Steel ladle refining monitoring method based on vibration measurement

Abstract

The invention discloses a steel ladle refining monitoring method based on vibration measurement, at least two vibration sensors are respectively arranged at the distance and angle symmetrical positions of a steel ladle, and the monitoring method comprises the steps of data acquisition, vibration signal preprocessing, feature extraction, judgment analysis and the like. Whether the steel ladle stirring is uniform or not is judged according to the analysis of the vibration quantity (a first characteristic value) and the dispersion degree (a second characteristic value) of the vibration sequence detected by the symmetrically mounted vibration sensors, the method is simple and easy to implement, the precise monitoring of the steel ladle refining with different quality requirements can be realized, the steel ladle refining stirring efficiency can be improved, and the production cost can be reduced. And the rejection rate of ladle refining products can also be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of steel smelting and control, and particularly relates to a ladle refining monitoring method based on vibration measurement. Background Art

[0002] The ladle degassing refining method is the main method of secondary steelmaking (also known as secondary steelmaking, simply referred to as ladle refining in the present invention), that is, the molten steel smelted in a primary smelting furnace (electric furnace and converter) is transferred to another high-temperature container (mainly a ladle) for refining. Its main classifications are: (1) ladle treatment type: such as argon blowing in ladle, powder injection in ladle, and vacuum circulation degassing (RH); (2) ladle refining type, such as vacuum oxygen decarburization (VOD), argon oxygen decarburization (AOD), ladle refining (LF, ASEA-CKF, VAD, etc.). Gas stirring is used to improve the mixing effect and promote chemical reactions, and the control of gas intensity (or flow rate, flow velocity) and time has a great influence on the entire process.

[0003] Since the ladle refining process is in a high-temperature state, it is difficult to obtain an image that effectively reflects the stirring effect at high temperature by general video image detection. Of course, even if a field image can be obtained, it is difficult to extract the stirring effect information. At present, there is still a lack of effective monitoring means and methods in the ladle refining process. It is mainly based on on-site operation experience to manually judge whether the stirring is uniform. In fact, most of the time, the stirring effect is ensured by experiencing sufficient stirring time at the empirical stirring intensity. That is to say, there is an over-stirring state in many cases, the refining efficiency is difficult to improve, the requirement for manual observation is high, and once unqualified products are detected afterwards, it will cause greater losses, and all the products cast in the whole furnace (ladle) may become waste products.

[0004] At present, although there are some research on methods for monitoring the ladle stirring state by using vibration measurement, due to the influence of many factors such as the ladle type (including the lining), molten steel height (or weight), shape of the gas blowing outlet, gas flow rate and flow, sensor installation method and installation position, environmental vibration, type and addition speed of additives, and other interference factors, it is difficult to find an effective characterization method, and there is still a distance from actual engineering application. For example, some studies have shown that although the ladle response value (LRV) can better describe the gas stirring intensity, there is no reliable information on whether the additives and other components are stirred evenly after being added. And affected by the above factors, the ladle response values at different times, different locations, and different objects are poorly comparable. To obtain effective empirical judgment data, a large number of experiments are required, and the inheritance is poor. Changes in environmental conditions will affect the availability of empirical data. Therefore, so far, this technology is still in the laboratory research stage, there is no monitoring device and method for engineering practical use, and there is an urgent need for a convenient and effective refining effect monitoring method for ladle degassing refining. Summary of the Invention

[0005] By collecting and analyzing the vibration information of symmetrically installed vibration sensors, the present invention rationally characterizes the stirring state of the blowing gas during the ladle refining process by using vibration characteristic values, thereby enabling effective monitoring of the ladle refining process.

