Method for predicting rolling force of flat rolling mill in different passes in rough rolling area of hot continuous rolling
By combining the Sims formula and the Grey Wolf optimization algorithm to optimize the support vector machine regression model, a rolling force prediction method for the flat roll mill in the roughing zone of hot continuous rolling mill is constructed. This method solves the problems of low rolling force prediction accuracy and poor adaptability in the existing technology, and achieves high-precision and stable rolling force prediction, thereby improving production efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HENAN IRON & STEEL GROUP CO LTD
- Filing Date
- 2025-12-26
- Publication Date
- 2026-05-08
AI Technical Summary
Existing methods for predicting rolling force in the roughing zone of hot strip mills suffer from low accuracy, weak generalization ability, and poor interpretability, failing to meet the requirements of high-precision and intelligent production.
A theoretical model of rolling force is constructed by combining the Sims formula, and the support vector machine regression model is optimized by the gray wolf optimization algorithm. The theoretical model and the data-driven model are integrated to form a hybrid prediction framework, and rolling force is predicted by collecting and preprocessing industrial data.
It improves the accuracy and stability of rolling force prediction, enhances adaptability to complex working conditions, and improves production stability and equipment load control capabilities.
Smart Images
Figure CN121997154A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of automated control of metallurgical rolling, specifically to a method for predicting the rolling force of different passes in a hot continuous rolling roughing mill. Background Technology
[0002] As a key piece of equipment in the roughing mill of a hot continuous rolling mill, the accurate prediction of rolling force is crucial for achieving dynamic roll gap setting, stable shape control, and smooth production processes. Currently, the prediction of roughing mill rolling force still faces significant technical bottlenecks, making it difficult to meet the high precision and robustness requirements of modern hot rolling production.
[0003] Currently, traditional methods mainly rely on mechanistic models based on rolling theory (such as the Orowan formula and the Sims formula). While these models reflect the physical essence and have a clear theoretical basis, the nonlinear, time-varying, and strongly coupled relationships among various factors during the rolling process lead to significant errors in the calculated rolling force results compared to actual on-site conditions. This makes it difficult for the accuracy of mechanistic models to meet actual production requirements. In contrast to theoretical analysis methods, pure data-driven models rely on historical data to train input-output mapping relationships. They can achieve high accuracy when data is sufficient and operating conditions are stable. However, their generalization ability and interpretability have inherent limitations, poor adaptability to new steel grades and processes, and they cannot reveal the intrinsic physical mechanisms of rolling force changes, making it difficult to guide process parameter optimization.
[0004] Existing rolling force prediction methods have significant shortcomings in adaptability to complex working conditions and in the fusion of physical mechanisms and data, resulting in low prediction accuracy, weak generalization ability, and poor interpretability, which cannot meet the requirements of high-precision and intelligent hot rolling production. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the present invention aims to provide a method for predicting rolling force in different passes of a hot continuous rolling roughing mill. This method integrates rolling mechanism and real-time data, and takes into account the differences in pass characteristics. It is of great significance for improving production stability in the roughing zone, reducing equipment load fluctuations, and improving product dimensional accuracy.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] A method for predicting the rolling force of different passes in a hot continuous rolling mill roughing zone, comprising the following steps:
[0008] Industrial data from different passes of the flat roll mill in the roughing zone of hot continuous rolling mill were collected and preprocessed to form a complete dataset.
[0009] A theoretical model of rolling force in the roughing zone of the flat roll mill was constructed based on the Sims formula, and the theoretical rolling force of the flat roll mill in the roughing zone of the hot strip mill was obtained for different passes.
[0010] A rolling force prediction model for the roughing zone flat roll mill is constructed based on the theoretical model and data-driven model of the rolling force of the roughing zone flat roll mill.
[0011] The collected dataset is used to train the established rolling force prediction model to predict the rolling force of different passes in the roughing mill of hot continuous rolling.
