A prediction method for transient abnormal fluctuations in the liquid level of a crystallizer

Abstract

The present invention provides a method for predicting transient abnormal fluctuations in a crystallizer liquid level, and relates to the field of crystallizer technology. The method first collects liquid level fluctuation data and production process data of a slab continuous casting crystallizer under different process conditions in real time; then, by performing spectrum analysis on the liquid level fluctuation data, it is determined whether the bulge generated during continuous casting production has a significant impact on the transient abnormal fluctuations in the crystallizer liquid level; then, by analyzing the time-frequency characteristics of the production process data, it is determined the correlation between the stopper rod position and the transient abnormal fluctuations in the crystallizer liquid level, and further mining the high-frequency region time-frequency characteristics of the stopper rod position to predict the occurrence of transient abnormal fluctuations in the crystallizer liquid level. The method performs continuous wavelet transform analysis on the crystallizer liquid level fluctuation data and the stopper rod position data, clarifies the correlation between the two, and conducts in-depth mining of the high-frequency region characteristics, summarizing the stopper rod position changes before the transient abnormal fluctuations in the crystallizer liquid level, so as to achieve accurate prediction of the abnormal fluctuations in the crystallizer liquid level.

Description

Technical Field

[0001] The present invention relates to the technical field of crystallizers, and in particular to a method for predicting abnormal instantaneous fluctuations of a crystallizer liquid level. Background Art

[0002] The mold, known as the "heart" of continuous casting, is the starting point for controlling the purity and quality of the ingot. Statistics show that 80% of surface defects in ingots originate in the mold. Fluctuations in the mold's molten steel level, a crucial physical phenomenon in the continuous casting process, significantly impact the quality of the resulting ingot.

[0003] Abnormal fluctuations in the mold liquid level not only easily cause slag entrainment, increasing the inclusion content near the primary solidified shell and affecting the uniform growth of the primary shell, but also affect the lubrication properties of the mold slag. In severe cases, this can cause surface cracks on the continuous casting billet and even steel leakage accidents. Currently, the main prediction targets are mainly periodic abnormal liquid level fluctuations, and there is little research on non-periodic instantaneous abnormal liquid level fluctuations. In addition, the research methods of the above studies mainly focus on the time domain. Due to the complex relationship between the various processes in the crystallizer, many important phenomena in the crystallizer are subject to significant limitations in obtaining information in the time domain. With the rapid development of wavelet analysis technology in recent years, it has had a strong impact on both ancient natural sciences and emerging high-tech applied technology disciplines, and has profound theoretical significance and a wide range of applications. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to address the deficiencies of the above-mentioned prior art and provide a method for predicting instantaneous abnormal fluctuations of the crystallizer liquid level, thereby realizing the prediction of instantaneous abnormal fluctuations of the crystallizer liquid level.

[0005] To solve the above technical problems, the present invention adopts the following technical solutions: a method for predicting abnormal transient fluctuations in the mold liquid level, which collects liquid level fluctuation data and production process data of a slab continuous casting mold under different process conditions in real time; performs spectrum analysis on the liquid level fluctuation data to determine whether bulging generated during continuous casting production has a significant impact on abnormal transient fluctuations in the mold liquid level; and determines the correlation between the stopper rod position and abnormal transient fluctuations in the mold liquid level by analyzing the time-frequency characteristics of the production process data. The method further explores the time-frequency characteristics of the stopper rod position in the high-frequency region to predict the occurrence of abnormal transient fluctuations in the mold liquid level. The method specifically comprises the following steps:

[0006] Step 1: Real-time acquisition of liquid level fluctuation data of a slab continuous casting mold at a continuous casting production site, as well as corresponding production process data and roll spacing data at the continuous casting production site; the production process data includes stopper position data, tundish tonnage, argon flow at the stopper, argon pressure at the stopper, argon flow at the water inlet, argon pressure at the water inlet, argon flow at the slide plate, and argon pressure at the slide plate; Step 2: Preprocessing the collected mold liquid level fluctuation data, and performing a fast Fourier transform on the preprocessed data to obtain frequency domain characteristics of the mold liquid level fluctuation when transient abnormal fluctuations are present, and frequency domain characteristics when transient abnormal fluctuations are absent;

