Prediction method and system for rolling mill cross stiffness

Abstract

The present invention discloses a method and system for predicting the cross-stiffness of a rolling mill. Based on roll position measurement results, a three-dimensional numerical model of the mill's roll system and mill is established. The input of the three-dimensional numerical model is the roll system's spatial state and rolling pressure, and the output is the mill's stiffness. Finite element simulation experiments are conducted using orthogonal testing to determine the stiffness of the mill on the operating and working sides under the influence of different horizontal factors. The position of each mill liner, rolling pressure, and mill stiffness values measured on-site are combined with orthogonal simulation data, and a neural network is used to learn the mapping relationship between the spatial position of the roll system and the mill's stiffness. The offset position of the mill's arch liner is measured to predict the longitudinal stiffness of the mill under different roll system states. This method allows for a comprehensive assessment of the impact of each roll system's cross state on the degree of roll system cross-stiffness without requiring on-site configuration of various roll system cross states. This approach is more realistic and applicable to actual production environments.

Description

Technical Field

[0001] The present invention belongs to the field of hot rolling mill measurement and control technology, and in particular relates to hot rolling mill stiffness prediction related technology. Background Art

[0002] For rolling mills, especially hot rolling mills, the degree of roll crossover has a significant impact on the vertical stiffness of the mill. For thin slab rolling, the thickness control accuracy is particularly demanding, placing even higher demands on the consistency of the mill stiffness.

[0003] In daily production, the stiffness of the rolling mill housing does not change much. The main factor affecting the stiffness is the relative position of the support rolls and the working rolls. The extent to which different positions affect the stiffness of the rolling mill is currently mainly calculated using the finite element method.

[0004] However, rolling mills are enormous, and the degree of roll crossover is typically very small, with a minimum of just 0.002 degrees. This necessitates simulations that replicate the mill's actual operating conditions at both micro and macro scales, making the simulation process highly complex. Furthermore, it's impossible to set the rolls' spatial positions according to arbitrary ideal conditions in on-site production equipment, making it impossible to systematically study the spatial state of the rolls. Consequently, the impact of roll crossover on the mill's longitudinal stiffness cannot be effectively assessed, hindering improvements in the accuracy of on-site equipment.

[0005] Therefore, if we can comprehensively evaluate and predict the influence of the mill roll cross-over degree and the mill longitudinal stiffness under the existing conditions, we can be close to the actual production environment and truly and effectively improve the mill thickness precision control level. Summary of the Invention

[0006] In order to overcome the shortcomings of the existing technology, the first technical problem to be solved by the present invention is to provide a method for predicting the cross stiffness of the rolling mill, comprehensively evaluate the influence of the cross degree of the rolling mill roll system and the longitudinal stiffness of the rolling mill, predict the longitudinal stiffness of the rolling mill under different states of the rolling mill roll system, and truly and effectively improve the thickness accuracy control level of the rolling mill.

[0007] The second technical problem to be solved by the present invention is to provide a prediction system for the cross stiffness of a rolling mill, which can predict the longitudinal stiffness of the rolling mill under different states of the rolling mill roll system, and truly and effectively improve the thickness accuracy control level of the rolling mill.

[0008] The technical solution adopted in the present invention is:

[0009] A method for predicting the cross stiffness of a rolling mill, characterized by comprising at least the following steps:

[0010] Based on the roll position measurement results, a three-dimensional geometric model of the rolling mill roll system and the rolling mill is established, wherein the input of the three-dimensional geometric model is the spatial state of the roll system and the rolling pressure;

[0011] Using the geometric model established above, an orthogonal test method was used to establish a mapping data table showing the correspondence between the roll system state and the rolling mill stiffness. The inputs were the position of the liner at the upper support roll entrance, the position of the liner at the upper work roll exit, the position of the liner at the lower work roll exit, and the position of the liner at the lower support roll entrance. The stiffness of the rolling mill on the operating side and the working side affected by different horizontal factors was obtained.

[0012] The position of each mill liner, rolling pressure value, and mill stiffness value measured on site are mixed with data obtained from orthogonal simulation experiments, and a neural network is used to learn the mapping relationship between the spatial position of the roll system and the mill stiffness.

