A method for determining and grading the liquid level fluctuation of slab crystallizer with high surface requirements

Abstract

The present invention relates to a method for determining and grading liquid level fluctuations in a crystallizer of a slab with high surface requirements, and belongs to the technical field of iron and steel metallurgy. The technical scheme of the present invention is: a corresponding model of liquid level fluctuation deviation and billet length is established through the continuous casting drawing speed and the crystallizer liquid level fluctuation measurement time; through the correspondence between the actual billet length and the billet drawing length model, a one-to-one correspondence between the liquid level fluctuation deviation data and the actual billet position is achieved, and the point and line connection between time and space in the production process is achieved; through the actual billet cutting length, the liquid level fluctuation deviation of each billet is statistically analyzed, and the liquid level fluctuation deviation of a single billet is determined and graded. The beneficial effect of the present invention is: the slag roll type peeling defects of ultra-deep drawing steel plates with high surface requirements are controlled within the steel plant, so as to effectively prevent problematic billets from flowing into users and reduce users' quality complaints and quality objections.

Description

Technical Field

[0001] The invention relates to a method for determining and grading liquid level fluctuations of a slab crystallizer with high surface requirements, and belongs to the technical field of iron and steel metallurgical methods. Background Art

[0002] Ultra-deep drawn automotive plates with high surface requirements are mainly used for exposed outer panels of automobiles. This series of products has strict requirements on surface defects of steel plates. Most automobile manufacturers require "zero defects" for exposed outer panels, especially for long strips of "cracking" metallurgical defects on the surface of the plates, which seriously affect automobile painting and are one of the main quality complaints of automobile manufacturers against steel companies.

[0003] The main reason for the "skin bursting" defect of automobile deep-drawing plates is the liquid level fluctuation of the continuous casting crystallizer. For plate continuous casting machines, the liquid level fluctuation of the crystallizer is generally controlled at ±5mm. Excessive liquid level fluctuation, especially the liquid level fluctuation of ±20mm, will cause a large amount of protective slag and refractory impurities to be drawn into the molten steel. During the rolling process of the ingot, this type of inclusions, unlike metal, do not have fluidity and are crushed into long strips, forming a typical "skin bursting" defect.

[0004] At present, there are two problems with the automatic determination of liquid level fluctuations. First, there is the correspondence between the liquid level fluctuation points and the ingot. The liquid level fluctuation is the measured displacement in the continuous casting crystallizer, which is the state of liquid molten steel in the continuous casting crystallizer. The ingot length is the measured value of continuous positioning when the ingot is cut. There are temporal and spatial deviations between the two. The correspondence between events is a common technical difficulty in the industry. Second, there is the problem of building an intelligent determination model for the liquid level fluctuation of the ingot. At present, the liquid level fluctuations in most domestic steel mills are determined manually, and the determination rules are relatively arbitrary. There are problems such as inconsistent determination standards, poor accuracy, and high misjudgment rate.

[0005] Patent application number 201710192048.0 discloses "a method for monitoring liquid level fluctuations in a crystallizer using 3D video or photography". This method uses a surface distance measurement method to determine the liquid level fluctuations in the crystallizer, thereby providing feedback on whether there is a trend of slag rolling. However, this method does not correspond the liquid level fluctuations to the ingot, and cannot effectively guide the quality judgment of the ingot.

[0006] Patent application number 201510982869.5 discloses "a method for measuring mold liquid level fluctuation and flow field symmetry". This method inserts a single row or multiple rows of steel nail plates made according to the cross-sectional dimensions of the mold and the water inlet into the mold, thereby judging the mold liquid level fluctuation according to the melting of the steel nails. However, this method also does not consider the correspondence between liquid level fluctuation and the ingot, and cannot effectively guide the ingot quality judgment.

[0007] Patent application number 202010601222.4 discloses "a fast, accurate, and multi-dimensional evaluation method for liquid level fluctuations in continuous casting machine crystallizers". This method achieves the correspondence between liquid level fluctuations and ingots through the position of the head and tail of the ingot and the pouring time, and determines the quality of the ingots by setting the ratio of unqualified points to the total number of points. This method does not fully consider the accuracy of the correspondence by corresponding the position and pouring time, and makes judgments based on the ratio of unqualified points. It does not fully consider the interference signals caused by points with large fluctuations and equipment fluctuations, and does not meet the "zero defect" requirements of ultra-deep drawn automotive plates with high surface requirements.

