A rolling force control method for hot-rolled oriented silicon steel

Abstract

The present invention discloses a rolling force control method for hot-rolled oriented silicon steel, and relates to the technical field of plate and strip rolling. First, an initial silicon steel rolling force model is established based on a contact arc length model, a friction stress state coefficient model, a strip deformation resistance model, and a tension influence coefficient model. Then, the regression coefficient in the initial silicon steel rolling force model is optimized using a hawk optimization algorithm, thereby improving the rolling force setting accuracy of hot-rolled oriented silicon steel while avoiding a large number of simulation experiments and industrial tests. Finally, based on the optimized silicon steel rolling force model, the friction stress state coefficient, deformation resistance coefficient and other characteristics are calculated and combined with historical production data as input, and the measured rolling force is used as output to establish a deep neural network model to predict the set rolling force of silicon steel. This method improves the rolling force setting accuracy of hot-rolled oriented silicon steel, improves the silicon steel head thickness, and improves the yield rate.

Description

Technical Field

[0001] The invention belongs to the technical field of plate and strip rolling, and relates to a rolling force control method for hot-rolled oriented silicon steel. Background Art

[0002] Rolling force is an important process parameter in the hot rolling process, which has a direct impact on the size control of the strip. For the rolling force control of general steel grades, traditional analytical models such as the Sims formula are used for setting. After a certain amount of industrial experiments and manual trial and error, the setting accuracy of the rolling force can meet the production needs. For hot-rolled silicon steel, recrystallization softening will occur when rolling in the range of 900℃ to 1050℃, which makes the error between the set rolling force and the measured rolling force large, resulting in the thickness and plate shape of the hot-rolled silicon steel being affected. Therefore, silicon steel will have problems such as warping, narrowing, and even breaking during the rolling process of the finishing mill. If rolling is carried out in this temperature range, the process requirements of hot-rolled silicon steel cannot be met. The existing rolling force adjustment technology of the hot rolling mill cannot meet the production requirements of hot-rolled oriented silicon steel. It is necessary to propose a new rolling force control method to improve the setting accuracy of the rolling force of hot-rolled silicon steel and avoid the dimensional errors such as thickness and plate shape caused by the rolling force. Summary of the invention

[0003] The object of the present invention is to provide a rolling force control method for hot-rolled oriented silicon steel, so as to overcome the problem of low rolling force setting accuracy due to softening in the existing silicon steel finish rolling process.

[0004] The present invention provides a rolling force control method for hot-rolled oriented silicon steel, comprising:

[0005] Step 1: Collect historical production data of hot-rolled silicon steel, including basic rolling variables and measured rolling forces;

[0006] Step 2: Use the 3σ principle to detect outliers on basic rolling variables in historical production data and eliminate outliers;

[0007] Step 3: Establish an initial silicon steel rolling force model based on the contact arc length model, friction stress state coefficient model, strip deformation resistance model and tension influence coefficient model;

[0008] Step 4: Using the eagle optimization algorithm, the basic rolling variables and the measured rolling force are used to optimize the parameters in the initial silicon steel rolling force model to obtain the optimized silicon steel rolling force model;

[0009] Step 5: Calculate advanced rolling variables based on the optimized silicon steel rolling force model, including contact arc length, friction stress state coefficient, strip deformation resistance and tension influence coefficient;

[0010] Step 6: Combine the basic rolling variables, measured rolling forces and advanced rolling variables, normalize them and divide them into training set and test set;

[0011] Step 7: Establish a deep neural network model for silicon steel rolling force setting, use basic rolling variables as shallow inputs of the network, advanced rolling variables as deep inputs of the network, and measured rolling force as output of the network model;

[0012] Step 8: Use different combinations of advanced rolling variables as deep inputs of the network, train the network model using the training set, and test it using the test set to determine the structure of the network model and the combination of advanced rolling variables of the deep input;

[0013] Step 9: Use the trained deep neural network model to predict the set rolling force of silicon steel.