[0007] Step 3: Calculate the frequency of the effect of bulging on the abnormal fluctuation of the mold liquid level, and determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level;

[0008] Using the collected roll spacing data from the continuous casting production site, the wavelength of the bulge caused by the gap between the rolls at different locations was calculated. The wavelengths of the bulge at different locations were directly superimposed and calculated to obtain the frequency of the bulge's effect on abnormal fluctuations in the mold liquid level.

[0009] Compare the calculated bulging frequency with the frequency domain characteristics of the abnormal fluctuation of the mold liquid level calculated in step 2 to determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level;

[0010] The abnormal fluctuation of the mold liquid level caused by unstable bulging can be calculated by superimposing the volume changes of the shell caused by rollers at different positions. The frequency f of the effect of the bulge at each position on the abnormal fluctuation of the mold liquid level is calculated by the following formula:

[0011]

[0012] Among them, v c is the pulling speed, λ is the wavelength of the bulge;

[0013] The bulge wavelength λ at different positions is calculated by the following formula:

[0014]

[0015] Where p is the roller spacing, n is the serial number of the rollers at different positions, n = 1, 2, 3, ..., 10;

[0016] Step 4: Conduct frequency domain coherence tests on different production process data and mold level fluctuation data to determine the core factors affecting mold level fluctuation;

[0017] Frequency domain coherence is used to verify the relationship between two random signals or data and is calculated as follows:

[0018]

[0019] Among them, C xy (ω) is the coherence coefficient of the random signals x(t) and y(t), ω is the frequency component of x(t), S xy (ω) is the cross power spectral density of two random signals x(t) and y(t), S xx (ω) and S yy (ω) are the autopower spectral densities of signals x(t) and y(t), respectively;

[0020] According to formula (3), the range of the coherence coefficient is:

[0021] 0≤C xy (ω)≤1 (4)

[0022] Frequency domain coherence is an indicator that measures the phase consistency of two signals at different frequencies, and its value range is 0 to 1. When the coherence coefficient is 1, it indicates that the two signals have a linear response in the corresponding frequency band; when the coherence coefficient is 0, it indicates that the two signals are incoherent within a certain frequency band, that is, the two signals are completely independent.

[0023] The cross power spectral density of two random signals x(t) and y(t) is shown below:

[0024]

[0025] Among them, T represents the time range, F x (ω,T),F y (ω, T) are the continuous Fourier transforms of signals x(t) and y(t), * represents conjugate, and E represents mathematical expectation; F x (ω,T) and F y (ω,T) is calculated by the following formula:

[0026]

[0027]

[0028] Step 5: Perform continuous wavelet transform analysis on the mold level fluctuation data to obtain the time-frequency characteristic diagram of the mold level fluctuation;

[0029] Step 6: Perform continuous wavelet transform analysis on the stopper rod position data in the production process data to obtain a high-frequency region time-frequency characteristic diagram of the stopper rod position;

[0030] Step 7: Summarize, classify, and compare the time-frequency characteristics of the mold liquid level fluctuation obtained in steps 5 and 6 with the time-frequency characteristics of the stopper rod position in the high-frequency region, and perform mathematical fitting on different comparison results to observe the changing trends and amplitudes of the time-frequency characteristics of the mold liquid level fluctuation and the stopper rod position in the high-frequency region, and summarize the change pattern of the stopper rod position before the instantaneous abnormal fluctuation of the mold liquid level; and use the summarized change pattern of the stopper rod position to predict the instantaneous abnormal fluctuation of the mold liquid level.