[0008] Therefore, it is necessary to find a reasonable and accurate method to correspond the liquid level fluctuation deviation with the ingot position, and to make a reasonable judgment on the quality of the liquid level fluctuation according to the quality requirements of high-surface automotive plates, so as to effectively avoid the "skin bursting" type slag defects of the plates from flowing into the hands of customers, causing user quality complaints and quality objections. Summary of the invention

[0009] The purpose of the present invention is to provide a method for determining and grading the liquid level fluctuation of a crystallizer of a slab with high surface requirements. By adopting the continuous casting drawing speed and the measurement time of the liquid level fluctuation of the crystallizer and the overall casting length of the continuous casting machine, a corresponding model of the drawing length and the liquid level fluctuation is established, and the accurate correspondence between the liquid level fluctuation deviation and the position of the ingot is achieved, thereby realizing the point and line connection between time and space in the production process; according to the liquid level fluctuation deviation, the liquid level fluctuation is subdivided into quality grades, thereby realizing the determination and grading of the surface quality of each ingot, and finally achieving the purpose of controlling the slag rolling and peeling defects of ultra-deep drawing steel plates with high surface requirements within the steel plant, effectively preventing problematic ingots from flowing into users, and reducing users' quality complaints and quality objections, thereby effectively solving the above-mentioned problems existing in the background technology.

[0010] The technical solution of the present invention is: a method for determining and grading liquid level fluctuations of a slab crystallizer with high surface requirements, comprising the following steps:

[0011] (1) A billet length model is established by measuring the continuous casting speed and the liquid level fluctuation measurement time of the crystallizer, the liquid level fluctuation deviation under the theoretical billet length is determined by the liquid level fluctuation time, and a corresponding model between the liquid level fluctuation deviation and the billet length is established;

[0012] (2) By matching the actual billet length with the billet drawing length model, the one-to-one correspondence between the liquid level fluctuation deviation data and the actual billet position is achieved, thus realizing the point and line connection between time and space in the production process;

[0013] (3) The actual length of the slab is used to statistically analyze the liquid level fluctuation deviation of each slab, and the Laida criterion is used to pre-process the liquid level fluctuation data;

[0014] (4) The liquid level fluctuation deviation of a single ingot is graded through big data analysis of the production site, production experience, and user quality requirements.

[0015] In the step (1), the calculation formula of the casting length model is:

[0016]

[0017] Where: Δt is the time interval between the adjacent time periods collected by the system, Δt = t i -t i-1 ; V i t i The continuous casting machine pulling speed at the moment; n is the number of samples collected from the start of the continuous casting to the end of the casting, and the data collection interval is 300 milliseconds;

[0018] The correspondence between the billet drawing length model and the billet length refers to the dimensionless processing of the calculated length and the actual length data, and finally the liquid level fluctuation deviation data of a single billet is obtained. i The actual billet position corresponding to the moment is:

[0019]

[0020] Where: L X is the actual length of the billet; L is the theoretical calculated length of the billet; L S L is the actual length of the billet produced; i is the theoretical calculated length corresponding to it;

[0021] Therefore, the corresponding relationship between the length of the casting billet and the liquid level fluctuation deviation is realized, that is:

[0022] YL X =YL i

[0023] Where: YL X YL is the liquid level fluctuation deviation at the actual length X of the billet; i It is the theoretical billet length Li corresponding to the actual billet length Lx, and the liquid level fluctuation deviation at the corresponding moment.

[0024] In the step (2), the liquid level fluctuation deviation under the theoretical casting length refers to t i The difference between the liquid level fluctuation value and the reference set value at the moment is calculated by the billet drawing length model to achieve the correspondence between the fluctuation deviation and the actual billet position, that is:

[0025] Y Li =Y i -a

[0026] Where: Y Li Theoretically calculated ti The liquid level fluctuation deviation corresponding to the length of the billet at the moment; Y i t i The liquid level fluctuation value collected at every moment; a is the liquid level fluctuation reference setting value.