[0006] To achieve the above object, the technical solution of the present invention is a ladle refining monitoring method based on vibration measurement. At least two vibration sensors are respectively arranged at symmetric positions of distance and angle of the ladle. The monitoring method includes the following steps: S1. Data acquisition: Using the sliding window method to acquire the vibration quantity of the symmetrically installed vibration sensors within the window duration 、 … Among them, is the sampling serial number, is the sensor number. The window sliding time interval is greater than the sampling period, and the sampling period is determined according to the vibration frequency, the characteristics of the sensor, and the data acquisition instrument (data acquisition box); S2. Vibration signal preprocessing, including filtering out obvious abnormal amplitude values and / or filtering out obvious non-stirring related frequency components; S3. Feature extraction: Extracting the first characteristic value corresponding to the vibration quantity 、 … and / or the second characteristic value 、 … , the first characteristic value describes the magnitude of the vibration quantity, and the second characteristic value describes the discrete degree of the vibration sequence; S4. Judgment and analysis, including judgment method one, judgment method two, or judgment method three; The judgment method one includes that if all within the window duration, j≠k (such as ), it is considered that the ladle stirring is uniform, otherwise, stirring needs to continue. Among them, is the first threshold; The judgment method two includes that if all within the window duration, j≠k (such as ), and all are satisfied (such as and etc.), it is considered that the ladle stirring is uniform, otherwise, stirring needs to continue. Among them, is the second threshold; The judgment method three includes simultaneously monitoring the blowing gas flow rate (also known as the flow velocity or intensity, and the working gas is generally argon) and the vibration quantity in real time, and calculating the ladle response value (LRV) within the window duration as the third characteristic value , … The ladle response value is the ratio of the area under the vibration quantity-time curve to the area under the flow rate-time curve (or the ratio of the integral of the vibration quantity with respect to time to the integral of the flow rate with respect to time within the window duration); According to the pre-calibrated ladle response value curve (i.e., the blowing intensity-vibration signal curve), obtain the nominal value of the strongest vibration signal intensity corresponding to the measured blowing flow rate If within the window duration , j≠k, and for any , then it is considered that this blowing and stirring has reached the optimal state and the stirring is uniform, where is the third threshold, is the fourth threshold.

[0007] Furthermore, the first eigenvalue includes at least one of the mean value, median value, and norm; the second eigenvalue includes at least one of the variance (RMS, root mean square value), standard deviation, maximum value, minimum value, and extreme value difference, and the extreme value difference is the difference between the maximum value and the minimum value.

[0008] Furthermore, the vibration quantity includes at least one of displacement, velocity, or acceleration in at least one of the x, y, and z directions (where x and y are the two horizontal directions and z is the vertical direction) (such as the acceleration in the x direction).

[0009] Furthermore, the lower limit of the window duration is not less than 3 times the longest vibration period duration, and the upper limit is not greater than the minimum lag duration. The lag duration includes the transition period duration before the system enters the steady state caused by the change of the ladle stirring state. The change of the ladle stirring state includes state changes such as the change of the blowing gas flow rate and the addition of additives.

[0010] Furthermore, the window duration includes 10s, 21s, or 40s.

[0011] Furthermore, the window sliding time interval includes an integer number of seconds (such as every 1 second or every 2 seconds, etc.) or a unit time greater than or equal to the longest vibration period or greater than or equal to an integer multiple of the sampling period.

[0012] Furthermore, the first threshold, the second threshold, the third threshold, and the fourth threshold all include quasi-dynamic thresholds, and the nominal value includes a quasi-dynamic nominal value. The determination method of the quasi-dynamic threshold or the quasi-dynamic nominal value includes that when the stirring time with a constant blowing flow rate exceeds the maximum value of the empirical duration, and the first eigenvalue, the second eigenvalue, or the third eigenvalue cannot meet the requirements of the corresponding Judgment Method 1, Judgment Method 2, or Judgment Method 3, the measured calculated value is used as the quasi-dynamic threshold or the quasi-dynamic nominal value of the corresponding threshold.

[0013] Further, the filtering of significantly abnormal amplitude values includes the wild value elimination method or the wild value smoothing method; the filtering of significantly non-stirring-related frequency components includes performing time-frequency transformation on the sampled time series to obtain a frequency-domain series, eliminating the frequency components in the frequency-domain series that are not within the stirring-related characteristic frequency range, and then performing inverse transformation on the remaining frequency-domain series to obtain a new time series. The time-frequency transformation includes discrete Fourier transformation and inverse transformation.

[0014] Further, the judgment and analysis further include the continuous window judgment method, which includes performing the judgment method 1, judgment method 2, or judgment method 3 on a number of consecutive windows. When the consecutive windows simultaneously meet the threshold conditions, it is considered that the stirring is uniform.

