Hot rolled strip convexity prediction method based on weight distribution strategy hybrid modeling

Abstract

The invention provides a hot rolled strip convexity prediction method based on weight distribution strategy hybrid modeling, and relates to the technical field of metallurgical machinery and automation. The method comprises the following steps: acquiring historical production data in the rolling process of a hot-rolled strip, and establishing a convexity mechanism model of the hot-rolled strip; an improved ant colony algorithm is adopted to optimize the extreme learning machine, and a hot rolled strip convexity data driving model is constructed; calculating the information entropy of the mechanical model and the information entropy of the data driving model by adopting an entropy weight method; according to the model information entropy, calculating an initial weight coefficient of a mechanical model and an initial weight coefficient of a data-driven model, and establishing a hybrid model through a weighted summation method; and in hot-rolled strip production of different rolling units, the weight is dynamically adjusted according to the deviation value between the predicted value and the actual value of the mechanism model and the data driving model, and a final hot-rolled strip convexity prediction result is output. By adopting the method, the accuracy of convexity prediction of the hot-rolled strip can be improved.

Description

Technical Field

[0001] The present invention relates to the field of metallurgical machinery and automation technology, and in particular to a method and device for predicting the convexity of a hot-rolled plate and strip based on hybrid modeling with a weight distribution strategy. Background Art

[0002] Crown is a key indicator for measuring the cross-sectional profile quality of hot-rolled strip. Accurate crown prediction directly impacts product quality and reduces potential risks. Traditional crown mechanism models have a comprehensive theoretical framework, but their prediction accuracy is limited by simplifying assumptions during modeling. Data-driven models can ignore the influence of cross-coupling between input features and have higher prediction accuracy. However, due to their inherent black-box nature and lack of expert knowledge integration, the model's accuracy is limited by the quality and scale of the data, and can only be used to predict data close to the training samples.

[0003] Combining a mechanism-based model with a data-driven model can address the shortcomings of a single model and leverage the strengths of each. Patent CN.112749505.A discloses a method for predicting the cross-sectional shape of hot-rolled strip using a mechanism-based data fusion approach. This method incorporates mechanism-based calculation results, such as inter-stand strip temperature, as input features into a multi-output support vector regression model through dimensionality manipulation, effectively improving the accuracy of convexity prediction. However, the increase in dimensionality significantly challenges the data-driven model's nonlinear processing capabilities, easily leading to the curse of dimensionality and increasing the risk of overfitting. Patent CN.114021290.B discloses a method for predicting the convexity of plate and strip based on the fusion of data-driven and mechanism-based models. This method uses the convexity value calculated by the mechanism model as a baseline value, and the deviation between the calculated value and the actual value as the output of a deep neural network. The final convexity prediction result is obtained by summing the results. Prior art also discloses a combined prediction method for the convexity of hot-rolled strip based on a mechanism-based and data-driven approach. This method uses a randomly configured network to predict the convexity deviation and superimposes it with the calculated convexity value to construct a hybrid model for the convexity of hot-rolled strip. Hybrid modeling using deviation compensation can effectively improve the accuracy of hot-rolled strip crown prediction. However, the mechanism model has inevitable systematic errors due to over-simplified assumptions. In the above-mentioned hybrid addition process, the error transmission also brings inevitable systematic errors to the new model. Summary of the Invention

[0004] To address the technical issues of existing technologies, such as high fitting risks and inevitable systematic errors in the mechanism model due to oversimplified assumptions, the present invention provides a method and device for predicting the crown of a hot-rolled strip based on hybrid modeling with a weight distribution strategy. The technical solution is as follows:

[0005] In one aspect, a hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling is provided. The method is implemented by a hot-rolled strip crown prediction device based on weight distribution strategy hybrid modeling. The method includes:

[0006] S1. Obtain historical production data during the hot-rolled strip rolling process;

[0007] S2. Using the historical production data, establish a hot-rolled strip crown mechanism model;

[0008] S3. Use the improved ant colony algorithm to optimize the extreme learning machine and build a data-driven model for the crown of hot-rolled strips;

[0009] S4. Calculate the information entropy of the mechanism model and the information entropy of the data-driven model using an entropy weight method; calculate the initial weight coefficient of the mechanism model based on the information entropy of the mechanism model; calculate the initial weight coefficient of the data-driven model based on the information entropy of the data-driven model, and establish a hybrid model using a weighted summation method;

[0010] S5. Obtain the data of the current hot-rolled strip production of different rolling units and input them into the hybrid model. Obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

[0011] On the other hand, a hot-rolled strip crown prediction device based on weight distribution strategy hybrid modeling is provided, which is applied to the hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling, and the device includes:

[0012] The first acquisition unit is used to acquire historical production data during the hot-rolled strip rolling process;

[0013] A first establishing unit is used to establish a hot-rolled strip crown mechanism model using the historical production data;

[0014] A construction unit is used to optimize the extreme learning machine using an improved ant colony algorithm to construct a data-driven model for the crown of hot-rolled strips;

[0015] The second establishing unit is configured to calculate the information entropy of the mechanism model and the information entropy of the data-driven model by using an entropy weight method; calculate the initial weight coefficient of the mechanism model according to the information entropy of the mechanism model; calculate the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establish a hybrid model by a weighted summation method;

[0016] The output unit is used to obtain the data of the current hot-rolled strip production of different rolling units and input it into the hybrid model, obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

[0017] On the other hand, a hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling is provided, and the hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling includes: a processor; a memory, and the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, any one of the hot-rolled strip convexity prediction methods based on weight distribution strategy hybrid modeling is implemented.

[0018] On the other hand, a computer-readable storage medium is provided, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned methods for predicting the convexity of hot-rolled strips based on hybrid modeling of a weight distribution strategy.