[0014] Furthermore, the basic rolling variables include: chemical composition, strip width, inlet / outlet thickness of each stand, inlet / outlet temperature of each stand, rolling speed of each stand, front / rear tension of each stand, roller diameter of each stand and plastic deformation resistance of silicon steel in each stand.

[0015] Furthermore, the step 3 is specifically as follows:

[0016] Step 3.1: Ignore the width change during silicon steel finishing and calculate the contact arc length between silicon steel and the work roll based on the contact arc length model. l c , the expression of the contact arc length model is as follows:

[0017] ;

[0018] ;

[0019] ;

[0020] In the formula, l c Indicates the length of the contact arc, in mm; R ' represents the flattening roller diameter, in mm; R 0 Indicates the radius of the working roll, in mm; H Indicates the inlet thickness, in mm; h Indicates the outlet thickness, in mm; P Indicates the calculated value of rolling force, unit kN; B Indicates the width of the strip, in mm; E is the elastic modulus; is the Poisson coefficient, c Indicates the roller flattening coefficient, the unit is MPa-1 ;

[0021] Step 3.2: Based on the friction stress state coefficient model, calculate the friction stress state coefficient during silicon steel finishing rolling. The expression of the friction stress state coefficient model is as follows:

[0022] ;

[0023] ;

[0024] ;

[0025] In the formula, Q p is the friction stress state coefficient, ε Indicates the degree of deformation; h m is the average thickness of the strip, in mm; α 1 ~ α 5 is the regression coefficient of the friction stress state coefficient model;

[0026] Step 3.3: According to the deformation resistance model, the deformation temperature, deformation speed and deformation degree of silicon steel finishing rolling process, the deformation resistance of the strip is calculated. The expression of the deformation resistance model is as follows:

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] In the formula, K m is the strip deformation resistance, σ is the resistance to plastic deformation of metal, in MPa; u m Indicates the average degree of deformation; r Indicates the rolling reduction rate; v Indicates the rolling speed in m / s; t is the rolling temperature in °C; β 1 ~ β 5 is the regression coefficient of the deformation resistance model, T is the absolute temperature in mK;

[0032] Step 3.4: According to the tension influence coefficient model, calculate the influence coefficient of tension on the rolling force in the deformation zone. The tension influence coefficient model is as follows:

[0033] ;

[0034] ;

[0035] In the formula, K T represents the tension influence coefficient, τ b Indicates the post-tension, in MPa; τ f Indicates the front tension, the unit is MPa, K σ It represents the average deformation resistance in MPa;

[0036] Step 3.5: According to the advanced rolling variables calculated in steps 3.1 to 3.4, namely, contact arc length, friction stress state coefficient, strip deformation resistance, and tension influence coefficient, a silicon steel rolling force model is established:

[0037] ;

[0038] In the formula, P It represents the calculated value of rolling force.

[0039] Furthermore, the step 4 is specifically as follows:

[0040] Step 4.1: Initialize the secretary bird population and generate an initial secretary bird population of 100 individuals. Each secretary bird individual contains the regression coefficient of the friction stress state coefficient model. α 1 ~ α 5 and the regression coefficient of the silicon steel deformation resistance model β 1 ~ β 5 ;

[0041] Step 4.2: assign the corresponding regression coefficients of the secretary vulture individuals to the friction stress state coefficient model in step 3.2 and the silicon steel deformation resistance model in step 3.3;

[0042] Step 4.3: Establish a fitness function based on the initial silicon steel rolling force model in step 3, and use the historical production data of silicon steel to calculate the fitness value of each secretary vulture individual. The mathematical expression is:

[0043] ;

[0044] In the formula, frepresents the fitness value of an individual, N is the sample size of silicon steel; represents the rolling force calculated by the initial silicon steel rolling force model in step 3, in kN; P i It represents the measured rolling force in historical production data, in kN;

[0045] Step 4.4: Sort the initial population from high to low according to the fitness value, and retain the top 20 individuals. After the remaining individuals are explored and developed, a population of 100 secretary vultures is obtained again;

[0046] Step 4.5: Repeat steps 4.2 to 4.4 until the iteration termination condition is met;

[0047] Step 4.6: Assign the regression coefficient corresponding to the individual with the lowest fitness value obtained after the iteration is terminated to the friction stress state coefficient model and the silicon steel deformation resistance model to obtain the optimized silicon steel rolling force model.