[0015] Further, the symmetric positions include 180-degree symmetry (2 vibration sensors), 90-degree symmetry (4 vibration sensors), and 60-degree symmetry (6 vibration sensors) with the central axis of the ladle as the axis of symmetry. Among them, the equal-distance symmetry relative to the axis of symmetry is the default.

[0016] The advantages and beneficial effects of the present invention are as follows: The ladle refining monitoring method based on vibration measurement proposed by the present invention enables the application of vibration monitoring in the ladle refining process to move from theory to practice. Through the analysis of the vibration quantity characteristics of symmetrically arranged vibration sensors, it is possible to timely and accurately judge whether the gas stirring in the ladle refining process is uniform, that is, whether the stirring effect is achieved; it is more reliable than manually using empirical flow rates and empirical times, avoiding excessive stirring or insufficient stirring, with higher efficiency, and is conducive to the control of ladle refining; it is more convenient to obtain empirical parameters, without excessive theoretical analysis and complex experiments, and the corresponding threshold parameter empirical values can be obtained in the normal stirring state. The application of the present invention can not only improve the stirring efficiency of ladle refining but also reduce the rejection rate of refined products. Description of the Drawings

[0017] Figure 1 is the flow chart of the ladle refining monitoring method based on vibration measurement; Figure 2 is the schematic diagram of the arrangement of ladle refining monitoring sensors.

[0018] In the figure: 1, ladle; 2, hanging ear; 3, trolley; 4, sensor; 5, blow pipe; 6, valve. Detailed Embodiments

[0019] The following combines the drawings and embodiments to further describe the detailed embodiments of the present invention. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0020] Theoretical analysis shows that vibration signals can effectively characterize the gas stirring intensity. However, due to the difficulty in finding the law between the two, practical engineering applications face difficulties.

[0021] Example 1: To better illustrate the application scenario of the present invention, as Figure 2 shown, there is a ladle refining monitoring device based on vibration measurement. The ladle 1 is placed on the trolley 3. Vibration sensors 4 are respectively installed on the sides of the two hanging ears 2 of the ladle. The blowing pipeline 5 blows gas into the ladle through the bottom of the ladle, and a controllable valve 6 is arranged on the blowing pipeline.

[0022] As Figure 1 shown, the present invention designs a ladle refining monitoring method based on vibration measurement. At least two vibration sensors are respectively arranged at the symmetric positions of the distance and angle of the ladle. The monitoring method includes the following steps: S1. Data acquisition: Use the sliding window method to acquire the vibration amounts of the symmetrically installed vibration sensors within the window duration , … wherein, is the sampling serial number, is the sensor number. The window sliding time interval is greater than the sampling period, and the sampling period is determined according to the vibration frequency, the sensor, and the characteristics of the data acquisition instrument (data acquisition box); S2. Preprocessing of vibration signals, including filtering out obvious abnormal amplitude values and / or filtering out obvious non-stirring related frequency components, extracting the vibration signals related to the stirring vibration frequency, and filtering out most of the environmental vibration signals; The method of filtering out obvious abnormal amplitude values can adopt the threshold judgment method (removing outliers beyond the threshold) and / or the filtering method (suppressing the amplitude of outliers); The stirring related frequency components are generally related to factors such as gas flow rate, ladle model, and molten steel capacity. Normally, it is between 10 Hz and 500 Hz. If various influencing factors are determined, this frequency range can be smaller, and even can be limited to several fixed point frequencies. Therefore, the frequency components outside this frequency range belong to non-stirring related frequency components (generally, the response frequency of vibration sensors can be between 1 Hz and 5000 Hz), and can be removed, so as to make the extraction of vibration amount characteristic values more accurate; S3. Feature extraction: Extract the first characteristic value corresponding to the vibration amount , … and / or the second characteristic value , … , where the first characteristic value describes the magnitude of the vibration amount, and the second characteristic value describes the discrete degree of the vibration sequence; In this example, both the first characteristic value and the second characteristic value are extracted; S4. Judgment and analysis, including Judgment Method 1, Judgment Method 2, or Judgment Method 3. In this embodiment, the method is mainly described by taking two vibration sensors as an example. For more than two sensors, pairwise comparison is carried out, and the basic method is essentially the same; The Judgment Method 1 includes that if all , j≠k (such as ), it is considered that the ladle stirring is uniform, otherwise, stirring needs to continue. Among them, is the first threshold; that is, when the ladle vibration transmission characteristics at symmetric positions are consistent and last for a period of time (window duration), it can be considered that the molten steel in the ladle has been stirred evenly; the first threshold is the allowable difference in the vibration quantity characteristic values in the symmetric direction under the evenly stirred state. The smaller the threshold, the more uniform the stirring, but the higher the requirements for the stirring time and the control of the gas stirring flow rate.

