Large-tonnage bending machine parameterization optimization design method based on multi-target control

Abstract

The invention relates to the technical field of bending machines, in particular to a large-tonnage bending machine parameterization optimization design method based on multi-target control. Comprising the following steps of determining key structure design parameters of the large-tonnage bending machine, performing multi-factor experimental design and working condition analysis, performing simulation and test data correction, establishing a multi-element agent model, and performing nonlinear constraint optimization solution. According to the large-tonnage bending machine parameterization optimization design method developed on the basis of the multivariate agent model and the nonlinear constraint function, the bending machine optimization design efficiency is remarkably improved, and the structure research and development period is shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of bending machines, and particularly relates to a parametric optimization design method for large-tonnage bending machines based on multi-objective control. Background Art

[0002] At present, the proportion of non-standard parameters of large-tonnage bending machines is relatively high, and the design cycle is relatively long. Developing a parametric design method can effectively shorten the design cycle of non-standard products for designers. This design method analyzes the structural parameters of the workbench, slider, and wall panels of large-tonnage bending machines, aiming to minimize the total mass (cost) of the structure, while taking the stiffness and strength of the structure as constraints, and establishing a parametric optimal solution model based on multi-agent models and non-linear constraint equations. For this purpose, the parameters required for the instance model are determined, and the geometric relationship between objects is fully considered by using mathematical relationship expressions and the assembly relationship of the model itself. On this basis, by adjusting the size of the stiffeners, some positions with relatively large stress in the structure can be optimized, and a complex product structure can be expressed with as few parameters as possible. Therefore, how to achieve precise and rapid design of large-tonnage bending machines is a technical problem to be solved. Summary of the Invention

[0003] The problem to be solved by the present invention is to provide a parametric optimization design method for large-tonnage bending machines based on multi-objective control, which has fast design efficiency and high accuracy.

[0004] In view of the deficiencies of the prior art, the technical solution adopted by the present invention to solve its technical problems is: a parametric optimization design method for large-tonnage bending machines based on multi-objective control, specifically including the following steps:

[0005] S1. Determine the key structural design parameters of the large-tonnage bending machine: Based on the bending load F and the bending span L0, determine the key design parameters of the slider, workbench, and wall panels, including the slider height H hk , the thickness a of the stiffener at the oil cylinder yg , the height H of the neutral plate gzt , the thickness a of the stiffener at the connection between the wall panel and the workbench lj , the width K of the wall panel, and the thickness a of the throat stiffener hk1 ;

[0006] S2. Multi-factor experimental design and working condition analysis: Select sample points and set value ranges for the design variables in step S1, generate simulation and test experimental groups, and number them as h, h = 1, 2,..., Q, where Q is the total number of experiments that need to be simulated and tested;

[0007] S3. Simulation and test data correction: Perform structural static simulation on the simplified model through finite element analysis software to obtain the local stress σ under each working condition hand the structural deformation Δl h , reverse correct the simulation results in combination with the measured data, and calculate the estimated cost value P under each working condition h ;

[0008] S4. Establish a multi - variable surrogate model: Fit the local stress σ h , the structural deformation Δl h and the relationship with the input variable x to construct a surrogate model, and its response function is: G(x) = f T (x)β + z(x), where f T (x) is the regression polynomial basis function, β is the regression coefficient, and z(x) is the Gaussian random process;

[0009] S5. Non - linear constraint optimization solution: With the minimum cost value P h as the goal, combined with the structural strength constraint and the stiffness constraint g2(x), numerically iterate to solve the optimal solution of the key design parameters.

[0010] Preferably, in the step S1: The slider design parameters also include the slider thickness a 0hk , the cylinder shoulder length L1, the distance L from the cylinder center to the side end of the slider b , the cylinder shoulder height H1, the cylinder center distance L, the slider rib thickness a hk and the slider rib width b hk ;

[0011] The workbench design parameters also include the neutral plate thickness a 0gzt , the wall panel connection hole width W, the distance D from the wall panel connection hole to the top of the vertical plate, the vertical plate rib thickness a gzt and the vertical plate rib width b gzt ;

[0012] The wall panel design parameters also include the wall panel thickness a 0qb , the distance C from the upper surface of the cylinder to the top of the wall panel, the distance L from the cylinder center to the side end of the wall panel d , the depth A3 at the connection between the wall panel and the vertical plate, the height H at the connection between the wall panel and the vertical plate lj , the throat depth L b1 , the distance B from the lower end face of the throat to the bottom of the wall panel, the wall panel rib thickness a qb and the wall panel rib width b qb .