[0031] The beneficial effects of the above technical solution are as follows: the present invention provides a method for predicting abnormal transient fluctuations in the mold liquid level, combining time domain features with frequency domain features. By performing fast Fourier transform analysis on the mold liquid level fluctuation data collected on site, the abnormal fluctuation characteristics are highlighted, the frequency of the effect of bulging on the abnormal transient fluctuations in the mold liquid level is calculated, and the degree of its influence is clarified. The core influencing factors of the abnormal transient fluctuations in the mold liquid level are determined using the coherence test method, and the interference of other influencing factors is eliminated. The mold liquid level fluctuation data and the stopper rod position data are analyzed by continuous wavelet transform to clarify the relationship between the two and conduct in-depth exploration of the high-frequency region features. The stopper rod position changes before the abnormal transient fluctuations in the mold liquid level are summarized to achieve accurate prediction of abnormal fluctuations in the mold liquid level. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A flow chart of a method for predicting abnormal transient fluctuations in a crystallizer liquid level provided by an embodiment of the present invention;

[0033] Figure 2 Frequency domain diagram of the crystallizer liquid level fluctuation provided by an embodiment of the present invention, where (a) indicates abnormal fluctuation and (b) indicates no abnormal fluctuation.

[0034] Figure 3 A schematic diagram of the frequency of belly bulging provided by an embodiment of the present invention;

[0035] Figure 4 Figure 1 shows the characteristic diagram of abnormal transient fluctuation of the liquid level in a carbon steel mold provided by an embodiment of the present invention, wherein (a) is a time-frequency diagram of the liquid level fluctuation in the mold, (b) is a CWT diagram of the liquid level fluctuation in the mold, (c) is a time-frequency diagram of the stopper rod position, and (d) is a CWT diagram of the stopper rod position.

[0036] Figure 5 A diagram of CWT coefficients in high-frequency areas of stopper rod positions of different steel grades provided in an embodiment of the present invention;

[0037] Figure 6 This is a diagram of the CWT coefficients in the high-frequency region for different acceleration plug positions provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0039] This embodiment takes the slab continuous casting process of a steel plant as an example, and uses the prediction method of the instantaneous abnormal fluctuation of the crystallizer liquid level of the present invention to predict the instantaneous abnormal fluctuation of the crystallizer liquid level in the continuous casting process.

[0040] In this embodiment, a method for predicting abnormal instantaneous fluctuations of the crystallizer liquid level is provided. The method collects the liquid level fluctuation data and production process data of the slab continuous casting crystallizer of the plant under different process conditions (steel composition and casting speed) in real time. By performing spectrum analysis on the liquid level fluctuation data, it is determined whether the bulge generated in the continuous casting production has a significant impact on the abnormal instantaneous fluctuations of the crystallizer liquid level. By analyzing the time-frequency characteristics of the production process data, the correlation between the stopper rod position and the abnormal instantaneous fluctuations of the crystallizer liquid level is determined. The time-frequency characteristics of the high-frequency region of the stopper rod position are further explored to predict the occurrence of abnormal instantaneous fluctuations of the crystallizer liquid level. Figure 1 As shown, the specific steps include:

[0041] Step 1: An electromagnetic liquid level sensor is used to collect real-time liquid level fluctuation data from the slab continuous casting mold at the continuous casting site, along with corresponding production process data and roll spacing data. This production process data includes stopper position data, tundish tonnage, argon flow and pressure at the stopper, argon flow and pressure at the nozzle, and argon flow and pressure at the slide. The electromagnetic liquid level sensor's measurement principle is to determine the mold liquid level based on the interaction between the high-frequency magnetic field generated by the coil in the detection and tracking detector and the mold liquid level fluctuations. The electromagnetic liquid level sensor is fixed to the back plate of the mold, directly behind the submerged nozzle and near the meniscus.

[0042] Step 2: The collected crystallizer liquid level fluctuation data is preprocessed, and the preprocessed data is fast Fourier transformed to obtain the frequency domain characteristics of the crystallizer liquid level fluctuation when there is transient abnormal fluctuation, and the frequency domain characteristics when there is no transient abnormal fluctuation. The characteristics of transient abnormal fluctuation of the crystallizer liquid level are more obvious, and clearer information is obtained.