[0048] Furthermore, the chemical composition and mass percentage of the silicon steel are as follows:

[0049] C: 0.001%~0.004%, Mn: 0.12%~0.93%, Si: 0.54%~1.71%, Nb: 0.001%~0.008%, Ti: 0.0007%~0.0029%, V: 0.0007%~0.0083%, Cu: 0.002%~0.031%, Ni: 0.002%~0.008%, Cr: 0.023%~0.063%, Als: 0.1468%~0.5969%, and the rest are Fe and unavoidable impurities.

[0050] Furthermore, the slab thickness of the silicon steel ranges from 200 mm to 240 mm, and the width ranges from 1000 mm to 1300 mm; the finished product thickness ranges from 2 mm to 4 mm, and the width ranges from 1000 mm to 1300 mm.

[0051] A rolling force control method for hot-rolled oriented silicon steel has the following beneficial effects:

[0052] (1) The rolling force control method for hot-rolled oriented silicon steel of the present invention establishes a rolling force model based on hot rolling field knowledge, and uses an intelligent algorithm to optimize the regression coefficient in the rolling force model based on the historical production data of silicon steel, thereby obtaining a hot-rolled silicon steel rolling force model, while improving the calculation accuracy of the silicon steel deformation resistance model and the friction stress state coefficient model.

[0053] (2) A deep neural network model for setting the rolling force of hot-rolled silicon steel is constructed by using the advanced rolling variables calculated by the optimized silicon steel rolling force model and the basic rolling variables in the historical production data of silicon steel. The present invention solves the problem of inaccurate rolling force setting caused by recrystallization softening during the silicon steel finishing rolling process, and avoids the quality problems of hot-rolled silicon steel caused by rolling force errors. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a rolling force control method for hot-rolled oriented silicon steel of the present invention;

[0055] Figure 2 Flow chart for optimizing the regression coefficients in the rolling force model based on the eagle algorithm;

[0056] Figure 3 This is the fitness value change diagram of the optimal individual during the regression coefficient optimization process;

[0057] Figure 4 A schematic diagram of a deep neural network provided by the present invention;

[0058] Figure 5 It is a comparison chart of the coefficient of determination of the deep neural network under different combinations of advanced rolling variables;

[0059] Figure 6 The comparison chart of the root mean square error of the deep neural network under different combinations of advanced rolling variables;

[0060] Figure 7 This is a regression effect diagram of the set rolling force and the measured rolling force of the deep network model proposed in the invention. DETAILED DESCRIPTION

[0061] During the finishing rolling process, hot-rolled oriented silicon steel softens due to recrystallization, resulting in a large error in the setting of the rolling force of hot-rolled oriented silicon steel. The regression coefficients of the existing rolling force model are divided and assigned according to the steel family, and the softening phenomenon of hot-rolled oriented silicon steel is not taken into account, which cannot meet the production of silicon steel. In order to improve the rolling force setting accuracy of hot-rolled oriented silicon steel, the present invention designs a rolling force control method for hot-rolled oriented silicon steel. First, an initial silicon steel rolling force model is established based on the contact arc length model, the friction stress state coefficient model, the strip deformation resistance model and the tension influence coefficient model. Then, the regression coefficients in the initial silicon steel rolling force model are optimized by using the eagle optimization algorithm, and the rolling force setting accuracy of hot-rolled oriented silicon steel is improved while avoiding a large number of simulation experiments and industrial tests. Finally, based on the optimized silicon steel rolling force model, the friction stress state coefficient, the deformation resistance coefficient and other characteristics are calculated and combined with the historical production data as input, and the measured rolling force is used as the output to establish a deep neural network model. The method improves the rolling force setting accuracy of hot-rolled oriented silicon steel, improves the silicon steel head thickness, and improves the yield rate.

