Numerical control four-roller plate rolling machine accurate rolling method

Abstract

The application discloses a numerical control four-roller plate rolling machine precise rolling method, including obtaining the performance parameters and process parameters of the plate to be bent and rolling, inputting the parameters into a mathematical model to determine the initial roller position; moving the side roller to the latest roller position and starting rolling; collecting and processing images through a machine vision system to fit the real-time curvature radius; if the deviation of the real-time curvature radius of the plate and the expected curvature radius is not within the allowable error range, correcting the roller position according to the deviation and continuing rolling; after the precision is reached, storing all the precise parameters of the rolling in a database as a group of data; when a new round of plate rolling is started, if the data volume does not reach the set value, continuing to roll the plate by using the machine vision method, etc. The application can efficiently and accurately realize the bending and rolling of the plate, and realizes the automation and intelligentization of the plate bending and rolling by the method of fusing the machine vision and the machine learning algorithm.

Description

Technical Field

[0001] This invention relates to the field of CNC four-roll plate rolling machine technology, specifically to a precise rolling method for a CNC four-roll plate rolling machine. Background Technology

[0002] A CNC four-roll plate bending machine is a machine used to bend metal sheets into a specific curvature. It typically consists of an upper roll, a lower roll, and two side rolls. The upper roll is driven by a motor to rotate, the lower roll moves up and down via hydraulic or motor-driven mechanisms, and the side rolls are hydraulically adjustable to ensure rolling precision. CNC four-roll plate bending machines are also equipped with a digital control system, which allows for automated control through parameter input, improving production efficiency and product quality. CNC four-roll plate bending machines are widely used in shipbuilding, wind power, construction, petrochemical, and other industries for manufacturing various metal sheets, such as steel, aluminum, and stainless steel.

[0003] Although CNC four-roll plate rolling machines can improve production efficiency and product quality to a certain extent, based on numerous practical application cases, the precision of the plates rolled by CNC four-roll plate rolling machines still does not meet the expected requirements. The curvature radius of the rolled plates needs to be measured manually with the aid of tools and the precision needs to be improved through multiple rolling passes. This rolling method wastes a lot of manpower and time.

[0004] Furthermore, for sheet metal with known material performance parameters and forming process parameters, CNC four-roll plate bending machines with certain structural parameters still repeatedly roll the material using traditional methods. They cannot calculate the side roll positions based on a large amount of data from rolled sheet metal, and the level of automation and intelligence does not meet practical needs. Therefore, a new rolling method is still needed for CNC four-roll plate bending machines to make the bending process faster and more efficient, and the rolling results more accurate. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of this section, the abstract, and the title, and such simplifications or omissions should not be used to limit the scope of the invention.

[0006] Therefore, the purpose of this invention is to provide a precise rolling method for a CNC four-roll plate bending machine that integrates machine vision and machine learning algorithms. By integrating machine vision and an XGBoost-based intelligent coupled influence analysis side roll position prediction model, the bending process of the CNC four-roll plate bending machine becomes faster and more accurate, saving significant manpower and time costs, and achieving automation and intelligence in plate bending.

[0007] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution:

[0008] A method for precise rolling using a CNC four-roll plate rolling machine, comprising:

[0009] S1. Obtain the material performance parameters and forming process parameters of the sheet material to be bent and rolled;

[0010] S2. Input the known parameters into the mathematical model to determine the initial side roller position;

[0011] S3. The host computer controls the hydraulic servo through the motion control card, so that the side roller moves to the latest roller position;

[0012] S4. The upper roller of the four-roll plate rolling machine rotates to roll the plate and form the newly rolled plate section.

[0013] S5. The area array CCD camera acquires images of the newly rolled sheet material after springback.

[0014] S6. Perform image processing in the host computer vision system and fit the real-time radius of curvature of the latest rolled sheet material.

[0015] S7. Determine whether the deviation between the real-time radius of curvature of the latest rolled sheet portion and the expected radius of curvature is within the allowable error range.

[0016] S8. Correct the side roller position based on the deviation between the real-time curvature radius of the latest rolled sheet and the desired curvature radius;

[0017] S9. Store the material performance parameters, forming process parameters, and corresponding optimal roll position parameters of the rolled sheet as a set of data in the database of the host computer.