[0043] Step 3: Calculate the frequency of the effect of bulging on the abnormal fluctuation of the mold liquid level, and determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level;

[0044] Using the collected roll spacing data from the continuous casting production site, the wavelength of the bulge caused by the gap between the rolls at different locations was calculated. The wavelengths of the bulge at different locations were directly superimposed and calculated to obtain the frequency of the bulge's effect on abnormal fluctuations in the mold liquid level.

[0045] Compare the calculated bulging frequency with the frequency domain characteristics of the abnormal fluctuation of the mold liquid level calculated in step 2 to determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level;

[0046] The abnormal fluctuation of the mold liquid level caused by unstable bulging can be calculated by superimposing the volume changes of the shell caused by rollers at different positions. The frequency f of the effect of the bulge at each position on the abnormal fluctuation of the mold liquid level is calculated by the following formula:

[0047]

[0048] Among them, v c is the pulling speed, λ is the wavelength of the bulge;

[0049] The bulge wavelength λ at different positions is calculated by the following formula:

[0050]

[0051] Where p is the roller spacing, n is the serial number of the rollers at different positions, n = 1, 2, 3, ..., 10;

[0052] Step 4: Conduct frequency domain coherence tests on different production process data and mold level fluctuation data to determine the core influencing factors affecting mold level fluctuation;

[0053] Frequency domain coherence is used to verify the relationship between two random signals or data and is calculated as follows:

[0054]

[0055] Among them, C xy (ω) is the coherence coefficient of the random signals x(t) and y(t), ω is the frequency component of x(t), S xy (ω) is the cross power spectral density of two random signals x(t) and y(t), S xx (ω) and S yy (ω) are the autopower spectral densities of signals x(t) and y(t), respectively;

[0056] According to formula (3), the range of the coherence coefficient is:

[0057] 0≤C xy (ω)≤1 (4)

[0058] Frequency domain coherence is an indicator that measures the phase consistency of two signals at different frequencies, and its value range is 0 to 1. When the coherence coefficient is 1, it indicates that the two signals have a linear response in the corresponding frequency band; when the coherence coefficient is 0, it indicates that the two signals are incoherent within a certain frequency band, that is, the two signals are completely independent.

[0059] The cross power spectral density of two random signals x(t) and y(t) is shown below:

[0060]

[0061] Among them, T represents the time range, F x (ω,T),F y (ω, T) are the continuous Fourier transforms of signals x(t) and y(t), * represents conjugate, and E represents mathematical expectation; F x (ω,T) and F y (ω,T) is calculated by the following formula:

[0062]

[0063]

[0064] Step 5: Perform continuous wavelet transform analysis on the mold level fluctuation data to obtain the time-frequency characteristic diagram of the mold level fluctuation;

[0065] For a random signal x(t), the continuous wavelet transform (CWT) is defined as:

[0066] W x (a,b)=∫x(t)Ψ a,b (t)dt (8)

[0067] Among them, W x (a,b) is the wavelet transform WT of the signal x(t), a is the scale factor, b is the position factor, Ψ a,b (t) is the wavelet basis function, which is obtained by expanding and shifting a single prototype wavelet Ψ(t) called the "mother wavelet":

[0068]

[0069] Step 6: Perform continuous wavelet transform analysis on the stopper rod position data in the production process data to obtain a high-frequency region time-frequency characteristic diagram of the stopper rod position;

[0070] Step 7: Conduct in-depth data mining on the time-frequency characteristics of the mold liquid level fluctuation and the high-frequency time-frequency characteristics of the stopper rod position obtained in steps 5 and 6, that is, summarize, classify, and compare the time-frequency characteristics of the mold liquid level fluctuation and the high-frequency time-frequency characteristics of the stopper rod position, and perform mathematical fitting on different comparison results to observe the changing trends and amplitudes of the mold liquid level fluctuation and the high-frequency time-frequency characteristics of the stopper rod position, summarize the stopper rod position change law before the instantaneous abnormal fluctuation of the mold liquid level; and use the summarized stopper rod position change law to predict the instantaneous abnormal fluctuation of the mold liquid level.