[0062] The specific implementation of the present invention will be further described in detail below in conjunction with the drawings and embodiments. In this embodiment, a 7-row finishing mill of a hot rolling mill is taken as an example to set the rolling force of the silicon steel rolling process, and the mill rolls of the finishing mill are all flat rolls.

[0063] like Figure 1 As shown, a rolling force control method for hot-rolled oriented silicon steel of the present invention comprises:

[0064] Step 1: Collect historical production data of hot-rolled silicon steel, including basic rolling variables and measured rolling forces.

[0065] The basic rolling variables include: chemical composition, strip width, inlet / outlet thickness of each stand, inlet / outlet temperature of each stand, rolling speed of each stand, front / rear tension of each stand, roller diameter of each stand and plastic deformation resistance of silicon steel in each stand.

[0066] The chemical composition and mass percentage of the silicon steel are: C: 0.001%~0.004%, Mn: 0.12%~0.93%, Si: 0.54%~1.71%, Nb: 0.001%~0.008%, Ti: 0.0007%~0.0029%, V: 0.0007%~0.0083%, Cu: 0.002%~0.031%, Ni: 0.002%~0.008%, Cr: 0.023%~0.063%, Als: 0.1468%~0.5969%, and the rest are Fe and unavoidable impurities.

[0067] In this embodiment, 2000 pieces of historical production data of silicon steel are collected, and the grades and components of the collected silicon steel are shown in Table 1 below:

[0068] Table 1 The grades and corresponding components of the hot-rolled silicon steel in the examples.

[0069]

[0070] Step 2: Use the 3σ principle to detect outliers on the basic rolling variables in the historical production data and eliminate outliers.

[0071] Step 3: Establish the initial silicon steel rolling force model based on the contact arc length model, friction stress state coefficient model, strip deformation resistance model and tension influence coefficient model, specifically:

[0072] Step 3.1: Ignore the width change during silicon steel finishing and calculate the contact arc length between silicon steel and the work roll based on the contact arc length model. l c , the expression of the contact arc length model is as follows:

[0073] ;

[0074] ;

[0075] ;

[0076] In the formula, l c Indicates the length of the contact arc, in mm; R ' represents the flattening roller diameter, in mm; R 0 Indicates the radius of the working roll, in mm; H Indicates the inlet thickness, in mm; h Indicates the outlet thickness, in mm; P Indicates the calculated value of rolling force, unit kN; B Indicates the width of the strip, in mm; E is the elastic modulus; is the Poisson coefficient, c Indicates the roller flattening coefficient, the unit is MPa -1 .

[0077] Step 3.2: Based on the friction stress state coefficient model, calculate the friction stress state coefficient during silicon steel finishing rolling. The expression of the friction stress state coefficient model is as follows:

[0078] ;

[0079] ;

[0080] ;

[0081] In the formula, Q p is the friction stress state coefficient, ε Indicates the degree of deformation; h m is the average thickness of the strip, in mm; α 1 ~ α 5 is the regression coefficient of the friction stress state coefficient model.

[0082] Step 3.3: According to the deformation resistance model, the deformation temperature, deformation speed and deformation degree of silicon steel finishing rolling process, the deformation resistance of the strip is calculated. The expression of the deformation resistance model is as follows:

[0083] ;

[0084] ;

[0085] ;

[0086] ;

[0087] In the formula, K m is the strip deformation resistance, σ is the resistance to plastic deformation of metal, in MPa; u m Indicates the average degree of deformation; r Indicates the rolling reduction rate; v Indicates the rolling speed in m / s; t is the rolling temperature in °C; β 1 ~ β 5 is the regression coefficient of the deformation resistance model, T is the absolute temperature in mK.

