Roll bending force and roll shifting amount setting method for hot-rolled strip steel plate shape control

Abstract

The invention belongs to the technical field of rolling, and relates to a method for setting roll bending force and roll shifting amount for controlling the shape of a hot-rolled strip steel, which comprises the following steps of: acquiring historical production data of strip steel in actual production; the method comprises the following steps: performing abnormal value detection on strip steel historical production data by using a 3 sigma principle, removing abnormal values, and dividing the data into a training set and a test set; establishing a convexity prediction model based on Lasso regression, and training the convexity prediction model by using the training set data; testing the trained convexity prediction model based on Lasso regression by using the test set; establishing a plate shape objective function; constraint conditions of rolling variables are established according to all rack equipment in the finishing mill group and the plate shape theory; and a snake vulture optimization algorithm is adopted, the roll bending force and the roll shifting amount of each rack in the finishing mill group are searched within the constraint condition range, the plate shape target function is made to be minimum, and the roll bending force set value and the roll shifting amount set value are obtained.

Description

Technical Field

[0001] The invention belongs to the technical field of rolling, and in particular relates to a method for setting a bending roll force and a roll shifting amount for controlling the shape of a hot-rolled strip. Background Art

[0002] Hot-rolled strip products, with their extremely high strength, excellent plasticity and superior mechanical properties, occupy an extremely important position in the national industrial system and are widely used in high-end industries such as machinery, transportation, and aviation. The dimensional accuracy of strip steel has become an important indicator for measuring the quality of finished products. With the rapid development and mature application of various rolling technologies and equipment, the thickness of hot-rolled strip steel basically meets the market requirements, but there is still a problem of insufficient control accuracy in shape control, which causes the head of the rolled strip to have middle waves, side waves, and middle waves. However, the hot rolling shape control mechanism is very complex, with the characteristics of multivariable, strong coupling, and nonlinearity. The rolling force, bending roll force, roll shifting amount, rolling speed, rolling temperature and other factors of each finishing rolling unit in the finishing rolling unit will affect the shape of the strip steel. Therefore, it is a relatively difficult task to achieve accurate hot-rolled strip shape setting. The existing shape model is overly simplified and the factors affecting the shape are not fully considered, resulting in low shape prediction accuracy. In addition, the setting of the shape rolling force variable faces the problem of easily falling into local optimality and zero gradient. Therefore, it is necessary to improve the setting accuracy of the strip shape and reduce the shape defects of the hot-rolled strip. Summary of the invention

[0003] In order to overcome the wave shape problem caused by low accuracy in setting the shape of existing hot-rolled strip steel, the present invention provides a method for setting the bending roll force and the amount of roll shifting for controlling the shape of hot-rolled strip steel.

[0004] The present invention provides a method for setting a bending roll force and a roll shifting amount for controlling a hot-rolled strip shape, comprising the following steps: Step 1: Collect historical production data of strip steel in actual production; Step 2: Use the 3σ principle to detect outliers on the historical production data of strip steel, remove outliers, and divide them into training set and test set; Step 3: Establish a convexity prediction model based on Lasso regression and train the convexity prediction model using the training set data; Step 4: Use the test set to test the trained convexity prediction model based on Lasso regression; Step 5: Establish the plate shape objective function; Step 6: Establish the constraints of rolling variables according to the equipment of each stand in the finishing mill and the plate shape theory; Step 7: Using the eagle optimization algorithm, search for the bending roll force and roll shifting amount of each stand in the finishing mill within the constraints of step 6, so that the plate shape objective function is minimized and the setting values ​​of the bending roll force and roll shifting amount are obtained.

[0005] The method for setting the bending roll force and the roll shifting amount for controlling the shape of hot-rolled strip steel of the present invention has at least the following beneficial effects: First, the present invention provides a data-driven convexity prediction model, which can effectively quantify the influence of rolling variables in the rolling process on the convexity of the strip. Then, the plate shape objective function and the constraints of the bending roll force and the roll shifting amount in the finishing mill are given. Finally, an intelligent algorithm is used to search for the bending roll force and the roll shifting amount so as to minimize the plate shape objective function, and the optimal bending roll force and the roll shifting amount are obtained and set. The setting method of the plate shape control of the present invention solves the plate shape problems such as middle waves, edge waves, and intermediate waves. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 It is a flow chart of a method for setting the bending roll force and the roll shifting amount for controlling the shape of hot-rolled strip steel according to the present invention; Figure 2 This is a diagram showing the prediction effect of the convexity prediction model based on Lasso regression on the strip convexity on the test set in the present invention. DETAILED DESCRIPTION

[0007] The specific implementation of the present invention will be described in detail below in conjunction with the accompanying drawings. In this example, a 7-row finishing mill of a hot rolling mill is taken as an example, and the rolling rolls of the finishing mill are all flat rolls, and the rolling variables are set.