[0019] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0020] The embodiment of the present invention first obtains historical production data during the rolling process of hot-rolled plates and strips; uses the historical production data to establish a hot-rolled plate and strip convexity mechanism model; secondly, uses an improved ant colony algorithm to optimize the extreme learning machine and constructs a hot-rolled plate and strip convexity data-driven model; uses the entropy weight method to calculate the information entropy of the mechanism model and the information entropy of the data-driven model; calculates the initial weight coefficient of the mechanism model according to the information entropy of the mechanism model; calculates the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establishes a hybrid model through the weighted summation method; finally, obtains the data of the current hot-rolled plate and strip production of different rolling units and inputs it into the hybrid model, obtains a first prediction result through the mechanism model; obtains a second prediction result through the data-driven model; calculates the deviation between the first prediction result and the actual true value obtained in advance; calculates the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjusts the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and outputs the final hot-rolled plate and strip convexity prediction result.

[0021] Compared with the traditional hybrid modeling method, the embodiment of the present invention, after constructing the convexity mechanism model and the high-precision data-driven model, solves the information entropy of the error decision matrix through the entropy weight method and adds weight coefficients to the convexity mechanism model and the data-driven model respectively to correct the model results, effectively reducing the impact of error transmission on model accuracy during the hybrid modeling process. The present invention can improve the accuracy of hot-rolled strip convexity prediction. The hybrid model established by the embodiment of the present invention can give full play to the characteristics of the mechanism model with practical physical significance, avoid the model results from violating the actual physical laws, and at the same time give play to the advantage of the data-driven model in achieving high-precision prediction, providing theoretical guidance for achieving high-precision hot-rolled strip convexity control in the hot rolling industry, which is of great significance for achieving high-precision production of hot-rolled strip. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 This is a flow chart of a hot-rolled strip crown prediction method based on hybrid modeling with a weight distribution strategy provided by an embodiment of the present invention;

[0024] Figure 2 This is a CSP hot rolling production line equipment layout diagram provided by an embodiment of the present invention;

[0025] Figure 3 This is a comparison chart of the accuracy of hot-rolled strip crown prediction using a mechanism model provided by an embodiment of the present invention;

[0026] Figure 4 This is a comparison chart of the prediction accuracy of the hot-rolled strip crown using a hot-rolled strip crown data-driven model provided by an embodiment of the present invention;

[0027] Figure 5 This is a comparison chart of the accuracy of hot-rolled strip crown prediction using a hybrid model provided by an embodiment of the present invention;

[0028] Figure 6 This is a block diagram of a hot-rolled strip crown prediction device based on hybrid modeling with a weight distribution strategy provided by an embodiment of the present invention;

[0029] Figure 7 It is a structural schematic diagram of a hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0031] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0032] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0033] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0034] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0035] The embodiment of the present invention provides a hot-rolled strip convexity prediction method based on weight distribution strategy hybrid modeling, which can be implemented by a hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling, which can be a terminal or a server. Figure 1 The flowchart of the hot-rolled strip convexity prediction method based on the weight distribution strategy hybrid modeling is shown. The processing flow of the method may include the following steps:

[0036] S1. Obtain historical production data during the hot-rolled strip rolling process.

[0037] Optionally, historical production data during the hot strip rolling process of S1 includes:

[0038] Mill entrance thickness, exit thickness, rolling speed, entrance temperature, final rolling temperature, workpiece width, rolling force of each stand, bending roll force, roll shifting amount, roll thermal crown and wear parameters.

[0039] Among them, such as Figure 2The following is an equipment layout diagram of a CSP hot strip rolling line according to an embodiment of the present invention, including a thin slab continuous caster, a heating furnace, a descaling machine, and F1-F7 finishing mills. Table 1 shows historical production data during the hot strip rolling process.

[0040] Table 1

[0041]

[0042] Wherein, h represents the outlet thickness of hot-rolled strip; F1_RF represents the rolling force of F1 stand; F5_BF represents the bending roll force of F5 stand; F5_S represents the roll shifting amount of F5 stand; FDT represents the finishing temperature; C W Indicates the actual value of convexity.

[0043] S2. Use historical production data to establish a hot-rolled strip crown mechanism model.

[0044] Optionally, S2 uses historical production data to establish a hot-rolled strip crown mechanism model, including:

[0045] Based on the historical production data of hot-rolled strip rolling, the force-deformation behavior of the hot-rolled strip rolling process is analyzed, and a hot-rolled strip crown mechanism model is established, which is expressed by the following formula (1):

[0046] (1)

[0047] in, Indicates the outlet convexity of hot-rolled strip, Indicates the rolling mill serial number; Indicates the inlet strip convexity correction coefficient; Indicates rolling force; Indicates the bending roll force; Indicates the transverse stiffness of the rolling mill; Indicates the lateral stiffness of the bending roller; Indicates the amount of roller shifting; Indicates the width of hot rolled strip; Indicates the thermal crown of the work roll; Indicates wear convexity; Indicates the initial roller crown; Indicates the correction coefficient of roller shifting; Indicates the strip width correction factor; Indicates the roll crown correction coefficient.

[0048] S3. Use the improved ant colony algorithm to optimize the extreme learning machine and build a data-driven model for the convexity of hot-rolled strips.