[0018] S10. When starting a new round of sheet metal rolling, determine whether the amount of data for each parameter accumulated in the database meets the set value.

[0019] S11. Continue to use machine vision methods to roll up sheets with different material performance parameters and forming process parameters;

[0020] S12. Using machine learning methods to roll the sheet, construct an intelligent coupling influence analysis side roll position prediction model based on XGBoost, and use 70% of the database as the training set and 30% as the test set to obtain the trained and optimized roll position prediction model.

[0021] S13. Input the material performance parameters and forming process parameters of the plate to be bent and rolled into the roller position prediction model. The model outputs the side roller position, so that for a plate rolling machine with certain structural parameters, the plate can be bent and rolled in one step.

[0022] In a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S1, the material performance parameters of the plate to be bent and rolled include the yield strength σ of the plate. s The sheet thickness t, sheet width b, and forming process parameters include sheet rolling feed speed V and sheet desired radius of curvature ρ.

[0023] As a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S2, the mathematical model is as follows: P = Y - Y1

[0024] Where P is the side roller position parameter, Y is the hydraulic cylinder position corresponding to the side roller position in the initial state, and Y1 is the hydraulic cylinder position corresponding to the latest side roller position.

[0025]

[0026]

[0027] Where g, k, μ, and v are structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual;

[0028]

[0029]

[0030] The radius of the upper roller of the plate rolling machine is R. a The lower roller radius is R b The radius of the left roller is R. c The radius of the right roller is R. d ,R a =R b =R1,R c =R d =R2, d and r are the structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual;

[0031]

[0032]

[0033]

[0034] Where, ρ r This refers to the inner diameter of the sheet metal before springback.

[0035]

[0036] γ=δ-ε

[0037]

[0038] L'=(ρ r +t+R1)sinε

[0039]

[0040]

[0041] Where K e K is the relative strengthening coefficient of the sheet material, E is the elastic modulus of the sheet material, and K is the relative strengthening coefficient of the sheet material. s K is the shape factor of the plate section, generally for rectangular sections. s =1.5, ρ i K represents the inner diameter of the sheet metal after springback bending. e E is obtained by referring to the performance parameter tables of different materials;

[0042]

[0043] Where ρ is the desired radius of curvature of the sheet material to be bent.

[0044] In a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S3, an industrial control computer is used as the host computer, and a program is developed in Visual Studio to control the motion control card. The motion control card controls the hydraulic servo of the left and right side rolls to make the side rolls reach the predetermined position.

[0045] In a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S5, a binocular CCD camera is used to acquire images of the plate under adjustable light source illumination.

[0046] As a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S6, the image is preprocessed, grayscaled, binarized, and morphologically processed by the image processing program developed in the host computer, then the discrete curvature is calculated using the difference operator, and finally the curve is fitted to calculate the real-time radius of curvature.

[0047] As a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S7, it is determined whether the deviation between the real-time radius of curvature of the latest rolled plate part and the expected radius of curvature is within the allowable error range. The allowable error range is determined according to the actual situation such as the precision of the four-roll plate rolling machine and the actual rolling precision requirements. If the deviation is within the set allowable error range, it is considered to meet the expected radius of curvature, and then step S9 is performed.

[0048] If the deviation exceeds the set allowable error range, it is considered that the desired radius of curvature has not been met, and then step S8 is performed, whereby the host computer calculates the corrected side roller position, followed by steps S3, S4, S5, and S6. Then, the judgment is made again, and the cycle is repeated until the deviation reaches within the allowable range.

[0049] In a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S8, the deviation is corrected by controlling the side rollers to continuously advance through a motion control card. Each advance is accompanied by a visual recognition of the real-time curvature radius of the plate until the deviation is within the allowable error range.