[0088] Step 3.4: According to the tension influence coefficient model, calculate the influence coefficient of tension on the rolling force in the deformation zone. The tension influence coefficient model is as follows:

[0089] ;

[0090] ;

[0091] In the formula, K T represents the tension influence coefficient, τ b Indicates the post-tension, in MPa; τ f Indicates the front tension, the unit is MPa, K σ Indicates the average deformation resistance, the unit is MPa.

[0092] Step 3.5: According to the advanced rolling variables calculated in steps 3.1 to 3.4, namely, contact arc length, friction stress state coefficient, strip deformation resistance, and tension influence coefficient, a silicon steel rolling force model is established:

[0093] ;

[0094] In the formula, P It represents the calculated value of rolling force.

[0095] Step 4: Using the eagle optimization algorithm, the basic rolling variables collected in step 1 and the measured rolling force are used to optimize the parameters in the initial silicon steel rolling force model, that is, the regression coefficient of the friction stress state coefficient model α 1 ~α 5 and the regression coefficient of the silicon steel deformation resistance model β 1 ~ β 5 The optimization is performed to improve the accuracy of the rolling force, deformation resistance and friction stress state coefficient of silicon steel during rolling, and the optimized silicon steel rolling force model is obtained, such as Figure 2 As shown, the specific steps are as follows:

[0096] Step 4.1: Initialize the secretary bird population and generate an initial secretary bird population of 100 individuals. Each secretary bird individual contains the regression coefficient of the friction stress state coefficient model. α 1 ~ α 5 and the regression coefficient of the silicon steel deformation resistance model β 1 ~ β 5 .

[0097] Step 4.2: Assign the corresponding regression coefficients of the secretary vulture individuals to the friction stress state coefficient model in step 3.2 and the silicon steel deformation resistance model in step 3.3.

[0098] Step 4.3: Establish a fitness function based on the initial silicon steel rolling force model in step 3, and use the historical production data of silicon steel to calculate the fitness value of each secretary vulture individual. The mathematical expression is:

[0099] ;

[0100] In the formula, f represents the fitness value of an individual, N is the sample size of silicon steel; represents the rolling force calculated by the initial silicon steel rolling force model in step 3, in kN; P i Represents the measured rolling force in historical production data, in kN.

[0101] Step 4.4: Sort the initial population from high to low according to the fitness value, and retain the top 20 individuals. After the remaining individuals are explored and developed, a population of 100 secretary vultures is obtained again.

[0102] Step 4.5: Repeat steps 4.2 to 4.4 until the iteration termination condition is met. The fitness value changes during the iteration as follows: Figure 3 shown.

[0103] Step 4.6: Assign the regression coefficient corresponding to the individual with the lowest fitness value obtained after the iteration is terminated to the friction stress state coefficient model and the silicon steel deformation resistance model to obtain the optimized silicon steel rolling force model.

[0104] Step 5: Calculate the advanced rolling variables according to the silicon steel rolling force model optimized in step 4, including contact arc length, friction stress state coefficient, strip deformation resistance and tension influence coefficient;

[0105] Step 6: Combine the basic rolling variables, measured rolling forces and advanced rolling variables, normalize them and divide them into training set and test set;

[0106] Step 7: Figure 4 As shown, a deep neural network model for setting the rolling force of silicon steel is established, the basic rolling variables collected in step 1 are used as the shallow input of the network, the advanced rolling variables calculated in step 5 are used as the deep input of the network, and the measured rolling force collected in step 1 is used as the output of the network model.