[0012] Furthermore, the industrial data for different passes of the hot continuous rolling roughing mill include roll radius, inlet width, outlet width, inlet thickness, outlet thickness, rolling temperature, rolling speed, roll elastic modulus, roll Poisson's ratio, friction coefficient, flow stress, roll gap value, deformation resistance, and rolling force.
[0013] Furthermore, the process involves collecting industrial data from different passes of the hot continuous rolling roughing mill and preprocessing it to form a complete dataset; the specific steps are as follows:
[0014] The 3σ criterion is used to remove outliers from the acquired data, as shown in the following formula:
[0015]
[0016] in, S is the average value; x x is the standard deviation; n is the number of samples in the dataset; x i For the i-th data;
[0017] The rolling data after removing outliers is normalized to reflect its inherent distribution characteristics, as shown in the following formula:
[0018]
[0019] Where x represents the original data; x min and x max , , represent the minimum and maximum values of the data, respectively; y represents the standardized data.
[0020] Furthermore, the specific steps for constructing the theoretical model of rolling force in the roughing zone flat roll mill based on the Sims formula are as follows:
[0021] The theoretical model for the rolling force of the flat roll mill in the roughing zone of a hot strip mill adopts the Sims formula, as follows:
[0022]
[0023] Where P is the rolling force, Pa; K represents the unit rolling force.m For planar deformation resistance; B is the plate width, mm, B=(B0+B1) / 2, B0 is the plate width before rolling; B1 is the plate width after rolling; Q p l is the stress state coefficient. c ' is the contact arc length, in mm;
[0024] Calculate the plane deformation resistance K using the following formula. m :
[0025]
[0026] Where σ is the deformation resistance; C is the carbon content, %; T K T is the absolute temperature. K = t + 273, where t is the deformation temperature (°C); e is the degree of deformation; u is the deformation rate (m / s); and a1 to a7 are the deformation resistance model coefficients.
[0027] Stress state coefficient Q after roll flattening p The calculation formula is as follows:
[0028]
[0029] Due to the effect of roll elastic flattening on the contact arc length, the contact arc length l of roll elastic flattening is obtained. c 'As shown in the following formula:'
[0030]
[0031] Among them, p' m The unit average pressure (Pa) is taken into account after the roll is elastically flattened; R is the roll radius (mm); c is the coefficient for calculating the contact arc length considering the elastic flattening of the roll. R' is the elastic flattening radius of the roll, mm; E is the elastic modulus; Δh is the reduction, mm.
[0032] Furthermore, the specific steps for constructing a rolling force prediction model for the roughing zone flat roll mill based on the theoretical model and data-driven model of the rolling force of the roughing zone flat roll mill are as follows:
[0033] Using absolute error to describe the overall error, a support vector machine regression model is established;
[0034] The gray wolf optimization algorithm is used to optimize the kernel function parameter σ and penalty factor γ of the support vector machine regression model to form a data-driven model;
[0035] Using the theoretical rolling force calculation value as the input variable, a rolling force prediction model for the roughing zone flat roll mill is constructed based on the theoretical model.
[0036] Furthermore, the specific steps for establishing the data-driven model are as follows:
[0037] Set the gray wolf population size N; the maximum number of iterations t; the penalty factor c and kernel function g to be optimized; the upper and lower boundaries of the wolf pack positions; and the initial positions and fitness values of α wolves, β wolves, and δ wolves.
[0038] The positions and objective functions of α, β, and δ wolves are updated randomly based on the fitness function. The wolf pack positions are iteratively updated as the pack tracks, approaches, pursues, harasses, and attacks prey. In each iteration, α, β, and δ wolves are identified as the first, second, and third best global solutions, respectively. The positions of other gray wolves are updated based on the position information of these three best solutions. The search update of the positions of α, β, and δ wolves is as follows:
[0039]
[0040] Where t is the current iteration number; and These represent the position vectors of α wolf, β wolf, and δ wolf in the current population, respectively. These represent the distances between the current candidate gray wolf and the three optimal wolves, respectively. and Both are coefficient vectors;
[0041] The search update for other gray wolf locations is as follows:
[0042]
[0043] Where t is the current iteration number; For coefficient vectors; Represents the position vector of the prey; This represents the current position vector of the gray wolf; It is a random vector in [0,1]. The distance between the wolf pack and its prey. The component decreases linearly from 2 to 0 during the iteration process.