[0013] Preferably, in the step S3: The boundary constraints are included in the pre - processing of the structural static simulation. The boundary constraints include fixing the bottom of the workbench and the wall panel, the load application positions are the upper surface of the workbench and the lower surface of the slider, and the load value is the bending load F.

[0014] Preferably, in the step S4, the covariance function of the Kriging model is as follows:

[0015]

[0016] where is the variance of the Gaussian process; R(x i , x j , θ) is a vector with parameter θ, is the k-th element of the vector x i , is the k-th element of the vector x j , and θ k is the hyperparameter of the model.

[0017] Preferably, in the step S5, the objective function of the non-linear constraint is f(x): f(x) satisfies

[0018] where g(x) is the constraint function in the non-linear constraint equation, and g(x) mainly includes the constraint g1(x) on the value range of the key component variables and the constraint g2(x) on the structural strength and stiffness of the key components. [σ] is the allowable value of the local maximum stress, and [Δl] is the allowable value of the maximum structural deformation of the key component.

[0019] The beneficial effects of the present invention are as follows: By means of multi-factor experimental design and the combination of simulation and testing, the present invention develops a parametric optimization design method for large-tonnage bending machines based on a multi-agent model and a non-linear constraint function. It significantly improves the efficiency of the optimization design of the bending machine and shortens the structural R & D cycle by about 50% on the premise of reducing the production cost by 15%. Through this method, a variety of design schemes can be quickly generated and evaluated, optimizing performance and economic benefits and ensuring the consistency and standardization of the design process. On this basis, the parametric design method supports rapid iteration, can flexibly adapt to market changes, and promotes continuous innovation. At the same time, the research on this common technology can contribute to the design and analysis of other engineering equipment (such as stamping machines, laser cutting machines, etc.). BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the flow block diagram of the parametric optimization design method for large-tonnage bending machines based on multi-objective control of the present invention;

[0021] Figure 2 is the flow block diagram of the parametric solution design of the slider structure of the large-tonnage bending machine provided by the present invention;

[0022] Figure 3 is the flow block diagram of the parametric solution design of the workbench structure of the large-tonnage bending machine provided by the present invention;

[0023] Figure 4 This is the flow chart of the parametric solution design of the large-tonnage bending machine wallboard structure provided by the present invention. Specific embodiments

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The embodiments of the present invention are given for the purpose of illustration and description, and are not exhaustive or limit the present invention to the disclosed form. Many modifications and variations will be obvious to those of ordinary skill in the art. The embodiments are selected and described to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and design various embodiments with various modifications suitable for specific purposes.

[0025] To solve the problems raised in the background art, the present invention provides a parametric optimization design method for a large-tonnage bending machine based on multi-objective control. Figure 1 This is the flow chart of the parametric optimization design method for a large-tonnage bending machine based on multi-objective control of the present invention, which specifically includes the following steps:

[0026] S1. Determination of the key structural design parameters of the large-tonnage bending machine:

[0027] According to the working principle of the large-tonnage bending machine, the basis for structural design and selection is mainly the bending load and the bending span. The key components in the design include the slider, the workbench, and the wallboard, etc. Among them, the design parameters of the slider mainly include:

[0028] Slider structural parameters ∝(F, L0, H hk , a 0hk , L1, L b , H1, L, a yg , a hk , b hk )

[0029] Among them, F is the bending load, L0 is the bending span, H hk is the slider height (design parameter), a 0hk is the slider thickness, L1 is the length of the cylinder shoulder, L b is the distance from the cylinder center to the side end of the slider, H1 is the height of the cylinder shoulder, L is the cylinder center distance, a yg is the thickness of the reinforcing rib at the cylinder (design parameter), a hk is the thickness of the slider reinforcing rib, b hk is the width of the slider reinforcing rib.