[0027] In step (3), the liquid level fluctuation data is pre-processed using the Laida criterion, and the error data outside (μ-3σ, μ+3σ) is filtered to prevent the detection system deviation from causing judgment distortion, wherein:

[0028]

[0029]

[0030] In the step (4), the liquid level fluctuation deviation of a single ingot is divided into five levels: A+, A, B, C and D, wherein A+ is the highest quality level, and the probability of subsequent slag roll quality defects caused by liquid level fluctuation is the lowest; A is a good quality level, and the probability of quality defects is low; B is a poor quality level, which can be used for low-end contract delivery, or upgraded to A through flame cleaning; C is poor quality and must be flame cleaned before being used for low-end contract delivery; D indicates that the ingot is downgraded.

[0031] The A+ level means that the number of ±5mm peaks and troughs appearing on a single ingot is 0; the A level means that the number of ±5mm peaks and troughs appearing on a single ingot does not exceed 5 times; the B level means that the number of peaks and troughs appearing ≥±5mm and <±10mm is greater than 5 times, or the number of peaks and troughs appearing ≥±10mm is less than 5 times; the C level means that the number of peaks and troughs appearing ≥±10mm and <±15mm is greater than 5 times, or the number of peaks and troughs appearing ≥±15mm is less than 5 times; the D level means that the number of peaks and troughs appearing ≥±15mm and <±20mm is greater than 5 times, or the peaks or troughs of ≥±20mm appear.

[0032] For the determination of the wave crest, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of +5mm, +10mm, +15mm and +20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not lower than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is lower than the determination threshold value, the computer stops determining and accumulates the number of wave crests corresponding to the threshold value.

[0033] For the determination of troughs, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of -5mm, -10mm, -15mm and -20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not higher than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is higher than the determination threshold value, the computer stops determining and accumulates the number of troughs corresponding to the threshold value.

[0034] When the computer determines the total cumulative number of peaks and troughs that appear when the block billet corresponds to the threshold values ​​of ±5mm, ±10mm, ±15mm and ±20mm respectively, it will make the following judgments: First, when the total cumulative number of peaks and troughs that appear ≥±15mm and <±20mm is greater than 5 times, or the total cumulative number of peaks or troughs that appear ≥±20mm is greater than 0 times, it is judged as Class D; otherwise, when the total cumulative number of peaks and troughs that appear ≥±10mm and <±15mm is greater than 0 times, it is judged as Class D. If the total cumulative number of peaks and troughs ≥±15mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±15mm is less than 5 times, it is judged as Class C; otherwise, when the total cumulative number of peaks and troughs ≥±5mm and <±10mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±10mm is less than 5 times, it is judged as Class B; when the total cumulative number of ±5mm peaks and troughs on a single ingot does not exceed 5 times, it is judged as Class A; if none of the above situations occur, it is judged as Class A+.

[0035] The beneficial effects of the present invention are as follows: by adopting the continuous casting drawing speed and the crystallizer liquid level fluctuation measurement time and the overall casting length of the continuous casting machine, a corresponding model of the billet drawing length and the liquid level fluctuation is established, and the accurate correspondence between the liquid level fluctuation deviation and the billet position is achieved, thereby realizing the point and line connection between time and space in the production process; according to the liquid level fluctuation deviation, the liquid level fluctuation is subdivided into quality grades, thereby realizing the judgment and grading of the surface quality of each billet, and finally achieving the purpose of controlling the slag rolling and peeling defects of ultra-deep drawing steel plates with high surface requirements within the steel plant, effectively preventing problematic billets from flowing into users, and reducing users' quality complaints and quality objections. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a statistical diagram of liquid level fluctuation data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the invention implementation cases clearer, the technical solutions in the implementation cases of the present invention will be clearly and completely described below in conjunction with the drawings in the implementation cases. Obviously, the implementation cases described are only a small part of the implementation cases of the present invention, rather than all the implementation cases. Based on the implementation cases in the present invention, all other implementation cases obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0038] A method for determining and grading liquid level fluctuations of a slab crystallizer with high surface requirements comprises the following steps:

[0039] (1) A billet length model is established by measuring the continuous casting speed and the liquid level fluctuation measurement time of the crystallizer, the liquid level fluctuation deviation under the theoretical billet length is determined by the liquid level fluctuation time, and a corresponding model between the liquid level fluctuation deviation and the billet length is established;

[0040] (2) By matching the actual billet length with the billet drawing length model, the one-to-one correspondence between the liquid level fluctuation deviation data and the actual billet position is achieved, thus realizing the point and line connection between time and space in the production process;

[0041] (3) The actual length of the slab is used to statistically analyze the liquid level fluctuation deviation of each slab, and the Laida criterion is used to pre-process the liquid level fluctuation data;

[0042] (4) The liquid level fluctuation deviation of a single ingot is graded through big data analysis of the production site, production experience, and user quality requirements.

