues to increase, increasing by 0.5-1 mm; the initial liquid level is 680±5 mm; the first liquid level setting is 752 mm; the second liquid level setting is 775 mm; the first stopper opening value is k×10 mm, where k is the ratio of the current minimum cross-sectional width of the billet to the standard cross-sectional width. The standard cross-sectional width is 1300 mm.

[0126] An automatic start-up method for tundish in slab continuous casting, wherein the start-up method uses the above-mentioned control method to control the liquid level in the crystallizer.

[0127] Example 3

[0128] A method for controlling the liquid level in the crystallizer of a slab continuous casting machine

[0129] The specific details of the slab continuous casting machine are as follows: No. 1 slab continuous casting machine has 1 machine and 2 strands, with an arc radius of 9.5m, a slab cross-sectional size of 900~1900mm×230~250mm, a slab length of 9~11m, a metallurgical length of 35.1m, an effective crystallizer length of 800mm, a casting speed range of 0.4~2.0m / min, and 11 secondary cooling zones, all of which adopt air-water mixed cooling.

[0130] 1. Data Acquisition and Preprocessing

[0131] Parameters were collected from 52 normal tundish casting starts. A normal tundish casting start means a relatively short mixed casting section and good billet quality. The main parameters collected included: steel grade, cross-sectional width, tundish temperature, tundish weight, crystallizer level, stopper opening, submerged entry nozzle depth, and casting speed, as shown in Table 1.

[0132] Table 1 Main parameters for data acquisition

[0133]

[0134] The collected data is cleaned, denoised, and normalized to lay a solid foundation for subsequent model training and prediction.

[0135] 2. Principal Component Analysis (PCA)

[0136] Principal Component Analysis (PCA) projects high-dimensional data onto low-dimensional data. The PCA model projects data samples onto principal component subspaces and residual subspaces, and establishes Ti values ​​in both subspaces. 2 Statistics and the squared prediction error (SPE) statistic.

[0137] Using the singular value decomposition method, the data matrix X n×p The singular value decomposition of X is X n×p =UΛV T , where U=[u1,u2,…,u n ]∈R n×n V = [v1, v2, ..., v n ]∈R p×p ,

[0138]

[0139] The column vectors of matrices U and V are mutually orthogonal, and the length of each vector is 1 (unit vector). The singular values ​​σ after eigenvalue decomposition... i The eigenvalues ​​λ of the covariance matrix i The square root of λ. Each eigenvalue λ represents the amount of variance information carried by the corresponding principal component, and they are arranged in the order λ1≥λ2≥…≥λ. k The order is sequential. Therefore, it can be represented in another way. Where σ1 is the matrix X n×p The first principal component, v1 is the first principal direction, and so on.

[0140] Cumulative Percent Variance (CPV) is the proportion of data variation explained by the first h principal components relative to the total data variation. The cumulative contribution rate of the first h principal components can be expressed as... Where, λ j This represents the eigenvalue corresponding to the j-th principal component. Generally, when the cumulative variance contribution rate reaches 85% or higher, it is considered that sufficient original data information has been captured. The number of the first four principal components is selected based on specific needs, model accuracy, and visualization requirements. The results of PCA analysis on the 52 samples of automatic pouring parameters for intermediate tundishes collected in this study are as follows: Figure 5 As shown.

[0141] 3. T 2 SPE statistic

[0142] (1)T 2 Statistic

[0143] T 2 The steps for calculating the statistic are as follows:

[0144] S1.1 The original data of the parameters are centered by subtracting the mean of each variable, so that the mean of the data is zero;

[0145] S1.2. Calculate the covariance matrix of the data to reflect the correlation between the variables;

[0146] S1.3 Perform eigenvalue decomposition on the covariance matrix to obtain its eigenvalues ​​and eigenvectors;

[0147] S1.4 Multiply the original data by the eigenvector matrix to obtain the projected data, which is the principal component;

[0148] S1.5 Calculation T 2 The statistic, with the formula: Among them, z i It is the i-th principal component, λ i is the i-th eigenvalue, and k is the number of principal components;

[0149] S1.6 Calculate T 2 The control limits for the statistic are given by the following formula: Where n is the number of samples, k is the number of principal components, α is the significance level, and F0 is the number of samples. α (k, n―k) is the α quantile of an F distribution with degrees of freedom k and n―k.