[0071] This embodiment collects data at the continuous casting site, and uses an electromagnetic liquid level sensor to collect real-time data on the liquid level fluctuation of the on-site slab continuous casting crystallizer. The electromagnetic liquid level sensor is produced by the Czech VUHZ company, and the sensor model is SH-D. In this embodiment, the size of the crystallizer is approximately (910–1300) mm×210 mm×900 mm, and the size of the electromagnetic liquid level sensor is 540 mm×150 mm×63 mm. The sensor is fixed on the back plate of the crystallizer, directly behind the immersion water nozzle, close to the meniscus. The measurement area is approximately 600 mm wide, 80 mm high, and has a response time of 0.1 s. In addition, this embodiment also collects production process data under different steel grades and different drawing speeds in real time, with a sampling time of 0.1 s.

[0072] The fast Fourier transform method is used to convert time domain features into frequency domain features, such as Figure 2 In order to find the influence range of the instantaneous abnormal fluctuation of the crystallizer liquid level in the frequency domain, the abnormal liquid level fluctuation data in the original liquid level fluctuation data is removed, and then the fast Fourier transform method is used to convert it into the frequency domain, as shown in Figure 2. Figure 2 (b) is shown. Figure 2 It can be seen that after removing the instantaneous fluctuation data of the crystallizer liquid level from the data, the peak amplitude in the low-frequency area of ​​the frequency domain diagram is significantly reduced, indicating that although the number of abnormal fluctuation data is small, it has a very obvious impact on the overall frequency domain distribution of the liquid level fluctuation, and this impact is mainly concentrated in the low-frequency area.

[0073] The wavelengths of bulges at different positions are directly superimposed for calculation. The final calculated frequency of the bulge on the abnormal fluctuation of the mold liquid level is 1.25Hz, as shown in Figure 3 As shown. Figure 3 It can be seen that the amplitude corresponding to 1.25 Hz is very low, indicating that the influence of bulging on the abnormal fluctuation of the crystallizer liquid level is very small, so it is not considered.

[0074] The frequency domain coherence test was performed on different process data and the crystallizer liquid level fluctuation data. When the casting speed was stable, the stopper rod position had a significant impact on the instantaneous abnormal fluctuation of the crystallizer liquid level in the low-frequency region.

[0075] The time-frequency characteristics of the instantaneous abnormal mold level fluctuation and stopper position under different conditions were studied using continuous wavelet transform. Figure 4 (a) shows the mold level fluctuation curve of medium carbon steel at a pulling speed of 1.2m / min. Figure 4 In (a), the crystallizer liquid level fluctuation consists of two parts: one is a stable change, and the other is an instantaneous abnormal fluctuation. Figure 4 The CWT results of the crystallizer level fluctuation in (a) are as follows Figure 4(b) The horizontal axis represents the sampling point, and the vertical axis represents the frequency, which is the main frequency of the wavelet at different scales. Three obvious areas are marked in the figure, corresponding to the rapid generation and disappearance of transient abnormal fluctuations in the crystallizer liquid level. At the 45th, 228th, and 515th sampling points, the CWT coefficients are 3.88, 3.40, and 6.61, respectively, while Figure 4 As shown in (a), significant transient abnormal fluctuations in the mold level occurred at sampling points 45, 228, and 515. This indicates that the CWT method can clearly reflect the characteristics of transient abnormal fluctuations in the mold level. In addition, the light blue area (with a CWT coefficient ranging from 1.5 to 2.0) is distributed throughout the entire region, indicating that the mold level fluctuations during this period were disordered and relatively stable, with no significant abnormal fluctuations.