[0107] Step 8: Use different combinations of advanced rolling variables as deep inputs of the network, train the network model using the training set, and test it using the test set to determine the structure of the network model and the combination of advanced rolling variables for deep input.

[0108] The deep network models with different combinations of advanced rolling variables are trained and tested, and their performance is compared. Figure 5 and Figure 6 As shown in the figure, it can be seen that when the friction stress state coefficient and the strip deformation resistance are combined, the determination coefficient R between the rolling force output by the network and the measured rolling force 2 The maximum and root mean square errors are the smallest. Therefore, the input features of the deep network are the friction stress state coefficient and the strip deformation resistance.

[0109] from Figure 7 It can be seen that there is a good regression effect between the set rolling force and the measured rolling force of the deep network model proposed in the present invention, and the rolling force accuracy of silicon steel is significantly improved, which solves the problem of low rolling force hit rate caused by the softening of silicon steel during the finishing rolling process.

[0110] Step 9: Use the trained deep neural network model to predict the set rolling force of silicon steel.

[0111] The silicon steel slab has a thickness range of 200 mm to 240 mm and a width range of 1000 mm to 1300 mm; the finished product has a thickness range of 2 mm to 4 mm and a width range of 1000 mm to 1300 mm.

[0112] 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 rolling force control method for hot-rolled oriented silicon steel, characterized in that: include: Step 1: Collect historical production data of hot-rolled silicon steel, including basic rolling variables and measured rolling forces; Step 2: Use the 3σ principle to detect outliers on basic rolling variables in historical production data and eliminate outliers; Step 3: Establish an initial silicon steel rolling force model based on the contact arc length model, friction stress state coefficient model, strip deformation resistance model and tension influence coefficient model; Step 4: Using the eagle optimization algorithm, the basic rolling variables and the measured rolling force are used to optimize the parameters in the initial silicon steel rolling force model to obtain the optimized silicon steel rolling force model; Step 5: Calculate advanced rolling variables based on the optimized silicon steel rolling force model, including contact arc length, friction stress state coefficient, strip deformation resistance and tension influence coefficient; Step 6: Combine the basic rolling variables, measured rolling forces and advanced rolling variables, normalize them and divide them into training set and test set; Step 7: Establish a deep neural network model for silicon steel rolling force setting, use basic rolling variables as shallow inputs of the network, advanced rolling variables as deep inputs of the network, and measured rolling force as output of the network model; Step 8: Use different combinations of advanced rolling variables as deep inputs of the network, train the network model using the training set, and test it using the test set to determine the structure of the network model and the combination of advanced rolling variables of the deep input; Step 9: Use the trained deep neural network model to predict the set rolling force of silicon steel.

2. The rolling force control method for hot-rolled grain-oriented silicon steel according to claim 1, characterized in that: The basic rolling variables include: chemical composition, strip width, inlet / outlet thickness of each stand, inlet / outlet temperature of each stand, rolling speed of each stand, front / rear tension of each stand, roller diameter of each stand and plastic deformation resistance of silicon steel in each stand.