[0049] Optionally, the specific implementation process of S3 includes S31-S37:

[0050] S31. Obtain input characteristic variables of a hot-rolled strip crown data-driven model; wherein the input variables include: hot-rolled strip inlet thickness, outlet thickness, inlet temperature, final rolling temperature, strip width, rolling force, bending force, roll shifting, rolling speed, roll thermal crown, and roll wear;

[0051] Among them, the input characteristic variables of the hot-rolled strip crown data-driven model are expressed as:

[0052]

[0053] Among them, H represents the inlet thickness of hot-rolled strip; h represents the outlet thickness of hot-rolled strip; ET represents the inlet temperature; FDT represents the final rolling temperature; B represents the strip width; P represents the rolling force; F represents the bending roll force; S represents the roll shifting amount; V represents the rolling speed; Indicates the thermal crown of the roll; Indicates roller wear;

[0054] Among them, the output variables of the hot-rolled strip crown data-driven model are expressed as:

[0055]

[0056] in, Indicates the convexity value;

[0057] S32, normalizing the input characteristic variables of the hot-rolled strip crown data-driven model to obtain normalized input characteristic variables;

[0058] A feasible implementation method is to obtain the normalized input feature variables through the following formula (2):

[0059] (2)

[0060] in, Represents the normalized input feature variable; Represents the original data; Indicates the maximum data; Indicates minimum data.

[0061] S33, initializing the parameters of the extreme learning machine and the improved ant colony algorithm, and randomly setting the initial positions of the ants; wherein the parameters of the extreme learning machine include: weight, bias, and number of hidden layer nodes; wherein the parameters of the improved ant colony algorithm include: number of ant populations, maximum number of iterations, pheromone volatility coefficient, transition probability constant, and pheromone release amount;

[0062] S34. Perform path search based on the randomly set initial position of the ants, and calculate the state transition probability based on the concentration of pheromones on the path, which is expressed by the following formula (3):

[0063] (3)

[0064] in, represents the probability that the kth ant moves from the i-th position to the j-th position at time t; represents the pheromone weight coefficient; represents the weight coefficient of the heuristic function; and ; s represents the remaining nodes to be visited except i; represents the set of remaining nodes to be visited by the k-th ant; represents the pheromone concentration on the path (i, j) at time t; represents a heuristic information, indicating the expected level of an ant moving from position i to position j;

[0065] During the search process, ants communicate with other ants through the mutual deposition and volatilization of pheromones. The concentration of pheromones on the path will affect the probability of ants choosing a path and move towards the direction of high concentration.

[0066] S34. Update the ant position according to the state transition probability; update the pheromone according to the updated ant position; wherein the updated pheromone is the sum of the volatilized pheromone and the newly generated pheromone, which is expressed by the following formula (4) to formula (6):

[0067] (4)

[0068] (5)

[0069] (6)

[0070] in, represents the amount of pheromone released by the kth ant on the path (i, j); Q represents the total amount of pheromone released by the ant in one cycle; is the Euclidean distance length of the path passed by the kth ant; is the path taken by the kth ant in this iteration; γ is the volatility coefficient of pheromone, 0<γ<1; Indicates updated pheromone; Indicates the original pheromone; Indicates newly produced pheromones;

[0071] S35. A dynamic adjustment mechanism for pheromone volatility is used to improve the original fixed pheromone volatility coefficient, so that the volatility changes with the iterative process of the algorithm and different search strategies are used at different stages. The improved pheromone volatility is expressed by the following formula (7):

[0072] (7)

[0073] in, The minimum pheromone volatilization rate is set, which is usually used in the early stage of the algorithm to enhance exploration; The maximum pheromone volatilization rate is set, which is usually used in the later stage of the algorithm to consolidate the optimal path; Indicates the current iteration number; Indicates the maximum number of iterations; Indicates the improved pheromone volatilization rate;

[0074] S36, inputting the normalized input feature variables into the data-driven model to obtain a predicted value, taking the root mean square error between the predicted value and the actual value as the fitness function, and calculating the fitness value; using the number of iterations as the current constraint condition to determine whether the preset maximum number of iterations is met, if it is met, then the iteration is stopped and the optimal solution is output; if it is not met, then the iteration counting is continued;

[0075] S37. Update the input weights and biases of the extreme learning machine according to the optimal solution, calculate the output weights of the hidden layer, and complete the construction process of the hot-rolled strip crown data-driven model.

[0076] S4. Use the entropy weight method to calculate the information entropy of the mechanical model and the information entropy of the data-driven model; calculate the initial weight coefficient of the mechanical model based on the information entropy of the mechanical model; calculate the initial weight coefficient of the data-driven model based on the information entropy of the data-driven model, and establish a hybrid model through the weighted summation method.

[0077] Optionally, the specific implementation process of S4 includes S41-S44:

[0078] S41. Obtain the calculated value of the convexity mechanism according to the hot-rolled strip convexity mechanism model; obtain the predicted value of the convexity according to the hot-rolled strip convexity data-driven model; calculate the absolute error between the calculated value of the convexity mechanism and the actual value of the hot-rolled strip, calculate the absolute error between the predicted value of the convexity and the actual value, and create an error decision matrix, which is expressed by the following formula (8):

[0079] (8)

[0080] Where n represents the number of samples; m represents the number of models to be evaluated; X represents the error decision matrix;

[0081] S42. Normalize the error data of each model by calculating the relative weight of the error decision matrix to obtain a normalized error matrix, and calculate the information entropy of the normalized error matrix, which is expressed by the following formulas (9)-(10):

[0082] (9)

[0083] (10)

[0084] in, represents the normalized error matrix; represents the information entropy of the normalized error matrix;

[0085] S43. Calculate the initial weight coefficient of the hot-rolled strip crown mechanism model based on the information entropy of the normalized error matrix; calculate the weight coefficient of the hot-rolled strip crown data-driven model based on the information entropy of the normalized error matrix; wherein the formula for calculating the weight coefficient is expressed by the following formula (11):

[0086] (11)

[0087] in, represents the weight coefficient;

[0088] S44. Based on the initial weight coefficient of the hot-rolled strip crown mechanism model and the initial weight coefficient of the hot-rolled strip crown data-driven model, a weighted summation method is used to construct a hybrid model, which is expressed by the following formula (12):

[0089] (12)

[0090] Where Y represents the output of the hybrid model; Represents the crown mechanism model of hot rolled strip; Represents the data driven model of hot rolled strip crown; represents the initial weight coefficient of the hot-rolled strip crown mechanism model; Represents the initial weight coefficient of the hot-rolled strip crown data-driven model.