[0008] like Figure 1 As shown, a method for setting the bending roll force and the roll shifting amount for controlling the shape of a hot-rolled strip according to the present invention comprises the following steps: Step 1: Collect historical production data of strip steel in actual production, including: strip steel width, inlet temperature of each stand, inlet thickness of each stand, rolling force of each stand, bending roll force of each stand, roll shifting amount of each stand, wear convexity of each stand, thermal convexity of each stand, inlet convexity of each stand, original convexity of the rolls of each stand, measured strip steel outlet convexity of each stand and target convexity of finished products.

[0009] Step 2: The 3σ principle is used to detect outliers in the historical production data of strip steel, remove outliers, and divide them into training set and test set.

[0010] Step 3: Establish a convexity prediction model based on Lasso regression and use the training set data to train the convexity prediction model. Specifically: Step 3.1: Establish a mathematical relationship model between rolling variables and predicted crown: ; in, For the i The predicted convexity of the strip exit of the rack, W i For the i The strip steel width of the rack, in mm; For the i The efficiency factor of the strip width of the rack, in μm / mm; P i For the i The rolling force of the stand, in kN; For the i The efficiency coefficient of the rolling force of the stand, in μm / kN; F i For the i The bending roll force of the frame, in kN; For the i The efficiency coefficient of the frame bending roll force, in μm / kN; S i For the i The roller shifting amount of the frame, in mm; For the i The efficiency coefficient of the roll shifting of the frame, in μm / mm; For the i The entrance convexity of the rack, in μm; For the i The efficiency coefficient of the rack inlet convexity; For the i The inlet temperature of the rack, in °C; For the i Efficiency factor for rack inlet temperature; For the i The inlet thickness of the rack, in mm; For the i The efficiency coefficient of the rack inlet thickness, in μm / mm; For the i The wear convexity of the stand roll, in μm; No. i Thermal crown of the stand roll, in μm; For the i The original crown of the stand roll, in μm; No. i The efficiency coefficient of the roll gap crown of the stand;

[0011] Step 3.2: Establish the loss function of Lasso regression: ; in, For the i vector of efficiency coefficients of rolling variables of the stands; is calculated based on the mathematical relationship model in step 3.1. i Rack No. j The predicted convexity of the strip steel outlet, in μm; It is i Rack No. j The measured convexity of the strip outlet is in μm; N is the number of strips in the training set; λ is the penalty coefficient; is the square of the L1-norm.

[0012] Step 3.3: Initialize i Efficiency coefficient vector of rolling stand variables β i , substitute the mathematical relationship model in step 3.1 to calculate the current predicted convexity of the strip export.

[0013] Step 3.4: Take one of the efficiency coefficients in the efficiency coefficient vector as a variable and the other efficiency coefficients as constants, bring the efficiency coefficient vector into the mathematical relationship model, calculate the current strip export prediction convexity and the loss function value of the Lasso regression, and continuously update the efficiency coefficient as a variable to minimize the loss function value of the Lasso regression. According to the above process, loop through all the efficiency coefficients. When the maximum number of iterations is reached, stop the iteration and obtain a set of optimal efficiency coefficient vectors that minimize the loss function value of the Lasso regression.

[0014] Step 3.5: Assign the obtained optimal efficiency coefficient vector to the mathematical relationship model of step 3.1 to obtain the convexity prediction model based on Lasso regression.

[0015] Step 4: Use the test set to test the trained convexity prediction model based on Lasso regression. Figure 2 As shown in the figure, the convexity value calculated by the convexity model based on Lasso regression has a good fitting effect with the measured convexity value.

[0016] Step 5: In order to reduce the strip wave shape as much as possible and make the finished product convexity as close to the target convexity as possible, the plate shape objective function is established as follows: ; ; in, F i For the i The bending roll force of the frame, in kN; S i For the iThe roller shifting amount of the frame, in mm; is the first value calculated based on the convexity prediction model in step 3. i The predicted strip crown of the rack, in mm; h i For the i The outlet thickness of the rack, in mm; C tar is the target convexity, α 1 and α 2 is the weight coefficient.

[0017] Step 6: Establish the constraints of rolling variables according to the equipment of each stand in the finishing mill and the plate shape theory, specifically: Step 6.1: Set the maximum and minimum roll shifting allowed for each finishing stand.

[0018] Step 6.2: Set the maximum and minimum bending roll forces allowed for each finishing stand.