[0023] The Judgment Method 2 includes that if all , j≠k (such as ), and all are satisfied (such as and etc.), it is considered that the ladle stirring is uniform, otherwise, stirring needs to continue. Among them, is the second threshold; that is to say, in addition to judging the difference in the first characteristics of the symmetrically installed vibration sensors, Judgment Method 2 also judges whether the second characteristic values of the vibration sensors are within a certain range. For example, the second characteristic value takes the variance of the respective vibration quantities within the window duration. When the variance is small enough, it indicates that the local molten steel being stirred has entered a steady state. At the same time, if the different local vibration characteristics are basically the same, it can be determined that this stirring is uniform, which makes sense; the Judgment Method 2 is essentially a judgment method with higher quality requirements for steelmaking products; The Judgment Method 3 includes simultaneously and real-time monitoring the blowing gas flow rate (also known as the flow rate or intensity, and the working gas is generally argon) and the vibration quantity, and calculating the ladle response value (LRV) within the window duration as the third characteristic value , … . The ladle response value is the ratio of the area under the vibration quantity-time curve to the area under the flow rate-time curve (or the ratio of the integral of the vibration quantity with respect to time to the integral of the flow rate with respect to time within the window duration), and the mean value of the ladle response values of the symmetrically installed vibration sensors can also be used as the third characteristic value of the overall stirring state of the ladle within the window duration; According to the pre-calibrated ladle response value curve (i.e., the blowing intensity-vibration signal curve), the nominal value of the strongest vibration signal intensity corresponding to the measured blowing gas flow rate is obtained . If within the window duration , j≠k, and any , it is considered that the blowing and stirring at this time has reached the optimal state and the stirring is uniform, where is the third threshold value, is the fourth threshold value. That is, if the pairwise difference of the ladle response values of all symmetrically installed vibration sensors is less than the third threshold value , and the ladle response values of each symmetrically installed vibration sensor and the nominal vibration signal intensity value under the corresponding blowing intensity have a deviation less than the fourth threshold value , it can be considered that the ladle has been stirred evenly.

[0024] Note that the characteristic values of the vibration amounts in different directions should be statistically analyzed separately and cannot be mixed; in this embodiment, the judgment method two is adopted.

[0025] Preferably, the first characteristic value includes at least one of the mean value, median value, and norm. In this embodiment, the mean value is taken as the first characteristic value, that is, the mean value of the vibration amount within the window duration is used as the basic information for analysis and judgment; the second characteristic value includes at least one of the variance (RMS, root mean square value), standard deviation, maximum value, minimum value, and extreme value difference. The extreme value difference is the difference between the maximum value and the minimum value. In this embodiment, the variance is taken as the second characteristic value.

[0026] Preferably, the vibration amount includes at least one of displacement, velocity, or acceleration in at least one of the x, y, and z directions (where x and y are the two horizontal directions and z is the vertical direction) (such as the acceleration in the x direction). Generally, the vibration sensor measures acceleration, and the velocity is obtained by integrating the acceleration, and the displacement is obtained by integrating the velocity.

[0027] In this embodiment, the vibration sensor measures the acceleration amounts in the x, y, and z directions simultaneously, and the vibration velocity amount and displacement amount can be obtained by integrating the acceleration; when calculating the characteristic values, generally one of displacement, velocity, and acceleration is selected to calculate simultaneously for the x, y, and z directions, that is, the judgment method is executed simultaneously for these three types of characteristic values. Note that when performing statistical analysis on the three directions, their threshold values are generally different. In this embodiment, the characteristic value statistics and analysis are mainly performed on the velocity amounts in the x, y, and z directions, and the first characteristic value takes the mean value and the second characteristic value takes the variance.