[0043] In the step (1), the calculation formula of the casting length model is:

[0044]

[0045] Where: Δt is the time interval between the adjacent time periods collected by the system, Δt = t i -t i-1 ; V i t i The continuous casting machine pulling speed at the moment; n is the number of samples collected from the start of the continuous casting to the end of the casting, and the data collection interval is 300 milliseconds;

[0046] The correspondence between the billet drawing length model and the billet length refers to the dimensionless processing of the calculated length and the actual length data, and finally the liquid level fluctuation deviation data of a single billet is obtained. i The actual billet position corresponding to the moment is:

[0047]

[0048] Where: L X is the actual length of the billet; L is the theoretical calculated length of the billet; L S L is the actual length of the billet produced; i is the theoretical calculated length corresponding to it;

[0049] Therefore, the corresponding relationship between the length of the casting billet and the liquid level fluctuation deviation is realized, that is:

[0050] YL X =YLi

[0051] Where: YL X YL is the liquid level fluctuation deviation at the actual length X of the billet; i It is the theoretical billet length Li corresponding to the actual billet length Lx, and the liquid level fluctuation deviation at the corresponding moment.

[0052] In the step (2), the liquid level fluctuation deviation under the theoretical casting length refers to t i The difference between the liquid level fluctuation value and the reference set value at the moment is calculated by the billet drawing length model to achieve the correspondence between the fluctuation deviation and the actual billet position, that is:

[0053] Y Li =Y i -a

[0054] Where: Y Li Theoretically calculated t i The liquid level fluctuation deviation corresponding to the length of the billet at the moment; Y i t i The liquid level fluctuation value collected at every moment; a is the liquid level fluctuation reference setting value.

[0055] In step (3), the liquid level fluctuation data is pre-processed using the Laida criterion, and the error data outside (μ-3σ, μ+3σ) is filtered to prevent the detection system deviation from causing judgment distortion, wherein:

[0056]

[0057]

[0058] In the step (4), the liquid level fluctuation deviation of a single ingot is divided into five levels: A+, A, B, C and D, wherein A+ is the highest quality level, and the probability of subsequent slag roll quality defects caused by liquid level fluctuation is the lowest; A is a good quality level, and the probability of quality defects is low; B is a poor quality level, which can be used for low-end contract delivery, or upgraded to A through flame cleaning; C is poor quality and must be flame cleaned before being used for low-end contract delivery; D indicates that the ingot is downgraded.

[0059] The A+ level means that the number of ±5mm peaks and troughs appearing on a single ingot is 0; the A level means that the number of ±5mm peaks and troughs appearing on a single ingot does not exceed 5 times; the B level means that the number of peaks and troughs appearing ≥±5mm and <±10mm is greater than 5 times, or the number of peaks and troughs appearing ≥±10mm is less than 5 times; the C level means that the number of peaks and troughs appearing ≥±10mm and <±15mm is greater than 5 times, or the number of peaks and troughs appearing ≥±15mm is less than 5 times; the D level means that the number of peaks and troughs appearing ≥±15mm and <±20mm is greater than 5 times, or the peaks or troughs of ≥±20mm appear.

[0060] For the determination of the wave crest, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of +5mm, +10mm, +15mm and +20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not lower than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is lower than the determination threshold value, the computer stops determining and accumulates the number of wave crests corresponding to the threshold value.

[0061] For the determination of troughs, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of -5mm, -10mm, -15mm and -20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not higher than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is higher than the determination threshold value, the computer stops determining and accumulates the number of troughs corresponding to the threshold value.