[0044] Furthermore, the gray wolf optimization algorithm is used to optimize the kernel function parameter σ and the penalty factor γ of the support vector machine regression model, specifically as follows:
[0045] When a support vector machine regression model uses the sum of squared errors as an empirical loss for a sample set Where, x i ∈R n Let y be the input vector of the training samples. i ∈R n To correspond to the predicted output, N is the number of samples, n is the vector dimension, and a nonlinear function is used. Mapping the samples to a high-dimensional feature space, we obtain the following regression prediction model:
[0046]
[0047] Among them, w T Let b be the weight vector in the feature space, and b be the bias, where b∈R;
[0048] When a support vector machine regression model is used for a regression task, the optimized model can be expressed as follows, based on the principle of minimizing structural risk:
[0049]
[0050] Where γ is the punishment parameter, ξ i As slack variables, Let be the value after mapping the original data, and st be the constraint condition. Then, using the Lagrange multiplier method and KKT conditions, the regression function can be obtained as follows:
[0051]
[0052] Where, α i K(x,x) is a Lagrange multiplier. i K(x,x) is the kernel function. i ) = exp(-||xx i || 2 / 2σ 2 ), σ is the kernel width value, x and x i The original sequence value;
[0053] Determine whether the support vector machine regression model has reached the maximum number of iterations, and output the optimal penalty factor c and kernel function g of the support vector machine regression model.
[0054] Furthermore, the process of training the established rolling force prediction model using the collected dataset to predict the rolling force of different passes in the roughing mill of a hot strip mill involves the following steps:
[0055] The rolling force prediction model is trained by taking production data and theoretical calculations of rolling force as inputs and roughing rolling force of different passes as outputs.
[0056] The trained rolling force prediction model was tested.
[0057] The accuracy of the prediction results output by the rolling force model is evaluated using preset evaluation criteria.
[0058] Using the trained rolling force prediction model, the rolling force of different passes in the roughing mill of hot strip mill is predicted.
[0059] The beneficial effects of this invention are as follows: This invention provides a method for predicting the rolling force of a flat roll mill in the roughing zone of a hot strip mill, which has significant technical advantages and engineering application value. By collecting and preprocessing industrial data from multiple rolling passes of the roughing mill, a complete dataset is formed; a theoretical model of the rolling force of the flat roll mill in the roughing zone is constructed based on the Sims formula (SIMS) to calculate the theoretical rolling force; a rolling force prediction model for the flat roll mill in the roughing zone is constructed based on the theoretical model and the data-driven model (GWO-SVR); the established rolling force prediction model is trained using the sample dataset to predict the rolling force of different passes of the flat roll mill in the roughing zone of a hot strip mill. This method retains the analytical ability of the theoretical model to explain the physical laws of the rolling process, while also using intelligent algorithms trained with industrial big data to capture dynamic characteristics under complex working conditions. It solves the problems of insufficient generalization of traditional pure mechanistic models and weak physical interpretability of pure data models, forming a complementary hybrid prediction framework. This lays the foundation for shape control in subsequent finishing rolling processes, improves rolling stability and production efficiency, and has important engineering application value. Attached Figure Description
[0060] Appendix Figure 1 This is a schematic diagram of a hot continuous rolling mill provided in an embodiment of the present invention;
[0061] Appendix Figure 2 A flowchart illustrating a method for predicting rolling force in different passes of a hot continuous rolling roughing mill, provided as an embodiment of the present invention;
[0062] Appendix Figure 3 A schematic diagram illustrating the process of establishing a rolling force prediction model for a roughing zone flat roll mill based on a theoretical model and a GWO-SVR model, as provided in an embodiment of the present invention.