[0050] In a preferred embodiment of the precision rolling method of the CNC four-roll plate rolling machine described in this invention, in step S10, when the judgment is "yes", the data volume of each parameter must reach: 5 different yield strengths σ s 5 different sheet thicknesses t, 5 different sheet widths b, 3 different feed speeds V, 5 different sheet desired radii of curvature ρ, 50 different side roller position parameters P; otherwise, judge as "no";

[0051] For data stored in the database, the data format is as follows:

[0052]

[0053] In a preferred embodiment of the precision rolling method for a CNC four-roll plate rolling machine described in this invention, in step S12, when using a machine learning algorithm, an intelligent coupling influence analysis side roll position prediction model based on XGBoost is constructed. The XGBoost ensemble learning method is adopted, and a regression tree is used as the base model to perform classification regression of the multi-factor coupling influence relationship in bending and rolling. Through joint decision-making using associated regression trees, a complex influence coupling analyzer (SC) is constructed based on the gradient boosting method. The constructed multi-factor coupling influence objective function is: P = SC(t, b, σ). s ,V,ρ)

[0054] Where SC is the forming relationship function under coupling effect. Data features of different orders of magnitude do not affect the model results and do not need to be normalized. 70% of the data in the database is used as the training set and the other 30% is used as the test set. The XGBoost-based roll position prediction model is trained with parameters. The evaluation index of the model is the mean square error of the side roll position. The training of roll position prediction is based on small sample data and requires the addition of a regularization term. The weights can be used as hyperparameters. Finally, the hyperparameters are adjusted by Bayesian optimization until the set number of iterations is reached.

[0055] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention not only solves the problems that the curvature radius of rolled plates needs to be measured manually with the aid of tools, and that the side roller positions cannot be calculated based on a large amount of rolled plate data for plates with known material performance parameters and forming process parameters, but also proposes a method that integrates machine vision and machine learning algorithms and applies it to CNC four-roll plate rolling machines. By using machine vision to roll plates, a large amount of data is obtained and stored in a database, with 70% used as a training set and the other 30% as a test set. The parameters in the XGBoost-based roller position prediction model are trained and optimized, and the hyperparameters are adjusted using Bayesian optimization, thereby improving the accuracy and generalization ability of the model. This enables plate rolling machines with certain structural parameters to accurately roll and form plates in one operation when the material performance parameters and forming process parameters are known, as well as automating and intelligentizing bending and rolling forming. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0057] Figure 1 A flowchart of a precision rolling method for a CNC four-roll plate rolling machine that integrates machine vision and machine learning algorithms is provided in an embodiment of the present invention.

[0058] Figure 2 This is a schematic diagram of the planar geometric relationship of the mathematical model for determining the initial roller position provided in an embodiment of the present invention;

[0059] Figure 3 A schematic diagram showing the changes in the CCD camera image acquisition area and the side roller position during the bending and forming of a sheet metal in a four-roll plate bending machine, provided in an embodiment of the present invention.

[0060] Figure 4 This is a schematic diagram of the method for predicting the position of the side roll in the intelligent coupling influence analysis of XGBoost provided in an embodiment of the present invention. Detailed Implementation

[0061] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0062] Secondly, the present invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not according to the usual scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of the present invention. In addition, actual fabrication should include three-dimensional spatial dimensions of length, width, and depth.

[0063] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0064] This invention provides a precise rolling method for a CNC four-roll plate bending machine that integrates machine vision and machine learning algorithms. By combining machine vision with an XGBoost-based intelligent coupled influence analysis side roll position prediction model, the bending process of the CNC four-roll plate bending machine becomes faster and more accurate, saving significant manpower and time costs, and achieving automation and intelligence in plate bending.

[0065] This invention provides a precise rolling method for a CNC four-roll plate rolling machine that integrates machine vision and machine learning algorithms, such as... Figure 1 As shown, it mainly includes:

[0066] S1. Obtain the material performance parameters and forming process parameters of the sheet material to be bent and rolled;

[0067] S2. Input the known parameters into the mathematical model to determine the initial side roller position;

[0068] S3. The industrial control computer (host computer) controls the hydraulic servo through the motion control card, so that the side roller moves to the latest roller position;

[0069] S4. The upper roller of the four-roll plate rolling machine rotates to roll the plate and form the newly rolled plate section.

[0070] S5. The area array CCD camera acquires images of the newly rolled sheet material after springback.

[0071] S6. Perform image processing in the host computer vision system and fit the real-time radius of curvature of the latest rolled sheet material.

[0072] S7. Determine whether the deviation between the real-time radius of curvature of the latest rolled sheet portion and the expected radius of curvature is within the allowable error range.

[0073] S8. Correct the side roller position based on the deviation between the real-time curvature radius of the latest rolled sheet and the desired curvature radius;

[0074] S9. Store the material performance parameters, forming process parameters, and corresponding optimal roll position parameters of the rolled sheet as a set of data in the database of the host computer.