[0062] When the computer determines the total cumulative number of peaks and troughs that appear when the block billet corresponds to the threshold values ​​of ±5mm, ±10mm, ±15mm and ±20mm respectively, it will make the following judgments: First, when the total cumulative number of peaks and troughs that appear ≥±15mm and <±20mm is greater than 5 times, or the total cumulative number of peaks or troughs that appear ≥±20mm is greater than 0 times, it is judged as Class D; otherwise, when the total cumulative number of peaks and troughs that appear ≥±10mm and <±15mm is greater than 0 times, it is judged as Class D. If the total cumulative number of peaks and troughs ≥±15mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±15mm is less than 5 times, it is judged as Class C; otherwise, when the total cumulative number of peaks and troughs ≥±5mm and <±10mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±10mm is less than 5 times, it is judged as Class B; when the total cumulative number of ±5mm peaks and troughs on a single ingot does not exceed 5 times, it is judged as Class A; if none of the above situations occur, it is judged as Class A+.

[0063] In practical applications, the specific implementation steps are as follows:

[0064] Step 1: Install a computer at the production site, and establish a corresponding model between the liquid level fluctuation deviation and the billet length in the computer by corresponding the billet drawing length model and the billet length.

[0065] Among them, the calculation formula of the casting length model is:

[0066]

[0067] Where: Δt is the time interval between the adjacent time periods collected by the system, Δt = t i -t i-1 ;

[0068] V i t i The continuous casting machine speed at the time;

[0069] n is the number of samples collected from the start to the end of the continuous casting cycle, and the sampling interval is 300 milliseconds.

[0070] The correspondence between the billet length and the billet drawing length model refers to the need to perform dimensionless processing on the calculated length and the actual length, and finally obtain the liquid level fluctuation deviation data of a single billet. i The actual billet position corresponding to the moment is:

[0071]

[0072] Where: L X is the actual length position of the billet;

[0073] L is the theoretical length of the billet;

[0074] L S The actual length of the billet produced;

[0075] L i is the corresponding theoretical calculated length.

[0076] Therefore, the corresponding relationship between the billet position and the liquid level fluctuation deviation can be realized, that is:

[0077] Y LX =Y Li

[0078] Where: Y LX is the liquid level fluctuation deviation at the actual length X of the billet;

[0079] Y Li It is the theoretical billet length Li corresponding to the actual billet length Lx, and the liquid level fluctuation deviation at the corresponding moment.

[0080] Step 2: Process and optimize the liquid level fluctuation data. Calculate the liquid level fluctuation deviation, which means ti The difference between the liquid level fluctuation value and the reference set value at the moment is calculated through the billet drawing length model to achieve the correspondence between the fluctuation deviation and the billet position. That is:

[0081] Y Li =Y i -a

[0082] Where: Y Li Theoretically calculated t i The liquid level fluctuation deviation corresponding to the length of the billet at the moment;

[0083] Y i t i Liquid level fluctuation values ​​collected at all times;

[0084] a is the liquid level fluctuation reference setting value.

[0085] The liquid level fluctuation deviation of each ingot is statistically analyzed, and the liquid level fluctuation data is pre-processed using the Laida criterion. The error data outside (μ-3σ, μ+3σ) is filtered to prevent the judgment distortion caused by the deviation of the detection system.

[0086] in:

[0087]

[0088] Step 3: Through the analysis of big data on the production site, production experience and user quality requirements, the liquid level fluctuation deviation of a single ingot is graded into five levels: A+, A, B, C and D. A+ is the highest quality level, and the probability of subsequent slag roll quality defects caused by liquid level fluctuation is the lowest; A is a good quality level with a low probability of quality defects; B is a poor quality level, which can be used for low-end contract delivery, or upgraded to A through flame cleaning; C is poor quality and must be flame cleaned before being used for low-end contract delivery; D indicates that the ingot has been downgraded.

[0089] The A+ level means that the number of ±5mm peaks and troughs appearing on a single ingot is 0; the A level means that the number of ±5mm peaks and troughs appearing on a single ingot does not exceed 5 times; the B level means that the number of peaks and troughs appearing ≥±5mm and <±10mm is greater than 5 times, or the number of peaks and troughs appearing ≥±10mm is less than 5 times; the C level means that the number of peaks and troughs appearing ≥±10mm and <±15mm is greater than 5 times, or the number of peaks and troughs appearing ≥±15mm is less than 5 times; the D level means that the number of peaks and troughs appearing ≥±15mm and <±20mm is greater than 5 times, or the peaks or troughs of ≥±20mm appear.

[0090] For the determination of the wave peak, the present invention mainly adopts the method that the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold value of +5mm, +10mm, +15mm, +20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not lower than the determination threshold value, the computer does not stop determining; when the actual value of the liquid level fluctuation is lower than the determination threshold value, the computer stops determining and accumulates the number of wave peaks corresponding to the threshold value.