[0063] Figure 4(a) is a scatter plot of the training set prediction data of the rolling force prediction model for the roughing zone flat roll mill provided in the embodiment of the present invention;
[0064] Figure 4(b) is a scatter plot of the test set prediction data of the rolling force prediction model of the roughing zone flat roll mill provided in the embodiment of the present invention;
[0065] Appendix Figure 5 This is a schematic diagram comparing the predicted rolling force values of different models provided in the embodiments of the present invention. Detailed Implementation
[0066] The present invention will now be further described with reference to the accompanying drawings and specific embodiments:
[0067] The rolling force prediction method for different passes of the hot strip roughing mill in this embodiment was applied to a large industrial hot strip roughing mill production line, which includes a roughing descaling mill, a width-fixing mill, a vertical roll mill E1, a two-high roughing mill R1, a vertical roll mill E2, and a four-high roughing mill R2. A schematic diagram of the hot strip roughing mill is shown below. Figure 1 As shown.
[0068] like Figure 2 The diagram shown is a flowchart illustrating a specific implementation of the present invention, which includes the following steps:
[0069] S1: Collect and preprocess industrial data from multiple passes of the roughing mill to form a complete dataset;
[0070] Specifically, some rolling data of the roughing production process obtained in this embodiment are shown in Table 1:
[0071] Table 1 Rolling data for the roughing section
[0072] Specifically, in this embodiment, the data processing in step S1 is implemented as follows:
[0073] S11: The 3σ criterion is used to remove outliers from the acquired data, as shown in the following formula:
[0074]
[0075] in, S is the average value; x Standard deviation;
[0076] S12: Normalize the rolling data after removing outliers to reflect the inherent distribution characteristics of the data. The formula is as follows:
[0077]
[0078] Where x represents the original data; x min and x max y represents the minimum and maximum values of the data, respectively; y represents the standardized data; after the above preprocessing, this embodiment finally obtains 2000 complete rolling sample data.
[0079] S2: Construct a theoretical model of rolling force for a flat roll mill in the roughing zone based on the Sims formula (SIMS) and calculate the theoretical rolling force;
[0080] In this embodiment, the data required to calculate the theoretical rolling force are shown in Table 2:
[0081] Table 2 Data required for calculating theoretical rolling force
[0082]
[0083] Taking a set of data in the table as an example, the theoretical rolling force is calculated. The basic parameters include: rolling speed v = 3.33 m / s, inlet temperature T = 1153℃, inlet thickness h0 = 137.69 mm, outlet thickness h1 = 99.31 mm, inlet width b0 = 1502.7 mm, outlet width b1 = 1539.2 mm, roll elastic modulus E = 176, roll Poisson's ratio is 0.29, and rolling temperature T K =T+273=1426K, carbon content is C=0.4%.
[0084] Indentation amount: Δh = h0 - h1 = 38.38 mm;
[0085] Contact arc length:
[0086] Contact angle:
[0087] True response:
[0088] Strain rate:
[0089] Planar deformation resistance K m The calculation formula is as follows:
[0090] K m =1.155σ=1.155exp(-0.2913-2.7933C-2.0142C 2 +(2972.2729
[0091] +3011.1204C-362.0715C 2 ) / T K )u -0.4999 e 0.2476 =82.9715MPa
[0092] Where σ is the deformation resistance; C is the carbon content, %; T K T is the absolute temperature. K = t+273, where t is the deformation temperature (°C); e is the degree of deformation; u is the deformation rate (m / s); and a1~a7 are the deformation resistance model coefficients.