[0076] Subsequently, the CWT results of the stopper rod position in the high-frequency region were further mined to characterize the transient abnormal fluctuation of the mold liquid level. In the time-frequency domain characteristics of the stopper rod position, the CWT coefficients of the high-frequency region of different steel grades from 20s to 10s (the 200th to 100th sampling points) before the peak of the transient abnormal fluctuation of the mold liquid level appeared were collected to further explore its change trend. Figure 5 As shown in the figure, before the instantaneous abnormal fluctuation of the crystallizer liquid level, the CWT coefficient of the high-frequency area of ​​the stopper rod position of different steel grades has an upward trend, and the growth rate of the CWT coefficient is very close in the same range. The CWT coefficient growth rate of peritectic steel is the lowest, which is 0.114 / s; the CWT coefficient growth rate of low carbon steel is the highest, which is 0.169 / s.

[0077] It can be seen that in the range of 20s-10s before the instantaneous abnormal fluctuation of the crystallizer liquid level, the CWT coefficient in the high-frequency area of ​​the stopper position has an upward trend, and the growth rate is in the range of 0.1 / s-0.2 / s.

[0078] This example also compares the predictions of abnormal instantaneous fluctuations in the mold liquid level at different casting speeds.

[0079] Figure 6 is the CWT coefficient of the high frequency area of ​​the plug rod position at different drawing speeds for low carbon steel. Figure 6 As can be seen from the figure, 20 to 10 seconds before the instantaneous abnormal fluctuation of the mold level, the CWT coefficient in the high-frequency region of the stopper position shows an upward trend at different casting speeds. Within the same range, the CWT coefficient growth rate is very similar, with the lowest CWT coefficient growth rate of 1.06 / s at a casting speed of 1.3 m / min and the highest CWT coefficient growth rate of 1.76 / s at a casting speed of 1.0 m / min. When the mold level fluctuation is relatively stable and no abnormal transient fluctuation occurs, the CWT coefficient in the high-frequency region of the stopper position does not show an upward trend.

[0080] It can be seen that within the first 20 to 10 seconds of an abnormal transient mold level fluctuation, the CWT coefficient in the high-frequency region of the stopper position shows an upward trend, with a growth rate ranging from 0.1 / s to 0.2 / s. In the absence of abnormal transient mold level fluctuations, the CWT coefficient does not show an upward trend and changes relatively steadily. Therefore, the trend of the CWT coefficient in the high-frequency region of the stopper position can be used to predict the occurrence of abnormal transient mold level fluctuations in advance.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A method for predicting abnormal transient fluctuations in a crystallizer liquid level, characterized by: Real-time collection of liquid level fluctuation data and production process data of slab continuous casting molds under different process conditions; By performing spectrum analysis on liquid level fluctuation data, it is determined whether the bulge generated during continuous casting production has a significant impact on the instantaneous abnormal fluctuation of the mold liquid level; By analyzing the time-frequency characteristics of production process data, the correlation between the stopper rod position and abnormal instantaneous fluctuations in the crystallizer liquid level is determined. The time-frequency characteristics of the high-frequency region of the stopper rod position are further exploited to predict the occurrence of abnormal instantaneous fluctuations in the crystallizer liquid level. The following steps are involved: Step 1: Real-time data collection of liquid level fluctuations in the slab continuous casting mold at the continuous casting production site is collected, along with corresponding production process data and roll spacing data at the continuous casting production site; Step 2: Preprocess the collected mold level fluctuation data and perform fast Fourier transform on the preprocessed data to obtain the frequency domain characteristics of the mold level fluctuation when there is transient abnormal fluctuation and the frequency domain characteristics when there is no transient abnormal fluctuation; Step 3: Calculate the frequency of the effect of bulging on the abnormal fluctuation of the mold liquid level, and determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level; Step 4: Conduct frequency domain coherence tests on different production process data and mold level fluctuation data to determine the core factors affecting mold level fluctuation; Step 5: Perform continuous wavelet transform analysis on the mold level fluctuation data to obtain the time-frequency characteristic diagram of the mold level fluctuation; Step 6: Perform continuous wavelet transform analysis on the stopper rod position data in the production process data to obtain a high-frequency region time-frequency characteristic diagram of the stopper rod position; Step 7: Perform data mining on the time-frequency characteristics of the mold liquid level fluctuation and the high-frequency area time-frequency characteristics of the stopper rod position, summarize the change pattern of the stopper rod position before the instantaneous abnormal fluctuation of the mold liquid level; and use the summarized change pattern of the stopper rod position to predict the instantaneous abnormal fluctuation of the mold liquid level.