[0091] For the determination of the trough, the present invention mainly adopts the following method: the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold value of -5mm, -10mm, -15mm, -20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not higher than the determination threshold value, the computer does not stop determining; when the actual value of the liquid level fluctuation is higher than the determination threshold value, the computer stops determining and accumulates the number of troughs corresponding to the threshold value.

[0092] When the computer determines the total cumulative number of peaks and troughs that appear when the block billet corresponds to the threshold values ​​of ±5mm, ±10mm, ±15mm, and ±20mm, the following judgment will be made: First, when the total cumulative number of peaks and troughs that appear ≥±15mm and <±20mm is greater than 5 times, or the total cumulative number of peaks or troughs that appear ≥±20mm is greater than 0 times, it is judged as Class D; otherwise, when the total cumulative number of peaks and troughs that appear ≥±10mm and <±15mm is greater than 0 times, it is judged as Class D. If the total cumulative number of peaks and troughs ≥±15mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±15mm is less than 5 times, it is judged as Class C; otherwise, when the total cumulative number of peaks and troughs ≥±5mm and <±10mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±10mm is less than 5 times, it is judged as Class B; when the total cumulative number of ±5mm peaks and troughs on a single ingot does not exceed 5 times, it is judged as Class A; if none of the above situations occur, it is judged as Class A+.

[0093] Example:

[0094] A steel plant has a 1-machine 2-stream slab continuous casting machine, producing 1 furnace and 8 billets, including 4 in the 1st stream and 4 in the 2nd stream. The computer calculated the billet length of the 1st stream to be 41.43m and the 2nd stream to be 40.21m, and the actual length of the 1st stream is 41.5m and the 2nd stream is 40.01m. The computer uses a dimensionless model to achieve the correspondence between the lengths of the two-stream billets, thereby obtaining the correspondence between the billets and the liquid level fluctuation values. The production furnace liquid level fluctuation reference value is 90mm. The computer calculates the liquid level fluctuation deviation value by the difference between the actual value and the reference value. The liquid level fluctuation deviation value is processed by the Laida criterion, and finally a total of 5433 liquid level fluctuation deviation values ​​of the 1st and 2nd streams corresponding to the billet positions are obtained.

[0095] The computer classifies the liquid level fluctuation state of the ingot by using the peak and trough judgment criteria, and ultimately determines the liquid level fluctuation status of stream 1 and stream 2.

[0096] Internal steel grade number 1 flow billet number Slab grading 2 stream billet number Slab grading BG0120E8 22184381010 D 22184381510 B BG0120E8 22184381020 A+ 22184381520 C BG0120E8 22184381030 A 22184381530 A BG0120E8 22184381040 A 22184381540 B