[0093] Stress state coefficient Q after roll flattening p The calculation formula is as follows:
[0094]
[0095] Contact arc length l of the roll elastically flattened c 'As shown in the following formula:'
[0096]
[0097] Among them, p' m The unit average pressure (Pa) is taken into account after the roll is elastically flattened; R is the roll radius (mm); c is the coefficient for calculating the contact arc length considering the elastic flattening of the roll. R' is the elastic flattening radius of the roll, mm; E is the elastic modulus; Δh is the reduction, mm.
[0098] The formula for calculating the rolling force of the flat roll mill in the roughing zone of a hot strip mill is as follows:
[0099]
[0100] Where P is the rolling force, Pa; K represents the unit rolling force. m For deformation resistance; B is the strip width, mm, B=(B0+B1) / 2, B0 is the strip width before rolling; B1 is the strip width after rolling; Q p l is the stress state coefficient. c ' is the contact arc length, in mm.
[0101] Similarly, the theoretical rolling force for each pass can be calculated using the same process for all other data.
[0102] S3: Construct a rolling force prediction model for a roughing zone flat roll mill based on theoretical models and data-driven models (GWO-SVR);
[0103] The initial parameter settings for the GWO-SVR model are shown in Table 3, including the number of wolves, the maximum number of iterations, and the upper and lower limits of the optimization parameters. After optimization, the penalty factor c is 8.44284, and the kernel function g is 0.68308. Based on this, a rolling force prediction model for the roughing zone flat roll mill is constructed based on the theoretical model and the GWO-SVR model, as follows: Figure 3 As shown.
[0104] Table 3 Initial parameter settings for the GWO-SVR model
[0105]
[0106] S4: Use the sample dataset to train the established rolling force prediction model and predict the rolling force of different passes of the hot continuous rolling roughing mill.
[0107] S41: The established rolling force prediction model is trained by taking production data and theoretical calculations of rolling force as inputs and roughing rolling force of different passes as outputs.
[0108] The model was trained using the following parameters as inputs: roll radius, inlet width, outlet width, inlet thickness, outlet thickness, rolling temperature, rolling speed, roll elastic modulus, roll Poisson's ratio, friction coefficient, flow stress, roll gap value, deformation resistance, and theoretical rolling force. The actual rolling force of passes 1-5 was used as the output. 80% of the dataset (1600 sets) was used to train the established roughing zone flat roll rolling force model.
[0109] S42: Test the trained rolling force prediction model;
[0110] The remaining 20% of the dataset (400 sets) was used to test the trained roughing zone flat roll rolling force model;
[0111] S43: The accuracy of the prediction results output by the rolling force model is evaluated using preset evaluation criteria;
[0112] The evaluation criteria used in this embodiment include the correlation coefficient R. 2 And Mean Absolute Percentage Error (MAPE):
[0113]
[0114] Where: y i and y * These represent the measured value and the predicted value, respectively; n is the total number of predicted data.
[0115] The prediction results of the rolling force prediction model proposed in this embodiment are compared with those of the theoretical model and the GWO-SVR model, as shown in Table 3. The rolling force prediction model proposed in this embodiment outperforms other models in all performance indicators, exhibiting better generalization performance and stability, while reducing human intervention in complex environments. Figures 4(a) and 4(b) show scatter plots of the prediction data of the rolling force prediction model proposed in this embodiment. As can be seen from the figures, the relative error between the training set and the test set is within 10%, indicating that the rolling force prediction model of this embodiment has stable reliability and can be better applied to the prediction of rolling force in the roughing mill of hot strip mills.
[0116] Table 4 Rolling force error for different models
[0117]
[0118] S44: Using the trained rolling force prediction model, predict the rolling force of different passes in the roughing mill of the hot strip mill.
[0119] The rolling force prediction model established in this embodiment was applied to an industrial rolling test on a roughing mill production line. The actual and predicted rolling forces of the roughing zone flat roll mill in passes 1-5 within a complete rolling unit were compared and analyzed. Figure 5As shown in the figure. The results show that the rolling force prediction model for the roughing mill flat roll mill proposed in this embodiment has significantly higher accuracy than the theoretical model. It can be applied more stably to the calculation of rolling force in the roughing mill flat roll mill, and can provide important guidance for online prediction of rolling force and high-precision rolling production.