[0030] The design parameters of the workbench mainly include:

[0031] Workbench structural parameters ∝(F, L0, H gzt , a 0gzt , W, D, L, alj , a gzt , b gzt )

[0032] Among them, F is the bending load, L0 is the bending span, H gzt is the height of the neutral plate (design parameter), a 0gzt is the thickness of the neutral plate, W is the width of the wall panel connection hole, D is the distance from the wall panel connection hole to the top of the vertical plate, L is the center distance of the oil cylinder, a lj is the thickness of the reinforcing rib at the connection between the wall panel and the workbench (design parameter), a gzt is the thickness of the vertical plate reinforcing rib, b gzt is the width of the vertical plate reinforcing rib.

[0033] The design parameters of the wall panel mainly include:

[0034] The wall panel structure parameters ∝(F, a 0qb , C, K, L d , A3, H lj , L b1 , B, a hk1 , a qb , b qb )

[0035] Among them, F is the bending load, a 0qb is the thickness of the wall panel, C is the distance from the upper surface of the oil cylinder to the top of the wall panel, K is the width of the wall panel (design parameter), L d is the distance from the center of the oil cylinder to the side end of the wall panel, A3 is the depth at the connection between the wall panel and the vertical plate, H lj is the height at the connection between the wall panel and the vertical plate, L b1 is the throat depth, B is the distance from the lower end face of the throat to the bottom end of the wall panel, a hk1 is the thickness of the throat reinforcing rib (design parameter), a qb is the thickness of the wall panel reinforcing rib, b qb is the width of the wall panel reinforcing rib.

[0036] S2. Analysis of the parameter working conditions of key components based on multi-factor experimental design:

[0037] According to the two design variable parameters of each key structural component determined in the above S1 step (the slider structure is the slider height H hk and the thickness a of the reinforcing rib at the oil cylinder; the workbench structure is the height H of the neutral plate yg and the thickness a of the reinforcing rib at the connection between the wall panel and the workbench; the wall panel structure is the width K of the wall panel and the thickness a of the throat reinforcing rib gzt ; lj ; hk1) Determine the number of sample points and value ranges of different design variables, and establish a simulation and test experiment table accordingly. A variety of different working condition test groups are obtained, and they are numbered as h, where h = 1, 2,..., Q, and Q is the total number of tests that need to be simulated and tested.

[0038] S3. Correction of the optimized values of different working condition parameters through simulation and test data:

[0039] According to the various different working condition test groups of the key structural components determined in the above S2 step, import the simplified model of the large-tonnage bending machine into ANSYS finite element software for structural static analysis. The mesh division of this simplified model is carried out by an automatic division method. Boundary constraints and the required bending load F are applied in the preprocessing. Based on the built-in part size parameterization solution function in the simulation software, the local stress σ h and the structural deformation Δl h are obtained for each group of working conditions in the test table, where h = 1, 2,..., Q, and Q is the total number of tests. On this basis, select some working conditions for test verification based on the strength and stiffness test methods, and reverse correct the finite element calculation results with the test and simulation error values to ensure the accuracy and reliability of the simulation data modeling.

[0040] At the same time, according to the value results of each variable of the key structural components in different working condition test groups, based on aspects such as the actual procurement of plates, processing and assembly costs, and waste nesting management, calculate the cost pre-estimation value P h under different working conditions, where h = 1, 2,..., Q, and Q is the total number of tests.

[0041] S4. Establishment of the objective function of key components based on the multi-agent model:

[0042] According to the parameter situations of different working condition combinations determined in the S2 step, combined with the local stress values and structural deformation values corresponding to each group of working conditions obtained by simulation in the S3 step, in the numerical analysis software, for the local stress values σ h and the structural deformation values Δl h corresponding to different working conditions, perform the fitting of the Kriging agent model. Divide each dimension into equal-probability intervals by Latin hypercube sampling and randomly extract sample points to improve the spatial coverage. On this basis, evaluate the model accuracy through the mean square error index. If the accuracy is insufficient, gradually increase the sample size, and update the relevant parameters through the maximum likelihood estimation method to improve the accuracy of the agent model until the model performance meets the requirements.