[0013] Using MATLAB software to extract neural network learning modules, a mill stiffness prediction software interface was developed. This software measures the offset position of the mill's arch liner and predicts the longitudinal stiffness of the mill under different roll system conditions. Conversely, the mill stiffness prediction software allows on-site personnel to predict the state of the mill's roll system based on changes in the mill's longitudinal stiffness, providing a theoretical basis for roll system adjustments.

[0014] In the above technical solution, a three-dimensional numerical model of the rolling mill is established, and the finite element method is used to analyze the mechanical properties of the model. Actual measurements are performed on the rolling mill, measuring the spatial state of the roll system. The cross-roll state is then mapped into the numerical model of the rolling mill. The input is the spatial state of the roll system and the rolling pressure, and the output is the rolling mill stiffness. The rolling mill stiffness is obtained through field tests. A total of n sets of data with different roll cross-roll states are extracted on site to verify the accuracy of the established finite element model. When the model does not meet the requirements, the model parameters are adjusted and re-verified until the requirements are met. Since the data is extracted over a continuous period of time, the state of the entire rolling mill arch does not change. It can be assumed that the main change in the rolling mill stiffness is caused by the change in the spatial state of the roll system.

[0015] In the above technical solution, the position of the liner at the entrance of the upper support roll, the position of the liner at the exit of the upper working roll, the position of the liner at the exit of the lower working roll, and the position of the liner at the entrance of the lower support roll all include the driving side and the working side of each roll.

[0016] In the above technical solution, the center line of the rolling mill arch is determined by using the base control point installed near the arch base or the center point of the arch base and the center point of the arch beam.

[0017] In the above technical solution, the positions of the lining plates of the rolling mill and the center line of the rolling mill arch are measured using a laser rangefinder.

[0018] In the above technical solution, the position of each liner of the rolling mill is measured by a laser rangefinder at several points on the liner plane, the liner plane is fitted, and the center position of the liner of the fitted liner plane is taken as the liner position point.

[0019] Preferably, three points are used horizontally and three points are used vertically on the plane, a total of nine points, to fit the lining plane. The plane fitting method uses the least squares method, and the center position of the lining of the fitted lining plane is taken as the lining position point.

[0020] Mill stiffness prediction software can also be used to predict the mill arch liner position based on the actual mill stiffness. The mill liner can then be adjusted based on the predicted position to reduce the degree of crossover of the mill roll system. The specific method is as follows.

[0021] Take the above working roller as an example, and assume that the position of the driving side liner is d 1uwb , with the center of the mill arch as d 0uwb , the standard distance is d 2uwb , then the relative position of the liner d uwb Indicated as d 1uwb -d 0uwb -d 2uwb , ideally d uwb =0, by adding or removing lining plates, the lining plate position can be made ≤0.1mm; the driving side position of the lining plate at the entrance of the upper support roller d uba , working side position of upper support roller inlet lining plate d ubb , Position d of the liner drive side at the exit of the upper working roll uwa , Liner working side position d at the upper working roll exit uwb , Liner drive side position d at the exit of the lower working roll dwa , Liner working side position d at the lower working roll exit dwb , Position d on the driving side of the liner at the entrance of the lower support roller dba , Position d on the working side of the liner at the entrance of the lower support roller dbb , the calculation formula is the same as d uwb same.

[0022] In the above technical solution, the orthogonal test method is used to carry out the finite element simulation experimental steps. According to the actual situation of liner wear on site, the number of liner change levels and the number of rolling pressure levels are set. The example of this solution is 9 factors and 3 levels, and the stiffness of the rolling mill on the operating side and the working side under the influence of different level factors is obtained.

[0023] In the above technical solution, a particle swarm neural network is used to learn the mapping relationship between the spatial position of the roll system and the stiffness of the rolling mill.

[0024] In the above technical solution, in the step of predicting the longitudinal stiffness of the rolling mill under different states of the rolling mill roll system, the working side stiffness K of the rolling mill is predicted. 1w and the rolling mill stiffness K on the transmission side 1d , the calculated rolling mill stiffness difference K1 = K 1w -K 1d, by judging whether the difference in the rolling mill stiffness exceeds the set range, it is determined whether the rolling mill roll system cross exceeds the standard.