[0091] S5. Obtain the data of the current hot-rolled strip production of different rolling units and input them into the hybrid model. Obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

[0092] Optionally, the specific implementation process of S5 includes S51-S52:

[0093] S51. When the deviation of the mechanism model is greater than the deviation of the data-driven model, the weight coefficient of the mechanism model is updated by reducing the weight coefficient, which is expressed by the following formula (13):

[0094] (13)

[0095] in, Represents the updated weight coefficient of the hot-rolled strip crown mechanism model. hour, ; Represents the gain coefficient, the value range is [0,1]; Indicates actual value; Represents the updated weight coefficient of the hot-rolled strip crown data-driven model;

[0096] S52. When the deviation of the data-driven model is greater than the deviation of the mechanism model, the weight coefficient of the data-driven model is updated by reducing the weight coefficient, which is expressed by the following formula (14):

[0097] (14)

[0098] Optionally, the process of outputting the final hot-rolled strip crown prediction result is expressed by the following formula (15):

[0099] (15)

[0100] in, It represents the predicted value of the next hot-rolled strip mixing model; Indicates the calculated value of the crown of the next hot-rolled strip; Represents the predicted value of the crown data-driven model for the next roll of hot-rolled strip.

[0101] In a feasible implementation manner, a hot-rolled strip convexity prediction method based on a weight distribution strategy hybrid modeling proposed in an embodiment of the present invention is adopted. Figure 3 、 Figure 4 as well as Figure 5The degree of fit between the prediction results of different models and the actual value of the crown is shown, including: a comparison chart of the crown prediction accuracy of hot-rolled strip using a mechanism model, a comparison chart of the crown prediction accuracy of hot-rolled strip using a data-driven model for the crown of hot-rolled strip, and a comparison chart of the crown prediction accuracy of hot-rolled strip using a hybrid model provided by an embodiment of the present invention. As can be seen from the figures, the mechanism model has a large degree of dispersion, with many samples having absolute errors exceeding 5μm. The absolute error of the hot-rolled strip crown data-driven model is basically within 4μm, with some samples having absolute errors exceeding 3μm. The hybrid model based on weight distribution effectively reduces the impact of error transfer on model accuracy in the hybrid modeling by adding weight coefficients to the single model to correct the model results. The absolute errors are basically within 2μm. The fitting effect is significantly better than that of a single mechanism model or data-driven model, proving the effectiveness of the hybrid model. Therefore, the hot-rolled strip crown prediction method based on weight distribution hybrid modeling proposed in the present invention achieves high-precision and high-reliability hot-rolled strip crown prediction, providing important theoretical guidance for high-precision control of hot-rolled strip crown.

[0102] The hybrid model established in the embodiment of the present invention can effectively overcome the model accuracy limitations of the mechanism model due to oversimplified assumptions, while avoiding the shortcomings of the data-driven model that lacks integration of domain knowledge and is easily distorted when faced with complex industrial modeling. By calculating the weight coefficient, not only is the negative impact of the poor prediction effect of the mechanism model on the final convexity prediction value weakened, but the high contribution of the high-precision data-driven model to the final convexity prediction value is also retained. By constructing a hybrid model through a weight distribution strategy, high-precision prediction of the convexity of hot-rolled plates and strips is achieved, providing important decision-making support for the optimal control of the convexity of hot-rolled plates and strips.

[0103] The embodiment of the present invention first obtains historical production data during the rolling process of hot-rolled plates and strips; uses the historical production data to establish a hot-rolled plate and strip convexity mechanism model; secondly, uses an improved ant colony algorithm to optimize the extreme learning machine and constructs a hot-rolled plate and strip convexity data-driven model; uses the entropy weight method to calculate the information entropy of the mechanism model and the information entropy of the data-driven model; calculates the initial weight coefficient of the mechanism model according to the information entropy of the mechanism model; calculates the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establishes a hybrid model through the weighted summation method; finally, obtains the data of the current hot-rolled plate and strip production of different rolling units and inputs it into the hybrid model, obtains a first prediction result through the mechanism model; obtains a second prediction result through the data-driven model; calculates the deviation between the first prediction result and the actual true value obtained in advance; calculates the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjusts the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and outputs the final hot-rolled plate and strip convexity prediction result.

[0104] Compared with the traditional hybrid modeling method, the embodiment of the present invention, after constructing the convexity mechanism model and the high-precision data-driven model, solves the information entropy of the error decision matrix through the entropy weight method and adds weight coefficients to the convexity mechanism model and the data-driven model respectively to correct the model results, effectively reducing the impact of error transmission on model accuracy during the hybrid modeling process. The present invention can improve the accuracy of hot-rolled strip convexity prediction. The hybrid model established by the embodiment of the present invention can give full play to the characteristics of the mechanism model with practical physical significance, avoid the model results from violating the actual physical laws, and at the same time give play to the advantage of the data-driven model in achieving high-precision prediction, providing theoretical guidance for achieving high-precision hot-rolled strip convexity control in the hot rolling industry, which is of great significance for achieving high-precision production of hot-rolled strip.