[0019] Step 6.3: Set the minimum and maximum values ​​of the relative convexity difference between the roll gaps of two adjacent frames caused by roll shifting. The relative convexity of the roll gap is the ratio of the roll gap convexity to the roll gap value.

[0020] Step 6.4: Set the minimum and maximum values ​​of the relative convexity change of the exit strips from two adjacent stands, which is allowed under the condition that each stand maintains good straightness. The relative convexity of the exit strip is the ratio of the convexity of the exit strip to the thickness of the exit strip.

[0021] Step 6.5: Set the minimum and maximum values ​​of the relative convexity of the exit strip allowed by the last effective stand in the finishing mill. The relative convexity of the exit strip is the ratio of the convexity of the exit strip to the thickness of the exit strip.

[0022] Step 7: Using the eagle optimization algorithm, search for the bending roll force and roll shifting amount of each stand in the finishing mill within the constraints of step 6, so that the plate shape objective function is minimized, and the setting values ​​of the bending roll force and roll shifting amount are obtained, specifically: Step 7.1: Generate an initial secretary bird population with 100 individuals, and each secretary bird individual contains the bending roll force and roll shifting amount of each stand of the finishing mill.

[0023] Step 7.2: Assign the bending roll force, roll shifting amount and other on-site production data collected in real time in each secretary vulture individual to the convexity prediction model of Lasso regression in step 3 and calculate the predicted strip outlet convexity of each rack, and substitute the predicted strip outlet convexity of each rack into the plate shape objective function in step 5 to calculate the fitness value of each secretary vulture individual.

[0024] Step 7.3: Sort the individuals from low to high according to their fitness values, and keep the top 30 individuals. After the remaining individuals perform aerial flight and underwater foraging operations, a population of 100 secretary bird species is obtained again.

[0025] Step 7.4: Repeat steps 7.2 to 7.3 until the iteration termination condition is met.

[0026] Step 7.5: The bending roll force and roll shifting amount corresponding to the individual with the lowest fitness value obtained after the iteration is terminated are the set values ​​of plate shape control.

[0027] The implementation of the present invention is carried out on a hot rolling production line, and the quality of the finished strip steel plate shape of the embodiment is shown in Table 1. As can be seen from Table 1, the convexity of the finished products of the embodiment is close to the target convexity, and the flatness of the finished products is less than 20IU, indicating that the finished product plate shape of the embodiment is good under the plate shape setting of the present invention.

[0028] Table 1 Plate quality of finished products of the examples.

[0029]

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the concept of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for setting the bending roll force and roll shifting amount for hot rolled strip shape control, characterized in that: The following steps are involved: Step 1: Collect historical production data of strip steel in actual production; Step 2: Use the 3σ principle to detect outliers on the historical production data of strip steel, remove outliers, and divide them into training set and test set; Step 3: Establish a convexity prediction model based on Lasso regression and train the convexity prediction model using the training set data; Step 4: Use the test set to test the trained convexity prediction model based on Lasso regression; Step 5: Establish the plate shape objective function; Step 6: Establish the constraints of rolling variables according to the equipment of each stand in the finishing mill and the plate shape theory; Step 7: Using the eagle optimization algorithm, search for the bending roll force and roll shifting amount of each stand in the finishing mill within the constraints of step 6, so that the plate shape objective function is minimized and the setting values ​​of the bending roll force and roll shifting amount are obtained.

2. The method for setting the bending roll force and roll shifting amount for hot-rolled strip shape control according to claim 1, characterized in that: The historical production data of the strip in step 1 include: strip width, inlet temperature of each stand, inlet thickness of each stand, rolling force of each stand, bending roll force of each stand, roll shifting amount of each stand, wear convexity of each stand, thermal convexity of each stand, inlet convexity of each stand, original convexity of the rolls of each stand, measured strip outlet convexity of each stand and target convexity of the finished product.