[0150] Compare T 2 Statistics and control limits, if T 2 If the statistic is greater than the control limit, then the null hypothesis is rejected, meaning the means are not equal to or do not equal to the given value; if T 2 If the statistic is less than or equal to the control limit, the null hypothesis is accepted, meaning the mean is equal to or equal to the given value.

[0151] (2) SPE statistic

[0152] The SPE statistic is a fault diagnosis method based on principal component analysis (PCA). It measures the projection of a sample onto the remaining subspace and reflects the amount of information in the sample that cannot be described by the principal component model. It can detect faults in non-principal variables or changes in the correlation between variables.

[0153] The steps for implementing the SPE statistic are as follows:

[0154] S2.1 Preprocess the raw data of the parameters, the preprocessing including centering, standardization or removal of outliers;

[0155] S2.2 Perform principal component analysis on the preprocessed data to obtain its principal component matrix and contribution rate;

[0156] S2.3 Determine the number of principal components based on the contribution rate, and select principal components with a cumulative contribution rate of 90% or higher;

[0157] S2.4 preprocesses the new observation data to make it have the same scale as the original data.

[0158] S2.5 Calculate the SPE statistic using the formula: SPE = |x 2 |=|(I―PP T )x| 2 Where x is the new observation data, P is the principal component matrix, and I is the identity matrix;

[0159] S2.6 The control limits for the SPE statistic are calculated using the following formula: In the formula, For X n×p The i-th power of the j-th eigenvalue of the covariance matrix, u α α is the critical value of a normal distribution at a significance level of α, g is the number of principal components retained in the model, and p is the number of variables.

[0160] Compare the SPE statistic with the control limits. If the SPE statistic is greater than the control limits, the observed data is considered faulty; if the SPE statistic is less than or equal to the control limits, the observed data is considered normal.

[0161] Specifically, the T2 statistic and SPE statistic of the 52 samples revealed outliers in samples 14 and 35. For example... Figure 6 As shown.

[0162] 4. Anomaly Contribution Analysis

[0163] T 2 Cumulative contribution value The cumulative contribution of the j-th variable at the i-th sample point to the h components is calculated. To determine which process variables caused T 2 The statistic exceeds the limit. The total contribution of the j-th variable from the i-th sample point to the h components. The formula for calculation is:

[0164] In the formula, x ij Let l represent the observed value of the j-th variable at the i-th sample point. jkLet s represent the j-th component of the k-th principal direction vector l. tk Let tk represent the standard deviation of the k-th principal component. Each sample point consists of p variables, and we need to calculate the sum of the projection values ​​of each variable in h directions, that is, the cumulative contribution of each variable to the h components.

[0165] The total contribution value Contr of SPE: SPE measures the distance of each sample point in the principal component space, reflecting the degree of deviation of the sample point from the normal process. By calculating the total contribution value Contr of the j-th variable at the i-th sample point to the SPE, we can assess which variable contributes more to the overall anomaly. When the squared prediction error (SPE) of the i-th sample point exceeds its control limit, the contribution value of the j-th variable at the i-th sample point to the SPE can be calculated and the SPE contribution plot can be output. The calculation formula is as follows: Where, x ij Let z be the j-th component of the outlier sample point x(i). ij Let z(i) be the j-th component of the reconstructed data of x(i). The outlier analysis results are as follows: Figure 7 As shown.

[0166] In the SPE contribution plot, the graphs with larger contribution values ​​are relatively tall, indicating that the main factor causing this anomaly is the anomaly of this variable. Through statistical control analysis of continuous casting process parameters, it was found that samples 14 and 35 exhibited T... 2 And the situation where the SPE control value is out of range. Taking sample point 35 as an example, the T value of this sample point... 2 The excessive SPE content was mainly due to variables 4, 5, and 6 exceeding the normal range, corresponding to the crystallizer level, stopper opening, and casting speed, respectively. To further improve the quality of tundish casting, the stopper opening was further optimized, and the optimization results are shown in Table 2 below.