[0120] In summary, the rolling force prediction method for different passes of a hot continuous rolling roughing mill proposed in this embodiment not only retains the ability of theoretical models to analyze the physical laws of the rolling process, but also uses intelligent algorithms trained by industrial big data to capture dynamic characteristics under complex working conditions. This solves the problems of insufficient generalization of traditional pure mechanism models and weak physical interpretability of pure data models, laying the foundation for improving rolling stability and production efficiency.
[0121] Furthermore, it should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.
[0122] Finally, it should be noted that the above description represents a preferred embodiment of the present invention. It should be pointed out that although preferred embodiments have been described, those skilled in the art, once they understand the basic inventive concept of the present invention, can make various improvements and modifications without departing from the principles described herein. These improvements and modifications should also be considered within the scope of protection of the present invention. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.
Claims
1. A method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone, characterized in that, The specific steps are as follows: Industrial data from different passes of the flat roll mill in the roughing zone of hot continuous rolling mill were collected and preprocessed to form a complete dataset. A theoretical model of rolling force in the roughing zone of the flat roll mill was constructed based on the Sims formula, and the theoretical rolling force of the flat roll mill in the roughing zone of the hot strip mill was obtained for different passes. A rolling force prediction model for the roughing zone flat roll mill is constructed based on the theoretical model and data-driven model of the rolling force of the roughing zone flat roll mill. The collected dataset is used to train the established rolling force prediction model to predict the rolling force of different passes in the roughing mill of hot continuous rolling.
2. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone as described in claim 1, characterized in that, The industrial data for different passes of the hot continuous rolling roughing mill include roll radius, inlet width, outlet width, inlet thickness, outlet thickness, rolling temperature, rolling speed, roll elastic modulus, roll Poisson's ratio, friction coefficient, flow stress, roll gap value, deformation resistance, and rolling force.
3. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone flat roll mill as described in claim 1, characterized in that, The process involves collecting industrial data from different passes of the hot continuous rolling roughing mill and preprocessing it to form a complete dataset. The specific steps are as follows: The 3σ criterion is used to remove outliers from the acquired data, as shown in the following formula: in, S is the average value; x x is the standard deviation; n is the number of samples in the dataset; x i For the i-th data; The rolling data after removing outliers is normalized to reflect its inherent distribution characteristics, as shown in the following formula: Where x represents the original data; x min and x max , , represent the minimum and maximum values of the data, respectively; y represents the standardized data.
4. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone as described in claim 1, characterized in that, The specific steps for constructing the theoretical model of rolling force in the roughing zone flat roll mill based on the Sims formula are as follows: The theoretical model for the rolling force of the flat roll mill in the roughing zone of a hot strip mill adopts the Sims formula, as follows: Where P is the rolling force, Pa; K represents the unit rolling force. m For planar deformation resistance; B is the plate width, mm, B=(B0+B1) / 2, B0 is the plate width before rolling; B1 is the plate width after rolling; Q p l is the stress state coefficient. c ' is the contact arc length, in mm; Calculate the plane deformation resistance K using the following formula. m : Where σ is the deformation resistance; C is the carbon content, %; T K T is the absolute temperature. K = t + 273, where t is the deformation temperature (°C); e is the degree of deformation; u is the deformation rate (m / s); and a1 to a7 are the deformation resistance model coefficients. Stress state coefficient Q after roll flattening p The calculation formula is as follows: Due to the effect of roll elastic flattening on the contact arc length, the contact arc length l of roll elastic flattening is obtained. c 'As shown in the following formula:' Among them, p' m The unit average pressure (Pa) is taken into account after the roll is elastically flattened; R is the roll radius (mm); c is the coefficient for calculating the contact arc length considering the elastic flattening of the roll. R' is the elastic flattening radius of the roll, mm; E is the elastic modulus; Δh is the reduction, mm.
5. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone flat roll mill as described in claim 1, characterized in that, The specific steps for constructing a rolling force prediction model for the roughing zone flat roll mill based on the theoretical model and data-driven model of rolling force in the roughing zone are as follows: Using absolute error to describe the overall error, a support vector machine regression model is established; The gray wolf optimization algorithm is used to optimize the kernel function parameter σ and penalty factor γ of the support vector machine regression model to form a data-driven model; Using the theoretical rolling force calculation value as the input variable, a rolling force prediction model for the roughing zone flat roll mill is constructed based on the theoretical model.
6. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone as described in claim 5, characterized in that, The specific steps for establishing the data-driven model are as follows: Set the gray wolf population size N; the maximum number of iterations t; the penalty factor c and kernel function g to be optimized; the upper and lower boundaries of the wolf pack positions; and the initial positions and fitness values of α wolves, β wolves, and δ wolves. The positions and objective functions of α, β, and δ wolves are updated randomly based on the fitness function. The wolf pack positions are iteratively updated as the pack tracks, approaches, pursues, harasses, and attacks prey. In each iteration, α, β, and δ wolves are identified as the first, second, and third best global solutions, respectively. The positions of other gray wolves are updated based on the position information of these three best solutions. The search update of the positions of α, β, and δ wolves is as follows: Where t is the current iteration number; and These represent the position vectors of α wolf, β wolf, and δ wolf in the current population, respectively. These represent the distances between the current candidate gray wolf and the three optimal wolves, respectively. and Both are coefficient vectors; The search update for other gray wolf locations is as follows: Where t is the current iteration number; For coefficient vectors; Represents the position vector of the prey; This represents the current position vector of the gray wolf; It is a random vector in [0,1]. The distance between the wolf pack and its prey. The component decreases linearly from 2 to 0 during the iteration process.
7. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone as described in claim 6, characterized in that, The gray wolf optimization algorithm is used to optimize the kernel function parameter σ and the penalty factor γ of the support vector machine regression model, specifically as follows: When a support vector machine regression model uses the sum of squared errors as an empirical loss for a sample set Where, x i ∈R n Let y be the input vector of the training samples. i ∈R n To correspond to the predicted output, N is the number of samples, n is the vector dimension, and a nonlinear function is used. Mapping the samples to a high-dimensional feature space, we obtain the following regression prediction model: Among them, w T Let b be the weight vector in the feature space, and b be the bias, where b∈R; When a support vector machine regression model is used for a regression task, the optimized model can be expressed as follows, based on the principle of minimizing structural risk: Where γ is the punishment parameter, ξ i As slack variables, Let be the value after mapping the original data, and st be the constraint condition. Then, using the Lagrange multiplier method and KKT conditions, the regression function can be obtained as follows: Where, α i K(x,x) is a Lagrange multiplier. i K(x,x) is the kernel function. i ) = exp(-||xx i || 2 / 2σ 2 ), σ is the kernel width value, x and x i The original sequence value; Determine whether the support vector machine regression model has reached the maximum number of iterations, and output the optimal penalty factor c and kernel function g of the support vector machine regression model.
8. The method for predicting rolling force in different passes of a hot continuous rolling mill roughing zone flat roll mill as described in claim 1, characterized in that, The specific steps are as follows: The model for predicting rolling force is trained using the collected dataset to predict the rolling force of different passes in the roughing mill of a hot strip mill. The rolling force prediction model is trained by taking production data and theoretical calculations of rolling force as inputs and roughing rolling force of different passes as outputs. The trained rolling force prediction model was tested. The accuracy of the prediction results output by the rolling force model is evaluated using preset evaluation criteria. Using the trained rolling force prediction model, the rolling force of different passes in the roughing mill of hot strip mill is predicted.