3. The method for setting the bending roll force and roll shifting amount for hot rolled strip shape control according to claim 1, characterized in that: The step 3 is specifically as follows: Step 3.1: Establish a mathematical relationship model between rolling variables and predicted crown: ； in, For the i The predicted strip exit crown of the rack, W i For the i The strip steel width of the rack, in mm; For the i The efficiency factor of the strip width of the rack, in μm / mm; P i For the i The rolling force of the stand, in kN; For the i The efficiency coefficient of the rolling force of the stand, in μm / kN; F i For the i The bending roll force of the frame, in kN; For the i The efficiency coefficient of the frame bending roll force, in μm / kN; S i For the i The roller shifting amount of the frame, in mm; For the i The efficiency coefficient of the roll shifting of the frame, in μm / mm; For the i The entrance convexity of the rack, in μm; For the i The efficiency coefficient of the rack inlet convexity; For the i The inlet temperature of the rack, in °C; For the i Efficiency factor for rack inlet temperature; For the i The inlet thickness of the rack, in mm; For the i The efficiency coefficient of the rack inlet thickness, in μm / mm; For the i The wear convexity of the stand roll, in μm; No. i Thermal crown of the stand roll, in μm; For the i The original crown of the stand roll, in μm; No. i The efficiency coefficient of the roll gap crown of the stand; Step 3.2: Establish the loss function of Lasso regression: ； in, For the i vector of efficiency coefficients of rolling variables of the stands; is calculated based on the mathematical relationship model in step 3.

1. i Rack No. j The predicted convexity of the strip steel outlet, in μm; It is i Rack No. j The measured convexity of the strip outlet is in μm; N is the number of strips in the training set; λ is the penalty coefficient; is the square of the L1-norm; Step 3.3: Initialize i Efficiency coefficient vector of rolling stand variables β i , bring into the mathematical relationship model of step 3.1, and calculate the current strip export forecast convexity; Step 3.4: Take one of the efficacy coefficients in the efficacy coefficient vector as a variable and the other efficacy coefficients as constants, bring the efficacy coefficient vector into the mathematical relationship model, calculate the current strip export forecast convexity and the loss function value of the Lasso regression, continuously update the efficacy coefficient as a variable to minimize the loss function value of the Lasso regression, repeat the above process to loop through all efficacy coefficients, and stop the iteration when the maximum number of iterations is reached to obtain a set of optimal efficacy coefficient vectors that minimize the loss function value of the Lasso regression; Step 3.5: Assign the obtained optimal efficiency coefficient vector to the mathematical relationship model of step 3.1 to obtain the convexity prediction model based on Lasso regression.

4. The method for setting the bending roll force and the roll shifting amount for hot rolled strip shape control according to claim 1, characterized in that: The plate shape objective function in step 5 is: ； ； in, F i For the i The bending roll force of the frame, in kN; S i For the i The roller shifting amount of the frame, in mm; is the first value calculated based on the convexity prediction model in step 3. i The predicted strip crown of the rack, in mm; h i For the i The outlet thickness of the rack, in mm; C tar is the target convexity, α 1 and α 2 is the weight coefficient.

5. The method for setting the bending roll force and roll shifting amount for hot rolled strip shape control according to claim 1, characterized in that: The step 6 is specifically as follows: Step 6.1: Set the maximum and minimum roll shifting allowed for each stand of the finishing rolling mill; Step 6.2: Set the maximum and minimum bending roll forces allowed for each stand of the finishing rolling mill; Step 6.3: Set the minimum and maximum values ​​of the relative convexity difference between the roll gaps of two adjacent stands caused by roll shifting. The relative convexity of the roll gap is the ratio of the roll gap convexity to the roll gap value. Step 6.4: Set the minimum and maximum values ​​of the relative convexity change of the strip steel at the exit of two adjacent stands that is allowed under the condition that each stand maintains good straightness. The relative convexity of the strip steel at the exit is the ratio of the convexity of the strip steel at the exit to the thickness of the strip steel at the exit; Step 6.5: Set the minimum and maximum values ​​of the relative convexity of the exit strip allowed by the last effective stand in the finishing mill. The relative convexity of the exit strip is the ratio of the convexity of the exit strip to the thickness of the exit strip.

6. The method for setting the bending roll force and the roll shifting amount for hot-rolled strip shape control according to claim 1, characterized in that: The step 7 is specifically as follows: Step 7.1: Generate an initial secretary bird population with 100 individuals, and each secretary bird individual contains the bending roll force and roll shifting amount of each stand of the finishing mill; Step 7.2: Assign the bending roll force, roll shifting amount and other on-site production data collected in real time in each secretary bird individual to the convexity prediction model of Lasso regression in step 3 and calculate the strip outlet predicted convexity of each rack, substitute the strip outlet predicted convexity of each rack into the flatness objective function in step 5 to calculate the fitness value of each secretary bird individual; Step 7.3: Sort the individuals from low to high according to their fitness values, and keep the top 30 individuals. After the remaining individuals perform aerial flight and underwater foraging operations, a population of 100 secretary vultures is obtained again; Step 7.4: Repeat steps 7.2 to 7.3 until the iteration termination condition is met; Step 7.5: The bending roll force and roll shifting amount corresponding to the individual with the lowest fitness value obtained after the iteration is terminated are the set values ​​of plate shape control.

Images