[0025] In the above technical solution, the setting range is 50-60. If K1 exceeds 50-60, the mill roll crossover is considered to be seriously excessive, and the liner position needs to be readjusted to reduce the mill stiffness to less than 50. The liner can be adjusted to the desired position without the need for mill operation testing. Mill stiffness prediction software uses the stiffness difference (obtained from on-site process data) to inversely determine the relative offset position of the liner. This relative offset position is used to adjust the liner position during equipment shutdown, eliminating the need for further measurement using a laser rangefinder; liner adjustments can be made directly.

[0026] A rolling mill cross stiffness prediction system is characterized by being used to implement the above steps.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention creatively ignores the state changes of the rolling mill arch. This is because the arch state will not be replaced or its components replaced during the entire production cycle. It focuses on the influence of the degree of cross-over of the rolling mill roll system and the longitudinal stiffness of the rolling mill, and fully combines artificial intelligence learning and on-site measured data in each link, which is more in line with the actual stress conditions and working conditions of the roll system.

[0029] The present invention proposes a method and system for measuring the spatial position of a roller using a laser tracker. The position measurement method, model setting, and position fitting setting are scientific, the data measurement method is simple, and it is convenient for rapid on-site detection and measurement, which complies with the control strategy of continuous production.

[0030] This method allows for a comprehensive assessment of the impact of the mill roll crossover state on the mill roll crossover degree, without requiring on-site configuration of various roll crossover states. This approach is more responsive to actual production environments and, by predicting mill stiffness differences, effectively improves mill thickness precision control, demonstrating its potential for widespread application. Conversely, the mill stiffness differences can be used to determine the roll crossover state, providing data support for on-site equipment maintenance.

[0031] The present invention uses a particle swarm neural network to learn the mapping relationship, and on this basis, uses MATLAB software to compile mill roll system longitudinal stiffness prediction software to predict the longitudinal stiffness of the mill under different roll system states.

[0032] By using the present invention, when the equipment is shut down, the relative offset position of the roller system is obtained by prediction software to adjust the position of the liner, without the need to use the laser rangefinder for measurement again to obtain the adjustment plan of the liner, thus saving the adjustment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0034] Figure 1 It is a stiffness prediction flow chart of the method for predicting the cross stiffness of a rolling mill according to the present invention.

[0035] Figure 2 It is a schematic diagram of the spatial position of the rollers of the present invention.

[0036] Figure 3 It is a top view of the roller bearing seat liner of the present invention.

[0037] Figure 4 It is a schematic diagram for determining the center line of the rolling mill arch of the present invention.

[0038] Figure 2-4 In the figure, the corresponding reference numerals are as follows:

[0039] 1-steel belt; 2-upper support roller; 3-upper working roller; 4-lower working roller; 5-lower support roller; a1-drive side of the inlet lining of the upper support roller; a2-working side of the inlet lining of the upper support roller; b 1- Drive side of lower support roller inlet lining; b2-operating side of lower support roller inlet lining; c1-drive side of upper working roller outlet lining; c2-working side of upper working roller outlet lining; d1-drive side of lower working roller outlet lining; d2-working side of lower working roller outlet lining; 6-drive side; 7-working side; 8-rolling mill arch; 9-arch crossbeam; 10-reference fixing point. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0041] Figure 1 This is a flow chart for predicting the rolling mill stiffness of the present invention. The first step is to measure the position of the roll system and establish a three-dimensional geometric model of the rolling mill roll system and the rolling mill. The second step is to verify the validity of the model using the roll system position and on-site stiffness measurement data. The input is the spatial state of the roll system and the rolling pressure, and the output is the rolling mill stiffness. The rolling mill stiffness is obtained from on-site tests. A total of n groups of data with different roll system cross states are extracted on-site (the preferred embodiment is 10 groups of data) to verify the accuracy of the established finite element model. When the model does not meet the requirements, the parameter settings of the model are adjusted and verified again until the requirements are met. Since the data is extracted for a continuous period of time, and the rolling mill arch is not replaced during the entire production cycle, the state of the entire rolling mill arch has not changed, so the main change in the rolling mill stiffness is caused by the change in the spatial state of the roll system.