[0075] S10. When starting a new round of sheet metal rolling, determine whether the amount of data for each parameter accumulated in the database meets the set value.

[0076] S11. Continue to use machine vision methods to roll up sheets with different material performance parameters and forming process parameters;

[0077] S12. Using machine learning methods to roll the sheet, construct an intelligent coupling influence analysis side roll position prediction model based on XGBoost, and use 70% of the database as the training set and 30% as the test set to obtain the trained and optimized roll position prediction model.

[0078] S13. Input the material performance parameters and forming process parameters of the plate to be bent and rolled into the roller position prediction model. The model outputs the side roller position, so that for a plate rolling machine with certain structural parameters, the plate can be bent and rolled in one step.

[0079] In some optional embodiments, in step S1, the material performance parameters of the sheet material to be bent and rolled include the sheet yield strength σ. s The sheet thickness t, sheet width b, and forming process parameters include sheet rolling feed speed V and sheet desired radius of curvature ρ.

[0080] In the embodiments, reference Figure 2 In step S2, the mathematical model is derived from the planar geometric relationship between the upper roller, lower roller, side roller, and the rolled sheet material of the CNC four-roll plate bending machine (side roller arc feed type). To simplify the mathematical model, it is assumed that the neutral layer remains unchanged during the bending process, the influence of the sheet width is ignored, the influence of the weight of the elongated end of the sheet material is not considered, the sheet material does not slip during the bending process, the sheet thickness remains unchanged during the bending process, and the initial state is defined as the close contact between the side roller and the upper roller. The mathematical model is as follows:

[0081] P = Y - Y1

[0082] Where P is the side roller position parameter, Y is the hydraulic cylinder position corresponding to the side roller position in the initial state, and Y1 is the hydraulic cylinder position corresponding to the latest side roller position.

[0083]

[0084]

[0085] Where g, k, μ, and v are structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual. μ = ∠ADH, v = ∠BDK;

[0086]

[0087]

[0088] The radius of the upper roller of the plate rolling machine is R. a The lower roller radius is R b The radius of the left roller is R. c The radius of the right roller is R. d ,R a =R b =R1,R c =R d =R2, where d and r are structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual.

[0089]

[0090]

[0091]

[0092] Where ρ r The inner diameter of the sheet metal before springback is the bending diameter.

[0093]

[0094] γ=δ-ε

[0095]

[0096]

[0097]

[0098] Where K e K is the relative strengthening coefficient of the sheet material, E is the elastic modulus of the sheet material, and K is the relative strengthening coefficient of the sheet material. s K is the shape factor of the plate section, generally for rectangular sections. s =1.5, ρ i K represents the inner diameter of the sheet metal after springback bending. e E is obtained by referring to the performance parameter tables of different materials;

[0099]

[0100] Where ρ is the desired radius of curvature of the sheet material to be bent.

[0101] In some alternative implementations, in step S3, an industrial control computer is used as the host computer, and a program is developed in Visual Studio to control the motion control card. The motion control card controls the hydraulic servo of the left and right side rollers to make the side rollers reach the predetermined position.

[0102] In some alternative implementations, refer to Figure 3In step S5, under adjustable light source illumination, a binocular CCD camera is used to acquire images of the board material. The CCD camera acquires the area of ​​the board material after bending and springing back.

[0103] In some optional implementations, in step S6, the image is preprocessed, grayscaled, binarized, and morphologically processed by an image processing program developed in the host computer, then the discrete curvature is calculated using a difference operator, and finally the curve is fitted to calculate the real-time radius of curvature.

[0104] In some optional implementations, in step S7, it is determined whether the deviation between the real-time radius of curvature of the latest rolled plate portion and the desired radius of curvature is within the allowable error range. The allowable error range is determined based on the actual situation such as the accuracy of the four-roll plate rolling machine and the actual rolling accuracy requirements. If the deviation is within the set allowable error range, it is considered that the desired radius of curvature is met, and then step S9 is performed. If the deviation exceeds the set allowable error range, it is considered that the desired radius of curvature is not met, and then step S8 is performed, whereby the host computer calculates the correction side roller position, and then steps S3, S4, S5, and S6 are performed. Then the judgment is made again, and the process is repeated iteratively until the deviation reaches within the allowable range.