[0028] If only one vibration amount is taken, it is recommended to preferably perform statistical calculation on the vertical direction velocity. On the one hand, because the vertical direction vibration velocity can better reflect the gas stirring intensity and is an ideal original information as the characteristic value. Experimental analysis shows that the root mean square (RMS) of the vertical direction velocity is the best characteristic for measuring the stirring intensity caused by the blowing gas flow rate in the ladle; on the other hand, relatively speaking, the vertical direction vibration velocity value is higher, the requirement for measurement accuracy is relatively lower, and the difference judgment is more accurate.

[0029] Preferably, the lower limit of the window duration is not less than the duration of 3 longest vibration periods, and the upper limit is not greater than the minimum lag duration. The lag duration includes the transition duration before the system enters the steady state caused by the change of the ladle stirring state. The change of the ladle stirring state includes the change of the blowing gas flow rate, the addition of additives and other state changes.

[0030] The lower limit of the window duration is set mainly to facilitate the statistical calculation of the vibration quantity. At least 3 complete vibration periods should be sampled to obtain at least 3 groups of effective amplitude values. Only in this way does the amplitude statistical calculation make sense. In fact, generally, it should be more than 10 or even 20 effective vibration periods to basically ensure the accuracy and reliability of the statistical calculation. That is, the sampling coverage duration of the vibration quantity corresponding to the calculated eigenvalue should be long enough (not less than the lower limit duration) to ensure the accuracy of the stirring characteristic judgment, and the shorter the window duration, the higher the real-time performance.

[0031] The upper limit of the window duration is set mainly because the window duration cannot be too long. Otherwise, the stirring time is too long. On the one hand, the time-varying characteristics of the stirring effect are averaged by time, which will make it difficult to judge whether the stirring effect really meets the standard, or the conclusion may not be accurate, or the conclusion loses its meaning. On the other hand, if the window duration is too long, the statistical result cannot reflect the real-time actual situation, the real-time performance is reduced, that is, the monitoring and judgment efficiency is reduced, which is also not conducive to the subsequent real-time control. Third, the longer the window duration, the greater the calculation workload. Fourth, taking the maximum lag duration as the reference benchmark for the upper limit of the window duration is mainly considered that if the molten steel is in the lag period transition stage (obviously non-steady state stage), the statistical characteristics should have obvious reactions. If the window duration exceeds this duration, too much steady state data is introduced, which will instead cause adverse effects on the statistical characteristic judgment, that is, abnormal phenomena may be averaged out. The lag duration includes actions such as changing the stirring gas flow rate, flow velocity or adding additives. Generally, it takes a certain lag time for the molten steel to enter the relatively steady state.

[0032] Accuracy and timeliness are sometimes a pair of contradictory requirements for the window duration and need to be considered in a balanced way.

[0033] Preferably, the window duration includes 10s, 21s or 40s. In this embodiment, the window duration is set to 10s. The main consideration is that the vibration period of the molten steel transmitted by the measured ladle is generally less than 1s. 10s, which is at least more than 100 vibration periods, is neither too small nor too large for sampling statistics and is relatively appropriate. This window duration also has a certain real-time performance. For the traditional molten steel stirring time of several minutes or even dozens of minutes, the real-time performance of the 10s detection and judgment period is still relatively obvious.

[0034] Preferably, the window sliding time interval includes integer seconds (such as every 1 second or every 2 seconds, etc.) or a unit time greater than or equal to the longest vibration period or greater than or equal to an integer multiple of the sampling period. Using an integer - second interval for the window sliding time interval conforms to the traditional sampling cognitive habit; using an interval greater than the longest vibration period is considered because at least one complete vibration period is required for each slide, and in fact, generally, multiple such vibration periods are slid at a time; using an integer multiple of the sampling period is convenient for data statistics and calculation, and various different choices have different advantages. In this embodiment, the sampling rate is 10 kHz, and a 1 - s sliding time interval is adopted, which simultaneously meets the requirement of being greater than the longest vibration period (for a vibration signal with a minimum of 2 Hz and a 0.5 - s period). Therefore, this setting essentially meets the above three conditions simultaneously.