[0167] Table 2. Main parameters for automatic tundish pouring after stopper opening optimization.

[0168]

[0169] This invention has been tested and verified in actual trials at steel mills. Statistical results show that this method can effectively improve the automatic start-up rate of tundish casting. Table 3 shows the verification records, with March 2023 showing the start-up rate using the traditional tundish method and November 2023 showing the start-up rate after applying this method. The statistical data comparison reveals that the method described in this invention has improved the automatic start-up rate for major steel grades, with a start-up success rate exceeding 90%.

[0170] Table 3 Verification Record of Automatic Casting Rate of Slab Tundish

[0171]

[0172] The embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for controlling the liquid level in the crystallizer of a slab continuous casting machine, the method being used for automatic start-up of the casting process when the tundish is replaced in slab continuous casting, characterized in that... The method includes the following steps: S1, Level control stage one, stage one adjusts the liquid level of the crystallizer by controlling the opening value of the stopper rod, raising the liquid level of the crystallizer from the initial liquid level value to the first liquid level set value; S2, Level control stage two, the second stage adjusts the crystallizer level by controlling the stopper opening value, raising the crystallizer level from the first level setting value to the second level setting value, the second level setting value being the level when the billet begins to be pulled out; S3, Billet Pulling Stage: After the liquid level in the crystallizer reaches the second liquid level setting value, the fan-shaped section is activated, the billet is pulled out, and automatic casting is completed. The adjustment and control of the stopper opening value includes: parameter acquisition and parameter analysis, determining whether the parameter is abnormal based on the result of the parameter analysis, and adjusting the stopper opening to control the liquid level of the crystallizer based on the determination result; The parameters include: steel grade, cross-sectional width, tundish temperature, tundish weight, crystallizer liquid level, stopper opening, submerged nozzle depth, and casting speed. The parameter analysis includes univariate analysis and multivariate analysis. The univariate analysis is used for univariate statistical process control, and the multivariate analysis is used for multivariate statistical process control. The univariate statistical process control and multivariate statistical process control are used for the adjustment control of the stopcock opening value. The univariate statistical process control includes: collecting parameter data in chronological order during the production process, determining control limits based on the parameter data, and controlling the corresponding parameter data to be within the control limits during the production process. The multivariate statistical process control includes: for production processes with multiple highly correlated variables, simultaneously collecting parameter data for multiple variables, and using multivariate statistical process control to comprehensively analyze the correlation between multiple variables and the production process and product quality, thereby achieving multivariate production process monitoring and quality anomaly analysis; the parameters are processed using PCA-T... 2 - The SPE analysis method is used to analyze and obtain control limits, and compare them with T. 2 And the SPE statistic and control limits, if T 2 If the SPE statistic is greater than the control limit, the observation data is considered faulty; if T 2 If the SPE statistic is less than or equal to the control limit, the observed data are considered normal. A process monitoring model is established by collecting operational data during the initial casting process. Data from the automatic casting process using an intermediate ladle with slab quality meeting requirements is selected as training data. This training data is first standardized, and then principal component analysis is performed to construct the T... 2 Using the SPE statistic, determine the corresponding control limits, which are the criteria for the model to judge anomalies; Process monitoring and anomaly detection are used to obtain new intermediate ladle pouring data and calculate T based on the new data. 2 The statistical quantity and SPE statistical quantity are compared with the control limit to determine whether the corresponding control limit is exceeded. This is used for online real-time detection of tundish opening anomalies and to realize automatic monitoring of tundish opening during continuous casting. Parameter optimization for the automatic pouring process T 2 The main parameters whose statistics and SPE statistics exceed the control limits are optimized. For the stopcock opening, the opening degree is selected. of The range is limited to selecting the stopper rod opening. The mean of the parameters, The standard deviation of the parameter is given.