[0105] In some alternative implementations, in step S8, the deviation is corrected by controlling the side rollers to continuously feed through a motion control card. Each feed is accompanied by a visual recognition of the real-time curvature radius of the board until the deviation is within the allowable error range.

[0106] In this embodiment, in step S10, when the determination is "yes", the data volume (set value) of each parameter must reach: 5 different yield strengths σ s The data stored in the database includes 5 different sheet thicknesses (t), 5 different sheet widths (b), 3 different feed speeds (V), 5 desired radii of curvature (ρ) of the sheet materials, and 50 different side roller position parameters. The data format is as follows:

[0107]

[0108] In this embodiment, reference Figure 4 In step S12, the prediction model refers to the XGBoost-based intelligent coupling influence analysis side roll position prediction model. The model employs the XGBoost ensemble learning method and uses a regression tree as the base model to classify and regress the coupling influence relationships of multiple factors in bending and rolling. Through joint decision-making using associated regression trees, a complex influence coupling analyzer (SC) is constructed based on the gradient boosting method. The constructed multi-factor coupling influence objective function is:

[0109] P=SC(t,b,σ s ,V,ρ)

[0110] Where SC is the shaping relationship function under coupling effect.

[0111] In this embodiment, data features of different orders of magnitude do not affect the model results and do not require normalization. 70% of the data in the database is used as the training set, and the other 30% is used as the test set. The XGBoost-based roll position prediction model is trained on the parameters, and the evaluation metric for the model is the mean square error of the side roll position. The training of roll position prediction is based on small sample data, so a regularization term needs to be added. The weights can be used as hyperparameters. Finally, the hyperparameters are adjusted through Bayesian optimization until the set number of iterations is reached.

[0112] This invention not only solves the problems of needing to manually measure the curvature radius of rolled plates with tools, and the inability to deduce the side roller positions based on a large amount of rolled plate data for plates with known material performance parameters and forming process parameters, but also proposes a method that integrates machine vision and machine learning algorithms for use in CNC four-roll plate rolling machines. By using machine vision to roll plates, a large amount of data is obtained and stored in a database. 70% of this data is used as a training set, and the remaining 30% as a test set. The parameters in the XGBoost-based roller position prediction model are trained and optimized. Bayesian optimization is used to adjust the hyperparameters, improving the model's accuracy and generalization ability. This enables plate rolling machines with certain structural parameters to accurately roll and form plates in one operation when the material performance parameters and forming process parameters are known, as well as automating and intelligentizing the bending and rolling process.

[0113] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for precise rolling using a CNC four-roll plate rolling machine, characterized in that, include: S1. Obtain the material performance parameters and forming process parameters of the sheet material to be bent and rolled; S2. Input the known parameters into the mathematical model to determine the initial side roller position; S3. The host computer controls the hydraulic servo through the motion control card, so that the side roller moves to the latest roller position; S4. The upper roller of the four-roll plate rolling machine rotates to roll the plate and form the newly rolled plate section. S5. The area array CCD camera acquires images of the newly rolled sheet material after springback. S6. Perform image processing in the host computer vision system and fit the real-time radius of curvature of the latest rolled sheet material. S7. Determine whether the deviation between the real-time radius of curvature of the latest rolled sheet portion and the expected radius of curvature is within the allowable error range; S8. Correct the side roller position based on the deviation between the real-time curvature radius of the latest rolled sheet and the desired curvature radius; S9. Store the material performance parameters, forming process parameters, and corresponding optimal roll position parameters of the rolled sheet as a set of data in the database of the host computer. S10. When starting a new round of sheet metal rolling, determine whether the amount of data for each parameter accumulated in the database meets the set value. S11. Continue to use machine vision methods to roll up sheets with different material performance parameters and forming process parameters; S12. Using machine learning methods to roll the sheet, construct an intelligent coupling influence analysis side roll position prediction model based on XGBoost, and use 70% of the database as the training set and 30% as the test set to obtain the trained and optimized roll position prediction model. S13. Input the material performance parameters and forming process parameters of the plate to be bent and rolled into the roller position prediction model. The model outputs the side roller position, so that for a plate rolling machine with certain structural parameters, the plate can be bent and rolled in one step.

2. The precise rolling method of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S1, the material performance parameters of the sheet material to be bent and rolled include the sheet yield strength σ. s The sheet thickness t, sheet width b, and forming process parameters include sheet rolling feed speed V and sheet desired radius of curvature ρ.