[0105] Figure 6 This is a block diagram of a hot-rolled strip convexity prediction device based on a weight distribution strategy hybrid modeling according to an exemplary embodiment. The device is used in a hot-rolled strip convexity prediction method based on a weight distribution strategy hybrid modeling. Figure 6 The device includes a first acquisition unit 610, a first establishment unit 620, a construction unit 630, a second establishment unit 640 and an output unit 650.

[0106] The first acquisition unit 610 is used to acquire historical production data during the hot strip rolling process;

[0107] A first establishing unit 620 is configured to establish a hot-rolled strip crown mechanism model using the historical production data;

[0108] A construction unit 630 is used to optimize the extreme learning machine using an improved ant colony algorithm to construct a hot-rolled strip crown data-driven model;

[0109] The second establishing unit 640 is configured to calculate the information entropy of the mechanism model and the information entropy of the data-driven model using an entropy weight method; calculate the initial weight coefficient of the mechanism model based on the information entropy of the mechanism model; calculate the initial weight coefficient of the data-driven model based on the information entropy of the data-driven model, and establish a hybrid model using a weighted summation method;

[0110] The output unit 650 is used to obtain the data of the current hot-rolled strip production of different rolling units and input it into the hybrid model, obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

[0111] Optionally, the historical production data during the hot-rolled strip rolling process includes:

[0112] Mill entrance thickness, exit thickness, rolling speed, entrance temperature, final rolling temperature, workpiece width, rolling force of each stand, bending roll force, roll shifting amount, roll thermal crown and wear parameters.

[0113] Optionally, the use of the historical production data to establish a hot-rolled strip crown mechanism model includes:

[0114] Based on the historical production data of hot-rolled strip rolling, the force-deformation behavior of the hot-rolled strip rolling process is analyzed, and a hot-rolled strip crown mechanism model is established, which is expressed by the following formula (1):

[0115] (1)

[0116] in, Indicates the outlet convexity of hot-rolled strip, Indicates the rolling mill serial number; Indicates the inlet strip convexity correction coefficient; Indicates rolling force; Indicates the bending roll force; Indicates the transverse stiffness of the rolling mill; Indicates the lateral stiffness of the bending roller; Indicates the amount of roller shifting; Indicates the width of hot rolled strip; Indicates the thermal crown of the work roll; Indicates wear convexity; Indicates the initial roller crown; Indicates the correction coefficient of roller shifting; Indicates the strip width correction factor; Indicates the roll crown correction coefficient.

[0117] Optionally, the construction unit is used to:

[0118] Obtaining input characteristic variables of a hot-rolled strip crown data-driven model; wherein the input variables include: hot-rolled strip inlet thickness, outlet thickness, inlet temperature, final rolling temperature, strip width, rolling force, bending force, roll shifting, rolling speed, roll thermal crown, and wear;

[0119] Normalizing the input characteristic variables of the hot-rolled strip crown data-driven model to obtain normalized input characteristic variables;

[0120] Initialize the parameters of the extreme learning machine and the improved ant colony algorithm, and randomly set the initial positions of the ants. The parameters of the extreme learning machine include weights, biases, and the number of hidden layer nodes. The parameters of the improved ant colony algorithm include the number of ant populations, the maximum number of iterations, the pheromone volatility coefficient, the transition probability constant, and the pheromone release amount.

[0121] The path search is performed based on the randomly set initial position of the ants. The state transition probability is calculated based on the concentration of pheromones on the path, which is expressed by the following formula (2):

[0122] (2)

[0123] in, represents the probability that the kth ant moves from the i-th position to the j-th position at time t; represents the pheromone weight coefficient; represents the weight coefficient of the heuristic function; and ; s represents the remaining nodes to be visited except i; represents the set of remaining nodes to be visited by the k-th ant; represents the pheromone concentration on the path (i, j) at time t; represents a heuristic information, indicating the expected level of an ant moving from position i to position j;

[0124] According to the state transition probability, the ant position is updated; according to the updated ant position, the pheromone is updated; wherein the updated pheromone is the sum of the volatilized pheromone and the newly generated pheromone, which is expressed by the following formula (3)-formula (5):

[0125] (3)

[0126] (4)

[0127] (5)

[0128] in, represents the amount of pheromone released by the kth ant on the path (i, j); Q represents the total amount of pheromone released by the ant in one cycle; is the Euclidean distance length of the path passed by the kth ant; is the path taken by the kth ant in this iteration; γ is the volatility coefficient of pheromone, 0<γ<1; Indicates updated pheromone; Indicates the original pheromone; Indicates newly produced pheromones;

[0129] A dynamic adjustment mechanism for pheromone volatility is adopted to improve the original fixed pheromone volatility coefficient, so that the volatility changes with the iterative process of the algorithm and different search strategies are used at different stages. The improved pheromone volatility is expressed by the following formula (6):

[0130] (6)

[0131] in, The minimum pheromone volatilization rate is set, which is usually used in the early stage of the algorithm to enhance exploration; The maximum pheromone volatilization rate is set, which is usually used in the later stage of the algorithm to consolidate the optimal path; Indicates the current iteration number; Indicates the maximum number of iterations; Indicates the improved pheromone volatilization rate;

[0132] The normalized input feature variables are input into the data-driven model to obtain the predicted value. The root mean square error between the predicted value and the actual value is used as the fitness function to calculate the fitness value. The number of iterations is used as the current constraint to determine whether the preset maximum number of iterations is met. If it is met, the iteration is stopped and the optimal solution is output. If it is not met, the iteration count continues.

[0133] The input weights and biases of the extreme learning machine are updated according to the optimal solution, the output weights of the hidden layer are calculated, and the construction process of the hot-rolled strip crown data-driven model is completed.