2. The control method according to claim 1, characterized in that, The adjustment and control of the stopper rod opening value includes: using regression analysis to fit a formula relating the stopper rod opening to the steel flow rate, and the steel flow rate... ,in For the stopper rod opening, The flow rate of molten steel. Let be the area of ​​the opening.

3. The control method according to claim 2, characterized in that, The PCA-T 2 -SPE analysis methods include T 2 The calculation of the statistic, namely T 2 The steps for calculating a statistic include: S1.1 The original data of the parameters are centered by subtracting the mean of each variable, so that the mean of the data is zero; S1.2 Calculate the covariance matrix of the data to reflect the correlation between the variables; S1.3 Perform eigenvalue decomposition on the covariance matrix to obtain its eigenvalues ​​and eigenvectors; S1.4 Multiply the original data by the eigenvector matrix to obtain the projected data, wherein the projected data is the principal component; S1.5 Calculation T 2 The statistic, with the formula: ,in, It is the first Principal components, It is the first 1 eigenvalue, It is the number of principal components; S1.6 Calculate T 2 The control limits for the statistic are given by the following formula: ,in, It is the sample size. Principal components It is the significance level. Is it subject to the degree of freedom? and of Distribution Quantiles.

4. The control method according to claim 2, characterized in that, The PCA-T 2 - The SPE analysis method includes the implementation of the SPE statistic, and the steps of the implementation method of the SPE statistic include: S2.1 Preprocess the raw data of the parameters, the preprocessing including centering, standardization or removal of outliers; S2.2 Perform principal component analysis on the preprocessed data to obtain its principal component matrix and contribution rate; S2.3 Determine the number of principal components based on the contribution rate, and select principal components with a cumulative contribution rate of 90% or higher; S2.4 preprocesses the new observation data to make it have the same scale as the original data; S2.5 The SPE statistic is calculated using the following formula: ,in, This is new observational data. It is a principal component matrix. It is the identity matrix; S2.6 Calculate the control limits for the SPE statistic.

5. The control method according to claim 4, characterized in that, The liquid level in the crystallizer is increased in a rapid-then-slow manner. The liquid level height in the crystallizer during the automatic casting process in the automatic casting model is: in, This refers to the liquid level height in the crystallizer. This refers to the time for pouring. This is the total length of the crystallizer; The length of the derrick. For actual pulling speed, To achieve the target speed.

6. The control method according to claim 5, characterized in that, The crystallizer liquid level control includes: during the initial casting process, after the billet is pulled out, controlling the periodic fluctuation of the crystallizer liquid level, with the fluctuation range being 800±3 mm.

7. The control method according to claim 6, characterized in that, The liquid level control stage includes four time periods: the first time period, the second time period, the third time period, and the fourth time period. Each of the four time periods is within the range of 1-10 seconds. Through these four time periods, the liquid level in the crystallizer rises from the initial liquid level value to the first liquid level set value. At the beginning of the first time period, the stopper rod opening is adjusted to the first stopper rod opening value and maintained at the first stopper rod opening value during the first time period; at the end of the first time period, the first stopper rod opening value is reduced by 3-3.5mm to reach the initial value of the second stopper rod during the second time period. During the second time period, the stopper rod opening value slowly increases. At the end of the second time period, the stopper rod opening value increases by 0.5-1mm relative to the initial value of the second stopper rod during the second time period, and the stopper rod opening reaches the tail value of the second stopper rod during the second time period. During the third time period, the stopper opening value continues to increase based on the end value of the second time period. During the third time period, the stopper opening value increases by 1-1.5 mm, and the stopper opening reaches the end value of the third time period. During the fourth time period, the stopper opening value continues to increase based on the end value of the stopper in the third time period, increasing by 0.5-1 mm during the fourth time period; the initial liquid level value is 680±5 mm; the first liquid level setting value is 752 mm; the second liquid level setting value is 775 mm; the first stopper opening value is k×10 mm, where k is the ratio of the current minimum cross-sectional width of the billet to the standard cross-sectional width, and the standard cross-sectional width is 1300 mm.

Images