[0043] Improving the accuracy of the Kriging model based on the maximum likelihood estimation method mainly optimizes the length scale parameter and the noise variance to maximize the joint probability density of the sample data, thereby better capturing the data characteristics and reducing the model bias. The maximum likelihood estimation method can effectively handle noisy data, distinguish signals from noise, and enhance the robustness of the model. Combining optimization methods such as gradient descent or genetic algorithms can efficiently solve the globally optimal hyperparameter combination. By optimizing the hyperparameters and enhancing the model fitting ability, the accuracy and generalization performance of the Kriging model are significantly improved.

[0044] According to the Kriging algorithm theory, assume that the relationship between the output response G(x) and the input variable x is expressed as:

[0045] G(x) = f T (x)β + z(x)

[0046] f T (x) is the basis function vector of the regression polynomial; β is the vector of regression coefficients; z(x) is a Gaussian random process with zero mean, and the covariance function is expressed as:

[0047]

[0048] Among them, is the variance of the Gaussian process; R(x i , x j , θ) is a vector with parameter θ, and its correlation function can be expressed as:

[0049]

[0050] Among them, is the k-th element of the vector x i , is the k-th element of the vector x j , θ k is the model hyperparameter.

[0051] Given a set of training samples with a capacity of N, the unbiased estimate and prediction error of G(x) are defined as follows:

[0052]

[0053] Among them, is the estimated value of β; r(x) is the correlation function vector between the training sample points and the prediction points; Y is the training sample response.

[0054] S5. Parametric optimization design method for a bending machine based on non-linear constraint control:

[0055] According to the local stress values σ corresponding to different working conditions obtained by fitting with the kriging algorithm in step S4 above h and the structural deformation value Δl h , combined with the estimated cost value P h A nonlinear constraint control equation under multiple variables is established. On the premise of ensuring the lowest estimated cost value P h , the optimal solution satisfying the local stress value and deformation value of the structure is iteratively solved based on the numerical simulation calculation platform. The main variables to be inversely solved include: the slider height H hk , the thickness a of the reinforcing rib at the oil cylinder yg , the height H of the neutral plate gzt , the thickness a of the reinforcing rib at the connection between the wall panel and the workbench lj , the width K of the wall panel and the thickness a of the throat reinforcing rib hk1 .

[0056]

[0057] Among them, f(x) is the objective function in the nonlinear constraint equation, with the minimum cost value P h as the goal; g(x) is the constraint function in the nonlinear constraint equation, which mainly includes the constraint g1(x) on the value range of key component variables and the constraint g2(x) on the structural strength and stiffness of key components. [σ] is the allowable value of the local maximum stress, and [Δl] is the allowable value of the maximum deformation of the key component structure.

[0058] As Figure 2 shown in the flowchart of the parametric solution design of the slider structure of the large-tonnage bending machine, according to the known parameters and performance index requirements, the slider height H hk is preferentially solved based on the method proposed in the present invention, and the thickness a of the reinforcing rib at the oil cylinder is adjusted yg to meet the requirements of the local strength at the oil cylinder installation of the slider.

[0059] As Figure 3 shown in the flowchart of the parametric solution design of the workbench structure of the large-tonnage bending machine, according to the known parameters and performance index requirements, the height H of the neutral plate gzt is preferentially solved based on the method proposed in the present invention, and the thickness a of the reinforcing rib at the connection between the wall panel and the workbench is optimized lj to meet the strength index requirements at this connection.

[0060] As Figure 4 shown in the flowchart of the parametric solution design of the wall panel structure of the large-tonnage bending machine, according to the known parameters and performance index requirements, the width K of the wall panel is preferentially solved based on the method proposed in the present invention, and the thickness a of the throat reinforcing rib is adjusted hk1 to meet the strength index requirements at the throat position at the lower end of the wall panel.