[0134] Optionally, the first establishing unit is configured to:

[0135] The convexity mechanism calculation value is obtained according to the hot-rolled strip convexity mechanism model; the convexity prediction value is obtained according to the hot-rolled strip convexity data-driven model; the absolute error between the convexity mechanism calculation value and the actual value of the hot-rolled strip is calculated, and the absolute error between the convexity prediction value and the actual value is calculated, and the error decision matrix is ​​created, which is expressed by the following formula (7):

[0136] (7)

[0137] Where n represents the number of samples; m represents the number of models to be evaluated; X represents the error decision matrix;

[0138] By calculating the relative weight of each model error data in the error decision matrix and performing normalization processing, a normalized error matrix is ​​obtained, and the information entropy of the normalized error matrix is ​​calculated, which is expressed by the following formulas (8)-(9):

[0139] (8)

[0140] (9)

[0141] in, represents the normalized error matrix; represents the information entropy of the normalized error matrix;

[0142] The initial weight coefficient of the hot-rolled strip crown mechanism model is calculated based on the information entropy of the normalized error matrix; the weight coefficient of the hot-rolled strip crown data-driven model is calculated based on the information entropy of the normalized error matrix; wherein, the formula for calculating the weight coefficient is expressed by the following formula (10):

[0143] (10)

[0144] in, represents the weight coefficient;

[0145] According to the initial weight coefficients of the hot-rolled strip crown mechanism model and the initial weight coefficients of the hot-rolled strip crown data-driven model, a weighted summation method is used to construct a hybrid model, which is expressed by the following formula (11):

[0146] (11)

[0147] Where Y represents the output of the hybrid model; Represents the crown mechanism model of hot rolled strip; Represents the data driven model of hot rolled strip crown; represents the initial weight coefficient of the hot-rolled strip crown mechanism model; Represents the initial weight coefficient of the hot-rolled strip crown data-driven model.

[0148] Optionally, dynamically adjusting the initial weight coefficients of the mechanism model and the initial weight coefficients of the data-driven model respectively by the deviation includes:

[0149] When the deviation of the mechanism model is greater than the deviation of the data-driven model, the weight coefficient of the mechanism model is updated by reducing the weight coefficient, which is expressed by the following formula (12):

[0150] (12)

[0151] in, Represents the updated weight coefficient of the hot-rolled strip crown mechanism model. hour, ; Represents the gain coefficient, the value range is [0,1]; Indicates actual value; Represents the updated weight coefficient of the hot-rolled strip crown data-driven model;

[0152] When the deviation of the data-driven model is greater than the deviation of the mechanism model, the weight coefficient of the data-driven model is updated by reducing the weight coefficient, which is expressed by the following formula (13):

[0153] (13)

[0154] Optionally, the process of outputting the final hot-rolled strip crown prediction result is expressed by the following formula (14):

[0155] (14)

[0156] in, It represents the predicted value of the next hot-rolled strip mixing model; Indicates the calculated value of the crown of the next hot-rolled strip; Represents the predicted value of the crown data-driven model for the next roll of hot-rolled strip.

[0157] The embodiment of the present invention first obtains historical production data during the rolling process of hot-rolled plates and strips; uses the historical production data to establish a hot-rolled plate and strip convexity mechanism model; secondly, uses an improved ant colony algorithm to optimize the extreme learning machine and constructs a hot-rolled plate and strip convexity data-driven model; uses the entropy weight method to calculate the information entropy of the mechanism model and the information entropy of the data-driven model; calculates the initial weight coefficient of the mechanism model according to the information entropy of the mechanism model; calculates the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establishes a hybrid model through the weighted summation method; finally, obtains the data of the current hot-rolled plate and strip production of different rolling units and inputs it into the hybrid model, obtains a first prediction result through the mechanism model; obtains a second prediction result through the data-driven model; calculates the deviation between the first prediction result and the actual true value obtained in advance; calculates the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjusts the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and outputs the final hot-rolled plate and strip convexity prediction result.

[0158] Compared with the traditional hybrid modeling method, the embodiment of the present invention, after constructing the convexity mechanism model and the high-precision data-driven model, solves the information entropy of the error decision matrix through the entropy weight method and adds weight coefficients to the convexity mechanism model and the data-driven model respectively to correct the model results, effectively reducing the impact of error transmission on model accuracy during the hybrid modeling process. The present invention can improve the accuracy of hot-rolled strip convexity prediction. The hybrid model established by the embodiment of the present invention can give full play to the characteristics of the mechanism model with practical physical significance, avoid the model results from violating the actual physical laws, and at the same time give play to the advantage of the data-driven model in achieving high-precision prediction, providing theoretical guidance for achieving high-precision hot-rolled strip convexity control in the hot rolling industry, which is of great significance for achieving high-precision production of hot-rolled strip.

[0159] Figure 7 Schematic diagram of a hot-rolled strip convexity prediction device based on hybrid modeling of weight distribution strategy provided by an embodiment of the present invention. Figure 7 As shown, the hot-rolled strip convexity prediction device based on the weight distribution strategy hybrid modeling may include the above Figure 6 The hot-rolled strip crown prediction device based on weight distribution strategy hybrid modeling is shown. Optionally, the hot-rolled strip crown prediction device 710 based on weight distribution strategy hybrid modeling may include a first processor 2001 .

[0160] Optionally, the hot-rolled strip convexity prediction device 710 based on weight distribution strategy hybrid modeling may further include a memory 2002 and a transceiver 2003 .

[0161] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0162] The following combination Figure 7 The components of the hot-rolled strip crown prediction device 710 based on the weight distribution strategy hybrid modeling are specifically introduced:

[0163] The first processor 2001 is the control center of the hot-rolled strip convexity prediction device 710 based on the weight distribution strategy hybrid modeling, and can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), or application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs) or one or more field programmable gate arrays (FPGAs).