[0042] In the first step, a three-dimensional geometric model of the mill roll system and the mill is established based on the roll system position measurement results. The input of the three-dimensional geometric model is the spatial state of the roll system and the rolling pressure (that is, in the three-dimensional geometric model, only the position of the roll system and the value of the rolling pressure in the geometric figure are changed, and the other geometric states of the model remain unchanged). The finite element simulation output of the geometric model is the mill stiffness. The mill stiffness expression is k = f(a1,a2,b1,b2,c1,c2,d1,d2,g) = k1a1+k2a2+k3b3+k4b4+h1c1+h2c2+h3d1+h4d2+pg, where:

[0043] a1-the driving side position of the upper support roller inlet lining; a2-the working side position of the upper support roller inlet lining; b 1- The position of the lower support roll inlet liner on the transmission side; b2-the position of the lower support roll inlet liner on the operating side; c1-the position of the upper work roll outlet liner on the transmission side; c2-the position of the upper work roll outlet liner on the working side; d1-the position of the lower work roll outlet liner on the transmission side; d2-the position of the lower work roll outlet liner on the working side; g-the rolling mill pressure; k-the rolling mill longitudinal stiffness; K1, k2, k3, k4, h1, h2, h3, h4, p are all constants, among which the coefficients of K1, k2, k3, k4 are close, 0.8≤k i / k j ≤1.2, (i, j=1,2,3,4), h1,h2,h3,h4 coefficients are close, 0.8≤h i / h j ≤1.2, (i, j=1,2,3,4), and k i / h j ≥10, p is the rolling mill pressure constant, h i / p≥2.

[0044] This is due to the special structure of the rolling mill. On-site maintenance experience shows that the support roll side liner has a much greater impact on the mill stiffness than the work roll side liner. At the same time, the rolling pressure of the mill has a smaller impact on the mill stiffness than other factors.

[0045] The established geometric model was validated using field-measured data. A total of no fewer than 10 sets of data were extracted for different roll crossover states (i.e., the roll position was measured every 10 days, and on-site rolling force and stiffness data were obtained). This was used to verify the accuracy of the established finite element model. When the model did not meet the requirements, the model parameters were adjusted and revalidated until it met the requirements. Because the data was extracted over a short, continuous period, the state of the entire mill arch did not change. Therefore, it can be assumed that the primary change in mill stiffness is caused by changes in the spatial state of the roll system.

[0046] The finite element simulation experiment was carried out using the orthogonal test method. A 9-factor 3-level structural design orthogonal simulation experiment was adopted. The input was the position of the liner a1 at the entrance of the upper support roll 2, the position of the liner c1 at the exit of the upper working roll 3, the position of the liner d1 at the exit of the lower working roll 4, and the position of the liner b1 at the entrance of the lower support roll 5. The above positions include the roll drive side 6 and the working side 7, a total of eight positions. According to the actual situation of the liner wear on site, the number of liner change levels is set to 3 and the number of rolling pressure levels is set to 3. The stiffness of the rolling mill on the roll drive side 6 and the working side 7 under the influence of different level factors is obtained. The on-site measured rolling mill liner position, rolling pressure value, and rolling mill stiffness value are mixed with the orthogonal simulation experiment data. The neural network is used to learn the mapping relationship between the spatial position of the roll system and the rolling mill stiffness. Then the neural network learning module is extracted using MATLAB software to compile the rolling mill stiffness prediction software interface.

[0047] The mill stiffness prediction software can also be used to predict the mill arch liner position based on the on-site mill stiffness. The mill liner can be adjusted based on the predicted position to reduce the crossover degree of the mill roll system. Figure 2-4 The mill lining position and the mill arch center line are measured by a laser rangefinder. The mill arch center line is determined by the base control point 10 installed near the arch base and the center point of the arch crossbeam 9. The liner position is measured by the laser rangefinder at 9 points on the liner plane. The liner plane is fitted and the liner center position is taken as the liner position point. Taking the working roll 3 as an example, the transmission side liner position is d 1uwb , with the center of the mill arch as d 0uwb , the standard distance is d 2uwb , then the relative position of the liner d uwb Represented as d 1uwb -d 0uwb -d 2uwb , ideally d uwb =0, by adding or removing lining plates, the lining plate position can be made ≤0.1mm; the upper support roller inlet lining plate drive side a1 position d uba 、Upper support roller inlet lining working side a2 position d ubb , Position d of c1 on the liner drive side at the exit of the upper working roll uwa 、C2 position d on the working side of the liner at the exit of the upper working roll uwb , liner drive side d1 position at the exit of the lower working roll dwa , d2 position on the working side of the liner at the exit of the lower working roll dwb , liner drive side b1 position d at the entrance of the lower support roller dba , b2 position d on the working side of the liner at the entrance of the lower support roller dbb .