[0035] Preferably, the filtering of significantly abnormal amplitude values includes the wild - value elimination method or the wild - value smoothing method. The wild - value elimination method includes setting an amplitude abnormal - value threshold, and in real - time judging whether the measured amplitude exceeds this threshold, and eliminating the exceeded ones, that is, the threshold - judgment method; the wild - value smoothing method includes the filtering method (suppressing the amplitude of abnormal values), or the wild - value filling method. That is, for the time series after eliminating wild values, points are filled. Mainly, the measured filling points of the eliminated points are obtained by interpolating and extrapolating using the previous measurement value or several previous measurement values. If the subsequent processing method can be compatible with the problem of missing sampling points, then no filling is required, that is, directly perform wild - value elimination; the filtering of significantly non - agitation - related frequency components includes performing a time - frequency transformation on the sampled time series to obtain a frequency - domain series, eliminating the frequency components in the frequency - domain series that are not within the agitation - related characteristic frequency range, and then performing an inverse transformation on the remaining frequency - domain series to obtain a new time series. The time - frequency transformation includes discrete Fourier transformation and inverse transformation.

[0036] Preferably, the symmetric positions include 180 - degree symmetry (2 vibration sensors), 90 - degree symmetry (4 vibration sensors), and 60 - degree symmetry (6 vibration sensors) with the center axis of the ladle as the symmetry axis. Among them, being symmetric at equal distances relative to the symmetry axis is the default.

[0037] Embodiment 2: The difference from Embodiment 1 is that in this embodiment, the vibration quantity is only statistically calculated by taking the vertical - direction velocity.

[0038] Embodiment 3: The difference from Embodiment 1 is that in this embodiment, the 2 - norm of the acceleration is taken as the first eigenvalue, and the standard deviation of the acceleration within the window duration is taken as the second eigenvalue.

[0039] The window duration is set to 21 s. Mainly, considering the analysis of actual test data, when performing central median smoothing filtering, choosing a window duration of 21 s, the correlation between the gas flow rate and the vibration characteristics is the strongest, and the best vibration characteristics of gas agitation can be obtained.

[0040] Example 4: The difference from Example 1 is that in this example, the 1-norm of the displacement is taken as the first eigenvalue, and the extreme difference of the displacement within the window duration is taken as the second eigenvalue.

[0041] The window duration is set to 40s. This is mainly considered based on empirical data that generally, it will enter a stable state after 40s. 40s is the shortest lag time of the vibration effect (of course, it is related to the stirring gas flow rate. This data is statistical data under normal relatively large flow rates. When the flow rate decreases, this time will be longer).

[0042] Example 5: The difference from Example 1 is that in this example, the first threshold, the second threshold, the third threshold, and the fourth threshold all include quasi-dynamic thresholds, and the nominal value includes a quasi-dynamic nominal value. The determination method of the quasi-dynamic threshold or the quasi-dynamic nominal value includes that when it is determined that the stirring time with a constant blowing gas flow rate exceeds the maximum value of the empirical duration, and the first eigenvalue, the second eigenvalue, or the third eigenvalue cannot meet the requirements of the corresponding Judgment Method 1, Judgment Method 2, or Judgment Method 3, the measured calculated value is used as the quasi-dynamic threshold or the quasi-dynamic nominal value of the corresponding threshold. That is, the measured value is used to correct the previously default used threshold to obtain a dynamic threshold. Note that after replacing the ladle or changing the stirring environment, the default threshold should still be used, unless the default prediction correction is made according to almost the same value during multiple actual uses, and a new dynamic threshold is obtained. At this time, the system should actively modify the default threshold. The specific calculation method of the correction amount can be calculated according to the corresponding judgment method.

[0043] This example mainly considers that due to the actual asymmetry of the installation position or the actual effect of the stirring gas injection port not being equivalent to the gas injection at the center of the bottom of the ladle, the vibration transmission and attenuation on both sides are inconsistent, and the vibration transmission and attenuation of the environmental vibration source to the target sensor are also inconsistent, etc. The above reasons combined with the dynamic change factors during stirring will cause a dynamic deviation when the eigenvalues of symmetrically installed sensors are subtracted. This dynamic deviation is related to the single installation of the sensor, the type of ladle, the real-time molten steel capacity (including the liquid level height), the shape and position of the gas injection port, the gas injection direction, the gas flow rate, the dynamic change of the additive addition amount, the dynamic change of the additive addition direction or position, the change of the same additive brand model (particle size), etc. The present invention uses quasi-dynamic thresholds and quasi-dynamic nominal values to characterize the influence of this dynamic deviation; the acquisition method of this dynamic deviation should mainly be based on the difference in the corresponding sensor eigenvalues under the condition that the stirring effect is confirmed to be good by the traditional stirring method in the same environment (including the installation environment and the stirring environment, etc.) as the benchmark for setting the quasi-dynamic threshold.