[0164] Optionally, the first processor 2001 can perform various functions of the hot-rolled strip convexity prediction device 710 based on weight distribution strategy hybrid modeling by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002.

[0165] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 7 CPU0 and CPU1 are shown in FIG.

[0166] In a specific implementation, as an embodiment, the hot-rolled strip convexity prediction device 710 based on the weight distribution strategy hybrid modeling may also include multiple processors, such as Figure 7 1 and 2. The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0167] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled by the first processor 2001 for execution. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0168] Alternatively, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or exist independently and accessed through the interface circuit ( Figure 7 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0169] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0170] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 7 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0171] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or can exist independently and be used to predict the convexity of the hot-rolled strip 710 through the interface circuit ( Figure 7 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0172] It should be noted that Figure 7 The structure of the hot-rolled strip convexity prediction device 710 based on hybrid modeling of weight distribution strategy shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0173] In addition, the technical effects of the hot-rolled strip convexity prediction device 710 based on weight distribution strategy hybrid modeling can refer to the technical effects of the hot-rolled strip convexity prediction method based on weight distribution strategy hybrid modeling described in the above method embodiment, and will not be repeated here.

[0174] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc.

[0175] It should also be understood that the memory in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0176] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0177] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0178] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0179] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0181] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0182] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0185] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0186] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A hot-rolled strip crown prediction method based on hybrid modeling of weight distribution strategy, characterized in that: The method comprises: S1. Obtain historical production data during the hot-rolled strip rolling process; S2. Using the historical production data, establish a hot-rolled strip crown mechanism model; S3. Use the improved ant colony algorithm to optimize the extreme learning machine and build a data-driven model for the crown of hot-rolled strips; S4. Calculate the information entropy of the mechanism model and the information entropy of the data-driven model using an entropy weight method; calculate the initial weight coefficient of the mechanism model based on the information entropy of the mechanism model; calculate the initial weight coefficient of the data-driven model based on the information entropy of the data-driven model, and establish a hybrid model using a weighted summation method; S5. Obtain the data of the current hot-rolled strip production of different rolling units and input them into the hybrid model. Obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

2. The hot-rolled strip convexity prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The historical production data of the hot-rolled strip rolling process of S1 includes: Mill entrance thickness, exit thickness, rolling speed, entrance temperature, final rolling temperature, workpiece width, rolling force of each stand, bending roll force, roll shifting amount, roll thermal crown and wear parameters.

3. The hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The S2 uses the historical production data to establish a hot-rolled strip crown mechanism model, including: Based on the historical production data of hot-rolled strip rolling, the force-deformation behavior of the hot-rolled strip rolling process is analyzed, and a hot-rolled strip crown mechanism model is established, which is expressed by the following formula (1): （1） in, Indicates the outlet convexity of hot-rolled strip, Indicates the rolling mill serial number; Indicates the inlet strip convexity correction coefficient; Indicates rolling force; Indicates the bending roll force; Indicates the transverse stiffness of the rolling mill; Indicates the lateral stiffness of the bending roller; Indicates the amount of roller shifting; Indicates the width of hot rolled strip; Indicates the thermal crown of the work roll; Indicates wear convexity; Indicates the initial roller crown; Indicates the correction coefficient of roller shifting; Indicates the strip width correction factor; Indicates the roll crown correction coefficient.

4. The hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The S3 uses an improved ant colony algorithm to optimize the extreme learning machine and build a data-driven model for the crown of hot-rolled strips, including: S31. Obtain input characteristic variables of a hot-rolled strip crown data-driven model; wherein the input variables include: hot-rolled strip inlet thickness, outlet thickness, inlet temperature, final rolling temperature, strip width, rolling force, bending force, roll shifting, rolling speed, roll thermal crown, and roll wear; S32, normalizing the input characteristic variables of the hot-rolled strip crown data-driven model to obtain normalized input characteristic variables; S33, initializing the parameters of the extreme learning machine and the improved ant colony algorithm, and randomly setting the initial positions of the ants; wherein the parameters of the extreme learning machine include: weight, bias, and number of hidden layer nodes; wherein the parameters of the improved ant colony algorithm include: number of ant populations, maximum number of iterations, pheromone volatility coefficient, transition probability constant, and pheromone release amount; S34. Perform path search based on the randomly set initial position of the ants, and calculate the state transition probability based on the concentration of pheromones on the path, which is expressed by the following formula (2): （2） in, represents the probability that the kth ant moves from the i-th position to the j-th position at time t; represents the pheromone weight coefficient; represents the weight coefficient of the heuristic function; and ; s represents the remaining nodes to be visited except i; represents the set of remaining nodes to be visited by the k-th ant; represents the pheromone concentration on the path (i, j) at time t; represents a heuristic information, indicating the expected level of an ant moving from position i to position j; S34. Update the ant position according to the state transition probability; update the pheromone according to the updated ant position; wherein the updated pheromone is the sum of the volatilized pheromone and the newly generated pheromone, which is expressed by the following formula (3) to formula (5): （3） （4） （5） in, represents the amount of pheromone released by the kth ant on the path (i, j); Q represents the total amount of pheromone released by the ant in one cycle; is the Euclidean distance length of the path passed by the kth ant; is the path taken by the kth ant in this iteration; γ is the volatility coefficient of pheromone, 0<γ<1; Indicates updated pheromone; Indicates the original pheromone; Indicates newly produced pheromones; S35. A dynamic adjustment mechanism for pheromone volatility is used to improve the original fixed pheromone volatility coefficient, so that the volatility changes with the iterative process of the algorithm and different search strategies are used at different stages. The improved pheromone volatility is expressed by the following formula (6): （6） in, The minimum pheromone volatilization rate is set, which is usually used in the early stage of the algorithm to enhance exploration; The maximum pheromone volatilization rate is set, which is usually used in the later stage of the algorithm to consolidate the optimal path; Indicates the current iteration number; Indicates the maximum number of iterations; Indicates the improved pheromone volatilization rate; S36, inputting the normalized input feature variables into the data-driven model to obtain a predicted value, taking the root mean square error between the predicted value and the actual value as the fitness function, and calculating the fitness value; using the number of iterations as the current constraint condition to determine whether the preset maximum number of iterations is met, if it is met, then the iteration is stopped and the optimal solution is output; if it is not met, then the iteration counting is continued; S37. Update the input weights and biases of the extreme learning machine according to the optimal solution, calculate the output weights of the hidden layer, and complete the construction process of the hot-rolled strip crown data-driven model.