3. The precise rolling method of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S2, the mathematical model is as follows: P = Y - Y1 Where P is the side roller position parameter, Y is the hydraulic cylinder position corresponding to the side roller position in the initial state, and Y1 is the hydraulic cylinder position corresponding to the latest side roller position. Where g, k, μ, and v are structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual; The radius of the upper roller of the plate rolling machine is R. a The lower roller radius is R b The radius of the left roller is R. c The radius of the right roller is R. d ,R a =R b =R1,R c =R d =R2, d and r are the structural parameters of the plate rolling machine, which can be obtained from the corresponding CNC four-roll plate rolling machine structural parameter manual; Where, ρ r This refers to the inner diameter of the sheet metal before springback. γ=δ-ε L'=(ρ r +t+R1)sinε Where K e K is the relative strengthening coefficient of the sheet material, E is the elastic modulus of the sheet material, and K is the relative strengthening coefficient of the sheet material. s K is the shape factor of the plate section, where the plate section is rectangular. s =1.5, ρ i K represents the inner diameter of the sheet metal after springback bending. e E is obtained by referring to the performance parameter tables of different materials; Where ρ is the desired radius of curvature of the sheet material to be bent.

4. The precise rolling method of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S3, the industrial control computer is used as the host computer, and a program is developed in Visual Studio to control the motion control card. The motion control card controls the hydraulic servo of the left and right side rollers to make the side rollers reach the predetermined position.

5. The precise rolling method of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S5, under adjustable light source illumination, a binocular CCD camera is used to acquire images of the board material.

6. The precise rolling method of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S6, the image is preprocessed, grayscaled, binarized, and morphologically processed by the image processing program developed in the host computer. Then, the discrete curvature is calculated using the difference operator, and finally, the curve is fitted to calculate the real-time radius of curvature.

7. A method for precise rolling using a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S7, it is determined whether the deviation between the real-time radius of curvature of the latest rolled plate part and the expected radius of curvature is within the allowable error range. The allowable error range is determined according to the accuracy of the four-roll plate rolling machine and the actual rolling accuracy requirements. If the deviation is within the set allowable error range, it is considered to meet the desired radius of curvature, and then proceed to step S9; If the deviation exceeds the set allowable error range, it is considered that the desired radius of curvature has not been met, and then step S8 is performed, whereby the host computer calculates the corrected side roller position, followed by steps S3, S4, S5, and S6. Then, the judgment is made again, and the cycle is repeated until the deviation reaches within the allowable range.

8. A method for precise rolling of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S8, the deviation is corrected by controlling the side rollers to continuously feed through the motion control card. Each feed is accompanied by a visual recognition of the real-time curvature radius of the board until the deviation is within the allowable error range.

9. A method for precise rolling using a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S10, when the judgment is "yes", the data volume of each parameter must reach: 5 different yield strengths σ s The data must include: 5 different sheet thicknesses (t), 5 different sheet widths (b), 3 different feed speeds (V), 5 desired radii of curvature (ρ) of the sheet material, and 50 different side roller position parameters (P). Otherwise, it is judged as "No". For data stored in the database, the data format is as follows:

10. A method for precise rolling of a CNC four-roll plate rolling machine according to claim 1, characterized in that, In step S12, when using machine learning algorithms, an intelligent coupling influence analysis side roll position prediction model based on XGBoost is constructed. The XGBoost ensemble learning method is adopted, and a regression tree is used as the base model to perform classification regression of the multi-factor coupling influence relationship in bending and rolling. Through joint decision-making using associated regression trees, a coupling analyzer for complex influences is constructed based on the gradient boosting method. The constructed multi-factor coupling influence objective function is: P = SC(t, b, σ). s ,V,ρ) Where SC is the forming relationship function under coupling effect. Data features of different orders of magnitude do not affect the model results and do not need to be normalized. 70% of the data in the database is used as the training set and the other 30% is used as the test set. The XGBoost-based roll position prediction model is trained with parameters. The evaluation index of the model is the mean square error of the side roll position. The training of roll position prediction is based on small sample data and requires the addition of a regularization term. The weights are used as hyperparameters. Finally, the hyperparameters are adjusted by Bayesian optimization until the set number of iterations is reached.

Images