[0044] The design of this embodiment actually reflects the practicability and convenience of the present invention. Since it is difficult to obtain empirical parameters for the application of many data processing and analysis methods, which leads to difficulties in their actual application. The method for obtaining quasi-dynamic thresholds and quasi-dynamic nominal values given in this embodiment is essentially a method for obtaining empirical parameters of thresholds or nominal values. However, the default empirical parameters should be tested through multiple practices and are relatively the most reasonable parameters.

[0045] Embodiment 6: The difference from Embodiment 1 is that the judgment and analysis in this embodiment further includes a continuous window judgment method. The previous embodiments are all vibration eigenvalue judgment methods based on a single window. When the window duration is short, the continuous window judgment method can be considered. The continuous window judgment method includes performing Judgment Method 1, Judgment Method 2, or Judgment Method 3 on a continuous number of windows. When the continuous windows simultaneously meet the threshold conditions, it is considered that the stirring is uniform.

[0046] The basic principle of this project is that when the ladle stirring is not yet uniform, due to the difference in the transmission path, the measurement data characteristics of the symmetrically installed vibration sensors affected by the stirring vibration must be different. Based on this basic principle, the present invention is designed to analyze the magnitude of the vibration amount (the first eigenvalue) and the degree of dispersion of the vibration sequence (the second eigenvalue) detected by the symmetrically installed vibration sensors to determine whether the ladle stirring is uniform. It is simple and easy to implement, and can achieve precise monitoring of ladle refining with different quality requirements, which can not only improve the ladle refining stirring efficiency but also reduce the scrap rate of refined products. The present invention reasonably utilizes the vibration amount eigenvalue, taking into account both the local and the global. The remaining difficulty lies in whether the empirical parameters can be conveniently obtained, which affects the practicability of this method.

[0047] Regarding the acquisition of empirical parameters, since the gas stirring effect is judged by taking the difference, there is no need to consider the consistency of too many environmental parameters (this is also the advantage of the method of the present invention). It only needs to monitor the ladle actually stirred under the traditional empirical stirring intensity and time (duration) in real time. According to the final product quality, extract the difference of characteristic values of the corresponding grade, which can be used as the threshold reference for products of different grades (performance grades) when the present invention is applied. In special cases, different thresholds can also be set for different product models (including different types of additives, addition amounts, and process sequences), different ladle sizes, different trolley environments, and different grades of products under different gas blowing methods. In any case, the comparison of differences can filter out the influence of various common factors, while the measurement of the absolute value of vibration, the ladle response value (LRV), or the rolled ladle response value (rLRV) will be affected by dynamic factors, and the differences may be very large, and there is almost no reference between them. It is difficult to analyze and separate the effective values, and the difficulty of obtaining accurate empirical parameters is much higher than that of obtaining empirical parameters of the present invention. Moreover, the difference threshold of the vibration quantity characteristic value of the empirical parameter of the present invention has little difference for different environments and different ladles (only the threshold requirements for steelmaking products of different models and qualities are different. Obviously, the smaller the threshold, the better the product performance, the higher the requirements for stirring time or gas flow intensity, and of course the higher the requirements for the measurement accuracy of the sensor). The referenceability between them is strong because many uncertain influencing factors have been eliminated during the process of taking the difference, and the remaining mainly is the influence brought by uneven stirring.

[0048] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and refinements can be made. For example, regarding the selection of the analysis object, including the measurement direction, the type of vibration quantity (acceleration, velocity, displacement), and the selection and combination method of the characteristic value type. In addition to the several methods given in the embodiments, there are several other choices. Regarding the installation position, layout method, window duration, window sliding time interval setting, and combination of the sensor, there are many variations. These improvements and changes should also be regarded as the protection scope of the present invention.