Claims

1. A method for determining and grading the liquid level fluctuation of a slab crystallizer with high surface requirements, characterized in that The following steps are involved: (1) A billet length model is established by measuring the continuous casting speed and the liquid level fluctuation measurement time of the crystallizer, the liquid level fluctuation deviation under the theoretical billet length is determined by the liquid level fluctuation time, and a corresponding model between the liquid level fluctuation deviation and the billet length is established; (2) By matching the actual billet length with the billet drawing length model, the one-to-one correspondence between the liquid level fluctuation deviation data and the actual billet position is achieved, thus realizing the point and line connection between time and space in the production process; (3) The actual length of the slab is cut to statistically analyze the liquid level fluctuation deviation of each slab, and the Laida criterion is used to pre-process the liquid level fluctuation data; (4) Through the analysis of big data on the production site, production experience and user quality requirements, the liquid level fluctuation deviation of a single ingot is analyzed; In the step (1), the calculation formula of the casting length model is: Where: Δt is the time interval between the adjacent time periods collected by the system, Δt = t i -t i-1 ; V i t i The continuous casting machine pulling speed at the moment; n is the number of samples collected from the start of the continuous casting to the end of the casting, and the data collection interval is 300 milliseconds; The correspondence between the billet drawing length model and the billet length refers to the dimensionless processing of the calculated length and the actual length data, and finally the liquid level fluctuation deviation data of a single billet is obtained. i The actual billet position corresponding to the moment is: Where: L X is the actual length of the billet; L is the theoretical calculated length of the billet; L S L is the actual length of the billet produced; i is the theoretical calculated length corresponding to it; Therefore, the corresponding relationship between the length of the casting billet and the liquid level fluctuation deviation is realized, that is: YL X =YL i Where: YL X YL is the liquid level fluctuation deviation at the actual length X of the billet; i is the theoretical length Li of the ingot corresponding to the actual length Lx of the ingot, and the liquid level fluctuation deviation at the corresponding moment; In the step (4), the liquid level fluctuation deviation of a single ingot is divided into five levels: A+, A, B, C and D, wherein A+ is the highest quality level, and the probability of subsequent slag roll quality defects caused by liquid level fluctuation is the lowest; A is a good quality level, and the probability of quality defects is low; B is a poor quality level, which can be used for low-end contract delivery, or upgraded to A through flame cleaning; C is poor quality, and must be flame cleaned before being used for low-end contract delivery; D indicates that the ingot is downgraded; The A+ level means that the number of ±5mm peaks and troughs on a single ingot is 0; the A level means that the number of ±5mm peaks and troughs on a single ingot does not exceed 5 times; the B level means that the number of peaks and troughs ≥±5mm and <±10mm is greater than 5 times, or the number of peaks and troughs ≥±10mm is less than 5 times; the C level means that the number of peaks and troughs ≥±10mm and <±15mm is greater than 5 times, or the number of peaks and troughs ≥±15mm is less than 5 times; the D level means that the number of peaks and troughs ≥±15mm and <±20mm is greater than 5 times, or the peaks or troughs ≥±20mm appear; For the determination of the wave crest, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of +5mm, +10mm, +15mm and +20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not lower than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is lower than the determination threshold value, the computer stops determining and accumulates the number of wave crests corresponding to the threshold value. For the determination of troughs, the computer cyclically scans the actual value of the liquid level fluctuation. When the absolute value of the actual value of the liquid level fluctuation reaches the determination threshold values ​​of -5mm, -10mm, -15mm and -20mm, the computer starts to determine. When the actual value of the liquid level fluctuation is not higher than the determination threshold value, the computer does not stop determining. When the actual value of the liquid level fluctuation is higher than the determination threshold value, the computer stops determining and accumulates the number of troughs corresponding to the threshold value. When the computer determines the total cumulative number of peaks and troughs that appear when the block billet corresponds to the threshold values ​​of ±5mm, ±10mm, ±15mm and ±20mm respectively, it will make the following judgments: First, when the total cumulative number of peaks and troughs that appear ≥±15mm and <±20mm is greater than 5 times, or the total cumulative number of peaks or troughs that appear ≥±20mm is greater than 0 times, it is judged as Class D; otherwise, when the total cumulative number of peaks and troughs that appear ≥±10mm and <±15mm is greater than 0 times, it is judged as Class D. If the total cumulative number of peaks and troughs ≥±15mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±15mm is less than 5 times, it is judged as Class C; otherwise, when the total cumulative number of peaks and troughs ≥±5mm and <±10mm is greater than 5 times, or the total cumulative number of peaks and troughs ≥±10mm is less than 5 times, it is judged as Class B; when the total cumulative number of ±5mm peaks and troughs on a single ingot does not exceed 5 times, it is judged as Class A; if none of the above situations occur, it is judged as Class A+.

2. The method for determining and grading the liquid level fluctuation of a slab crystallizer with high surface requirements according to claim 1, characterized in that: In the step (2), the liquid level fluctuation deviation under the theoretical casting length refers to t i The difference between the liquid level fluctuation value and the reference set value at the moment is calculated by the billet drawing length model to achieve the correspondence between the fluctuation deviation and the actual billet position, that is: AND Li =And i -to Where: Y Li Theoretically calculated t i The liquid level fluctuation deviation corresponding to the length of the billet at the moment; Y i t i The liquid level fluctuation value collected at every moment; a is the liquid level fluctuation reference setting value.

3. The method for determining and grading the liquid level fluctuation of a slab crystallizer with high surface requirements according to claim 1, characterized in that: In step (3), the liquid level fluctuation data is pre-processed using the Laida criterion, and the error data outside (μ-3σ, μ+3σ) is filtered to prevent the detection system deviation from causing judgment distortion, wherein:

Images