[0048] During on-site use, the offset position of the mill arch liner is measured and the data is input into the software to predict the mill stiffness K on the working side of the mill. 1w and the rolling mill stiffness K on the transmission side 1d , then the rolling mill stiffness difference K1=K 1w -K 1d By determining the mill stiffness difference, if K1 exceeds 50-60, the mill roll crossover is considered excessive, and the liner is readjusted. The software is then used to predict the situation, eliminating the need to start the mill for actual measurement and reducing on-site adjustment time. The mill stiffness prediction software uses the stiffness difference to reverse-calculate the relative offset position of the liner. When the equipment is shut down, the relative offset position calculated by the prediction software can be used to adjust the liner position, eliminating the need for further measurement using a laser rangefinder, and ultimately achieving a liner adjustment plan.

[0049] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A method for predicting the cross stiffness of a rolling mill, characterized in that At least the following steps are included: Based on the roll position measurement results, a three-dimensional numerical model of the rolling mill roll system and the rolling mill is established. The input of the three-dimensional numerical model is the spatial state of the roll system and the rolling pressure value, and the output is the rolling mill stiffness. Finite element simulation experiments were conducted using the orthogonal test method. The inputs were the position of the liner at the upper support roll entrance, the position of the liner at the upper work roll exit, the position of the liner at the lower work roll exit, and the position of the liner at the lower support roll entrance. The results showed that the influence of different horizontal factors on the stiffness of the rolling mill on the operating side and the working side was significant. The position of each liner of the rolling mill, rolling pressure value, and rolling mill stiffness value measured on site are mixed with orthogonal simulation experimental data, and the mapping relationship between the spatial position of the roll system and the rolling mill stiffness is learned using a neural network. Matlab software was used to extract the neural network learning module, and the mill stiffness prediction software interface was compiled to measure the offset position of each liner of the mill and predict the longitudinal stiffness of the mill under different states of the mill roll system.

2. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: The position of the liner at the entrance of the upper support roll, the position of the liner at the exit of the upper working roll, the position of the liner at the exit of the lower working roll, and the position of the liner at the entrance of the lower support roll all include the drive side and the working side of each roll.

3. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: The center line of the rolling mill arch is determined by the base control point installed near the arch base or the center point of the arch base and the center point of the arch beam.

4. The method for predicting the cross stiffness of a rolling mill according to claim 1, wherein: The position of each liner of the rolling mill is measured by a laser rangefinder at several points on the liner plane, generally several points in the horizontal and vertical directions of the plane, and the liner plane is fitted. The plane fitting method adopts the least squares method, and the center position of the liner of the fitted liner plane is taken as the liner position point.

5. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: In the finite element simulation experimental steps using the orthogonal test method, the number of liner change levels and the number of rolling pressure levels are set according to the actual situation of liner wear on site, and the stiffness of the rolling mill on the operating side and the working side under the influence of different level factors is obtained.

6. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: The mapping relationship between the spatial position of the roll system and the stiffness of the rolling mill is learned using particle swarm neural network.

7. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: Predict the mill stiffness K on the working side of the mill 1w and the rolling mill stiffness K on the transmission side 1d， The calculated rolling mill stiffness difference K1 = K 1w -K 1d Whether the mill roll system cross exceeds the standard is determined by judging whether the mill stiffness difference exceeds the mill stiffness difference setting range.

8. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: The setting range of the rolling mill stiffness difference is 50~60. If the rolling mill stiffness difference K1>50-60, it is considered that the rolling mill roll system crossover exceeds the standard seriously and the equipment needs to be shut down for maintenance. Otherwise, the rolling mill roll system crossover is not obvious and production can continue.

9. The method for predicting the cross stiffness of a rolling mill according to claim 1, characterized in that: The relative offset position of the liner is obtained by reverse calculation using the stiffness difference. When the equipment is shut down, the liner position is adjusted using the relative offset position without the need to use a laser rangefinder for measurement again to obtain a liner adjustment plan. Alternatively, the mill stiffness prediction software can be used to predict the mill arch liner position based on the on-site mill stiffness. The mill liner is adjusted based on the predicted mill arch liner position to reduce the degree of crossover of the mill roll system.

10. A prediction system for rolling mill cross stiffness, characterized in that Used to implement the steps described in any one of claims 1 to 9 above.

Images