Claims

1. A ladle refining monitoring method based on vibration measurement, characterized in that At least two vibration sensors are respectively arranged at symmetric positions of the ladle in terms of distance and angle. The monitoring method includes the following steps: S1. Data acquisition: Use the sliding window method to collect the vibration amounts of vibration sensors symmetrically installed within the window duration. , … Among them, is the sampling serial number, is the sensor number, and the window sliding time interval is greater than the sampling period. S2. Preprocessing of vibration signals, including filtering out obvious abnormal amplitude values and / or filtering out obvious frequency components not related to stirring; S3. Feature extraction, extracting the first eigenvalue corresponding to the vibration quantity , … and / or the second eigenvalue , … , where the first eigenvalue describes the magnitude of the vibration quantity, and the second eigenvalue describes the degree of discreteness of the vibration sequence; S4. Judgment and analysis, including Judgment Method 1, Judgment Method 2 or Judgment Method 3; The first determination method includes that if all , where j ≠ k, it is considered that the ladle stirring is uniform; otherwise, stirring needs to continue. Here, is the first threshold value. The second determination method includes that if all within the window duration , j≠k, and all are satisfied, it is considered that the ladle stirring is uniform; otherwise, stirring needs to continue. Among them, is the second threshold value; The third determination method includes simultaneously and real-time monitoring the blowing gas flow rate and the vibration amount, and calculating the ladle response value within the window duration as the third characteristic value 、 … , where the ladle response value is the ratio of the area under the vibration amount-time curve to the area under the flow rate-time curve; Obtain the nominal value of the strongest vibration signal intensity corresponding to the measured gas blowing flow rate according to the pre-calibrated ladle response value curve , if within the window duration , j≠k, and for any is satisfied, then it is considered that the gas blowing stirring for this time has reached the optimal state and the stirring is uniform, where is the third threshold value, is the fourth threshold value.

2. The ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that The first eigenvalue includes at least one of mean value, median value, and norm; the second eigenvalue includes at least one of variance, standard deviation, maximum value, minimum value, and extreme difference.

3. The ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that The vibration quantity includes at least one of displacement, velocity, or acceleration in at least one of the x, y, and z directions.

4. A ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, The lower limit of the window duration is not less than 3 times the longest vibration period duration, and the upper limit is not greater than the maximum lag duration. The lag duration includes the transition period duration before the system enters a steady state caused by the change of the ladle stirring state. The change of the ladle stirring state includes the change of the blowing gas flow rate and the addition of additives.

5. A ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, The window duration includes 10s, 21s, or 40s.

6. The ladle refining monitoring method based on vibration measurement according to claim 1, wherein, The window sliding time interval includes an integer number of seconds or a unit time greater than or equal to the longest vibration period or greater than or equal to an integer multiple of the sampling period.

7. A ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, The first threshold, the second threshold, the third threshold, and the fourth threshold all include a quasi-dynamic threshold, and the nominal value includes a quasi-dynamic nominal value. The method for determining the quasi-dynamic threshold or the quasi-dynamic nominal value includes that when it is determined that the stirring time with a constant blowing flow rate exceeds the maximum value of the empirical duration, and when the first eigenvalue, the second eigenvalue, or the third eigenvalue cannot meet the requirements of the corresponding Judgment Method 1, Judgment Method 2, or Judgment Method 3, the measured calculated value is used as the quasi-dynamic threshold or the quasi-dynamic nominal value of the corresponding threshold.

8. A ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, Filtering out obvious abnormal amplitude values includes the wild value elimination method or the wild value smoothing method; filtering out obvious frequency components not related to stirring includes performing time-frequency transformation on the sampled time series to obtain a frequency domain series, eliminating the frequency components in the frequency domain series that are not within the stirring-related characteristic frequency range, and then performing inverse transformation on the remaining frequency domain series to obtain a new time series.

9. The ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, The judgment and analysis further include the continuous window judgment method. The continuous window judgment method includes performing Judgment Method 1, Judgment Method 2 or Judgment Method 3 on a continuous number of windows. When the continuous windows simultaneously meet the threshold conditions, it is considered that the stirring is uniform.

10. A ladle refining monitoring method based on vibration measurement according to claim 1, characterized in that, The symmetric positions include 180-degree symmetry, 90-degree symmetry, and 60-degree symmetry with the ladle central axis as the symmetry axis.

Images