2. The method for predicting transient abnormal fluctuations of a crystallizer liquid level according to claim 1, characterized in that: The step 1: using an electromagnetic liquid level sensor to collect liquid level fluctuation data of a slab continuous casting crystallizer in a continuous casting production site in real time.

3. The method for predicting transient abnormal fluctuations of a crystallizer liquid level according to claim 1, wherein: The production process data includes stopper rod position data, tundish tonnage, argon flow at the stopper rod, argon pressure at the stopper rod, argon flow at the nozzle, argon pressure at the nozzle, argon flow at the slide plate, and argon pressure at the slide plate.

4. The method for predicting abnormal instantaneous fluctuation of a crystallizer liquid level according to claim 1, characterized in that: The specific method of step 3 is: Using the collected roll spacing data from the continuous casting production site, the wavelength of the bulge caused by the gap between the rolls at different locations was calculated. The wavelengths of the bulge at different locations were directly superimposed and calculated to obtain the frequency of the bulge's effect on abnormal fluctuations in the mold liquid level. Compare the calculated bulging frequency with the frequency domain characteristics of the abnormal fluctuation of the mold liquid level calculated in step 2 to determine the impact of bulging on the instantaneous abnormal fluctuation of the mold liquid level; The abnormal fluctuation of the mold liquid level caused by unstable bulging can be calculated by superimposing the volume changes of the shell caused by rollers at different positions. The frequency f of the effect of the bulge at each position on the abnormal fluctuation of the mold liquid level is calculated by the following formula: Among them, v c is the pulling speed, λ is the wavelength of the bulge; The bulge wavelength λ at different positions is calculated by the following formula: Wherein, p is the roller spacing, n is the serial number of the rollers at different positions, n=1, 2, 3, ..., 10.

5. The method for predicting abnormal instantaneous fluctuation of the mold liquid level according to claim 1, characterized in that: The frequency domain coherence is used to verify the relationship between two random signals or data and is calculated by the following formula: Among them, C xy (ω) is the coherence coefficient of the random signals x(t) and y(t), ω is the frequency component of x(t), S xy (ω) is the cross power spectral density of two random signals x(t) and y(t), S xx (ω) and S yy (ω) are the autopower spectral densities of signals x(t) and y(t), respectively; According to formula (3), the range of the coherence coefficient is: 0≤C xy (ω)≤1 (4) Frequency domain coherence is an indicator that measures the phase consistency of two signals at different frequencies, and its value range is 0 to 1. When the coherence coefficient is 1, it indicates that the two signals have a linear response in the corresponding frequency band; when the coherence coefficient is 0, it indicates that the two signals are incoherent within a certain frequency band, that is, the two signals are completely independent. The cross power spectral density of two random signals x(t) and y(t) is shown below: Among them, T represents the time range, F x (ω,T),F y (ω, T) are the continuous Fourier transforms of signals x(t) and y(t), * represents conjugate, and E represents mathematical expectation; F x (ω,T) and F y (ω,T) is calculated by the following formula:

6. The method for predicting abnormal instantaneous fluctuation of the mold liquid level according to claim 1, characterized in that: The data mining described in step 7 is to summarize, classify, and compare the time-frequency characteristics of the crystallizer liquid level fluctuation and the time-frequency characteristics of the stopper rod position in the high-frequency area, and perform mathematical fitting on different comparison results to observe the changing trends and amplitudes of the time-frequency characteristics of the crystallizer liquid level fluctuation and the stopper rod position in the high-frequency area, and then summarize the change pattern of the stopper rod position before the instantaneous abnormal fluctuation of the crystallizer liquid level.

Images