[0061] In summary, the method of the present invention realizes the development of a large-tonnage bending machine through steps of determining key structural design parameters of the large-tonnage bending machine, multi-factor experimental design and working condition analysis, simulation and test data correction, establishing a multi-variable surrogate model, and non-linear constraint optimization solution. The design method based on the multi-variable surrogate model and the non-linear constraint function significantly improves the optimization design efficiency of the bending machine, shortens the structural research and development cycle, and realizes the design of the bending machine with high efficiency and high accuracy.

Claims

1. A parametric optimization design method for large-tonnage bending machines based on multi-objective control, characterized in that, Specifically, it includes the following steps: S1. Determine the key structural design parameters of the large-tonnage bending machine: Based on the bending load F and the bending span L0, determine the key design parameters of the slider, workbench, and wallboard, including the slider height H hk , the thickness a of the reinforcing rib at the oil cylinder yg , the height H of the neutral plate gzt , the thickness a of the reinforcing rib at the connection between the wallboard and the workbench lj , the width K of the wallboard and the thickness a of the throat reinforcing rib hk1 ; S2. Multi-factor experimental design and working condition analysis: Select sample points and set value ranges for the design variables in step S1, generate simulation and test experimental groups, and number them as h, where h = 1, 2,..., Q, and Q is the total number of experiments that need to be simulated and tested; S3. Simulation and test data correction: Perform structural static simulation on the simplified model through finite element analysis software to obtain the local stress σ under each working condition h and the structural deformation Δl h , reverse-correct the simulation results in combination with the measured data, and calculate the cost prediction value P under each working condition h ; S4. Establish a multi - agent model: Based on the Kriging algorithm, fit the local stress σ h and the structural deformation Δl h to construct an agent model for the relationship with the input variable x. Its response function is: G(x) = f T (x)β + z(x), where f T (x) is the regression polynomial basis function, β is the regression coefficient, and z(x) is the Gaussian random process; S5. Nonlinear constraint optimization solution: taking the cost value P h to be the minimum, combined with the structural strength constraint and the stiffness constraint g2(x), the optimal solution of the key design parameters is obtained through numerical iteration.

2. The method according to claim 1, wherein In the step S1, the slider design parameters further include the slider thickness a 0hk , the cylinder shoulder length L1, the distance L from the cylinder center to the side end of the slider b , the cylinder shoulder height H1, the cylinder center distance L, the slider reinforcing rib thickness a hk and the slider reinforcing rib width b hk ; the workbench design parameters further include the neutral plate thickness a 0gzt , the wall panel connection hole width W, the distance D from the wall panel connection hole to the top of the vertical plate, the vertical plate reinforcing rib thickness a gzt and the vertical plate reinforcing rib width b gzt ; The design parameters of the wall panel also include the wall panel thickness a 0qb , the distance C from the upper surface of the oil cylinder to the top end of the wall panel, and the distance L from the center of the oil cylinder to the side end of the wall panel d , the depth A3 at the connection between the wall panel and the vertical plate, and the height H at the connection between the wall panel and the vertical plate lj , the throat depth L b1 , the distance B from the lower end face of the throat to the bottom end of the wall panel, the thickness a of the wall panel stiffener qb and the width b of the wall panel stiffener qb .

3. The method according to claim 1, wherein In the said step S3: Boundary constraints are included in the preprocessing of structural static simulation. The boundary constraints include fixing the bottom of the workbench and the wall panel, the load application positions are the upper surface of the workbench and the lower surface of the slider, and the load value is the bending load F.

4. The method according to claim 1, wherein In the step S4, the covariance function of the Kriging model is as follows: Among them, is the variance of the Gaussian process; R(x i , x j , θ) is a vector with parameter θ, is the k-th element of the vector x i , is the k-th element of the vector x j , and θ k are the model hyperparameters.

5. The method according to claim 1, wherein In the said step S5: The objective function of the non-linear constraint is f(x): f(x) satisfies where g(x) is the constraint function in the non-linear constraint equation. g(x) mainly includes the constraint on the value range of key component variables g1(x) and the constraint on the structural strength and stiffness of key components g2(x), [σ] is the allowable value of the local maximum stress, and [Δl] is the allowable value of the maximum deformation of the key component structure.

Images