3. The rolling force control method for hot-rolled grain-oriented silicon steel according to claim 1, characterized in that: The step 3 is specifically as follows: Step 3.1: Ignore the width change during silicon steel finishing and calculate the contact arc length between silicon steel and the work roll based on the contact arc length model. l c , the expression of the contact arc length model is as follows: ； ； ； In the formula, l c Indicates the length of the contact arc, in mm; R ' represents the flattening roller diameter, in mm; R 0 represents the radius of the working roll, in mm; H Indicates the inlet thickness, in mm; h Indicates the outlet thickness, in mm; P Indicates the calculated value of rolling force, unit kN; B Indicates the width of the strip, in mm; E is the elastic modulus; is the Poisson coefficient, c Indicates the roller flattening coefficient, the unit is MPa -1 ; Step 3.2: Based on the friction stress state coefficient model, calculate the friction stress state coefficient during silicon steel finishing rolling. The expression of the friction stress state coefficient model is as follows: ； ； ； In the formula, Q p is the friction stress state coefficient, ε Indicates the degree of deformation; h m is the average thickness of the strip, in mm; α 1~ α 5 is the regression coefficient of the friction stress state coefficient model; Step 3.3: According to the deformation resistance model, the deformation temperature, deformation speed and deformation degree of silicon steel finishing rolling process, the deformation resistance of the strip is calculated. The expression of the deformation resistance model is as follows: ； ； ； ； In the formula, K m is the strip deformation resistance, σ is the resistance to plastic deformation of metal, in MPa; u m Indicates the average degree of deformation; r Indicates the rolling reduction rate; v Indicates the rolling speed in m / s; t is the rolling temperature in °C; β 1~ β 5 is the regression coefficient of the deformation resistance model, T is the absolute temperature in mK; Step 3.4: According to the tension influence coefficient model, calculate the influence coefficient of tension on the rolling force in the deformation zone. The tension influence coefficient model is as follows: ； ； In the formula, K T represents the tension influence coefficient, τ b Indicates the post-tension, in MPa; τ f Indicates the front tension, the unit is MPa, K σ It represents the average deformation resistance in MPa; Step 3.5: According to the advanced rolling variables calculated in steps 3.1 to 3.4, namely, contact arc length, friction stress state coefficient, strip deformation resistance, and tension influence coefficient, a silicon steel rolling force model is established: ； In the formula, P It represents the calculated value of rolling force.

4. The rolling force control method for hot-rolled grain-oriented silicon steel according to claim 3, characterized in that: The step 4 is specifically as follows: Step 4.1: Initialize the secretary bird population and generate an initial secretary bird population of 100 individuals. Each secretary bird individual contains the regression coefficient of the friction stress state coefficient model. α 1~ α 5 and the regression coefficient of silicon steel deformation resistance model β 1~ β 5; Step 4.2: assign the corresponding regression coefficients of the secretary vulture individuals to the friction stress state coefficient model in step 3.2 and the silicon steel deformation resistance model in step 3.3; Step 4.3: Establish a fitness function based on the initial silicon steel rolling force model in step 3, and use the historical production data of silicon steel to calculate the fitness value of each secretary vulture individual. The mathematical expression is: ； In the formula, f represents the fitness value of an individual, N is the sample size of silicon steel; represents the rolling force calculated by the initial silicon steel rolling force model in step 3, in kN; P i It represents the measured rolling force in historical production data, in kN; Step 4.4: Sort the initial population from high to low according to the fitness value, and retain the top 20 individuals. After the remaining individuals are explored and developed, a population of 100 secretary vultures is obtained again; Step 4.5: Repeat steps 4.2 to 4.4 until the iteration termination condition is met; Step 4.6: Assign the regression coefficient corresponding to the individual with the lowest fitness value obtained after the iteration is terminated to the friction stress state coefficient model and the silicon steel deformation resistance model to obtain the optimized silicon steel rolling force model.

5. The rolling force control method for hot-rolled grain-oriented silicon steel according to claim 1, characterized in that: The chemical composition and mass percentage of the silicon steel are as follows: C: 0.001%~0.004%, Mn: 0.12%~0.93%, Si: 0.54%~1.71%, Nb: 0.001%~0.008%, Ti: 0.0007%~0.0029%, V: 0.0007%~0.0083%, Cu: 0.002%~0.031%, Ni: 0.002%~0.008%, Cr: 0.023%~0.063%, Als: 0.1468%~0.5969%, and the rest are Fe and unavoidable impurities.

6. The rolling force control method for hot-rolled grain-oriented silicon steel according to claim 1, characterized in that: The silicon steel slab has a thickness range of 200 mm to 240 mm and a width range of 1000 mm to 1300 mm; the finished product has a thickness range of 2 mm to 4 mm and a width range of 1000 mm to 1300 mm.

Images