5. The hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The step S4 uses an entropy weight method to calculate the information entropy of the mechanism model and the information entropy of the data-driven model; and calculates an initial weight coefficient of the mechanism model based on the information entropy of the mechanism model; Calculating the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establishing a hybrid model by a weighted summation method, including: S41. Obtain the calculated value of the convex mechanism according to the hot-rolled strip convex mechanism model; obtain the predicted value of the convex according to the hot-rolled strip convex data-driven model; calculate the absolute error between the calculated value of the convex mechanism and the actual value of the hot-rolled strip, calculate the absolute error between the predicted value of the convex and the actual value, and create an error decision matrix, which is expressed by the following formula (7): （7） Where n represents the number of samples; m represents the number of models to be evaluated; X represents the error decision matrix; S42. Normalize the error data of each model by calculating the relative weight of the error decision matrix to obtain a normalized error matrix, and calculate the information entropy of the normalized error matrix, which is expressed by the following formulas (8)-(9): （8） （9） in, represents the normalized error matrix; represents the information entropy of the normalized error matrix; S43. Calculate the initial weight coefficient of the hot-rolled strip crown mechanism model based on the information entropy of the normalized error matrix; calculate the initial weight coefficient of the hot-rolled strip crown data-driven model based on the information entropy of the normalized error matrix; wherein the formula for calculating the weight coefficient is expressed by the following formula (10): （10） in, represents the weight coefficient; S44. Based on the initial weight coefficient of the hot-rolled strip crown mechanism model and the initial weight coefficient of the hot-rolled strip crown data-driven model, a weighted summation method is used to construct a hybrid model, which is expressed by the following formula (11): （11） Where Y represents the output of the hybrid model; Represents the crown mechanism model of hot rolled strip; Represents the data driven model of hot rolled strip crown; represents the initial weight coefficient of the hot-rolled strip crown mechanism model; Represents the initial weight coefficient of the hot-rolled strip crown data-driven model.

6. The hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The S5 dynamically adjusts the initial weight coefficients of the mechanism model and the initial weight coefficients of the data-driven model respectively through the deviation amount, including: S51. When the deviation of the mechanism model is greater than the deviation of the data-driven model, the weight coefficient of the mechanism model is updated by reducing the weight coefficient, which is expressed by the following formula (12): （12） in, Represents the updated weight coefficient of the hot-rolled strip crown mechanism model. hour, ; Represents the gain coefficient, the value range is [0,1]; Indicates actual value; Represents the updated weight coefficient of the hot-rolled strip crown data-driven model; S52. When the deviation of the data-driven model is greater than the deviation of the mechanism model, the weight coefficient of the data-driven model is updated by reducing the weight coefficient, which is expressed by the following formula (13): （13）。 7. The hot-rolled strip crown prediction method based on weight distribution strategy hybrid modeling according to claim 1 is characterized in that: The process of outputting the final hot-rolled strip crown prediction result is expressed by the following formula (14): （14） in, It represents the predicted value of the next hot-rolled strip mixing model; Indicates the calculated value of the crown of the next hot-rolled strip; Represents the predicted value of the crown data-driven model for the next roll of hot-rolled strip.

8. A hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling, the hot-rolled strip convexity prediction device based on weight distribution strategy hybrid modeling is used to implement the hot-rolled strip convexity prediction method based on weight distribution strategy hybrid modeling according to any one of claims 1 to 7, characterized in that: The device comprises: The first acquisition unit is used to acquire historical production data during the hot-rolled strip rolling process; A first establishing unit is used to establish a hot-rolled strip crown mechanism model using the historical production data; A construction unit is used to optimize the extreme learning machine using an improved ant colony algorithm to construct a data-driven model for the crown of hot-rolled strips; The second establishing unit is configured to calculate the information entropy of the mechanism model and the information entropy of the data-driven model by using an entropy weight method; calculate the initial weight coefficient of the mechanism model according to the information entropy of the mechanism model; calculate the initial weight coefficient of the data-driven model according to the information entropy of the data-driven model, and establish a hybrid model by a weighted summation method; The output unit is used to obtain the data of the current hot-rolled strip production of different rolling units and input it into the hybrid model, obtain the first prediction result through the mechanism model; obtain the second prediction result through the data-driven model; calculate the deviation between the first prediction result and the actual true value obtained in advance; calculate the deviation between the second prediction result and the actual true value obtained in advance; dynamically adjust the initial weight coefficient of the mechanism model and the initial weight coefficient of the data-driven model according to the deviation, and output the final hot-rolled strip convexity prediction result.

9. A hot-rolled strip crown prediction device based on weight distribution strategy hybrid modeling, characterized in that: The hot-rolled strip crown prediction device based on weight distribution strategy hybrid modeling includes: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 7.

Images