A variable number of variable double-layer iteration optimization scheme determination method and system

Abstract

The application belongs to the technical field of engineering design optimization, and discloses a double-layer iterative optimization scheme determination method and system with variable number of variables, which comprises the following steps: determining an optimization object, setting an outer-layer iteration variable as the number of variables of different optimization objects, and setting an inner-layer iteration variable as the variable itself of different optimization objects; performing outer-layer iteration: determining a variable number scheme by using an orthogonal table, performing iteration one by one in the inner layer according to the variable number scheme, obtaining an optimal comprehensive score value and variable value under the current variable, and returning all inner-layer iteration results to the outer layer; the outer layer compares the comprehensive score values to obtain an optimal variable number and variable value of this iteration of the outer layer; and repeating the inner-layer and outer-layer iteration until a convergence condition is met to output an optimal scheme. The orthogonal table in the Taguchi method is used to reasonably reduce the number of tests and reduce the cost consumption caused by multiple iterations, so that the application is a feasible and practical optimization method for improving engineering efficiency.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of engineering design optimization, and particularly relates to a double-layer iterative optimization scheme determination method and system with variable variable number. BACKGROUND

[0002] At present, optimization algorithms in which the optimization result is determined by the number of design variables and the variables themselves are widely used in engineering design, such as composite material layer design, in which the number of layers and the layer angle determine the performance of the material; satellite orbit arrangement in space, in which the number of orbits and the running angle of the satellite determine the arrangement of wind power generators in a wind power plant, in which the number of arrangements and the type of the generator determine the power generation capacity, etc. In these problems, the variable number is an important index affecting the optimization result in the optimization design.

[0003] Nowadays, there are many existing optimization methods, such as gradient descent method, damped Newton method, genetic algorithm and ant colony algorithm, which are used to perform multivariable combination optimization for different optimization problems to obtain an optimal scheme. However, the above optimization schemes can only be applied to optimization scheme determination with a fixed variable number, and cannot obtain an optimal result in complex optimization scheme determination with a variable number. Meanwhile, the existing technology is complicated, has a large amount of calculation and is low in efficiency.

[0004] Through the above analysis, the problems and defects of the existing technology are that the existing engineering design scheme optimization scheme cannot obtain an optimal result in complex optimization scheme determination with a variable number, and the existing technology is complicated, has a large amount of calculation and is low in efficiency. SUMMARY

[0005] In view of the problems in the prior art, the present application provides a double-layer iterative optimization scheme determination method and system with a variable variable number.

[0006] The present application is implemented as follows. A double-layer iterative optimization scheme determination method with a variable variable number comprises the following steps.

[0007] Firstly, an optimization object is determined, the outer-layer iteration variable is set as the variable number of different optimization objects, and the inner-layer iteration variable is set as the variable itself of the different optimization objects.

[0008] Secondly, outer-layer iteration is performed, the variable number scheme is determined by using an orthogonal table, the inner layer is iterated one by one according to the variable number scheme, the optimal comprehensive score value and the variable value under the current variable are obtained, and all the inner-layer iteration results are returned to the outer layer. The outer layer compares the comprehensive score values to obtain the optimal variable number and variable value of this iteration of the outer layer. The inner and outer layer iteration is repeated until the optimal scheme is output when the convergence condition is met.

[0009] Further, the variable number variable double-layer iteration optimization scheme determination method comprises the following steps:

[0010] Step one, determining the optimization target and constraint condition and constructing the optimization model; determining the optimization object and the variable of each optimization object and the number of each variable and the design space;

[0011] Step two, determining the variable of the outer layer iteration, adopting the three-level orthogonal test method to establish the level table and the orthogonal table of the variable of the outer layer iteration, and determining the variable number scheme of each optimization object;

[0012] Step three, determining the variable of the inner layer iteration, adopting the three-level orthogonal test design method to establish the level table and the orthogonal table of the variable of the inner layer iteration based on the variable number scheme of the outer layer iteration which has been determined;

[0013] Step four, performing test calculation on each sample point of the inner layer iteration under each scheme to obtain the test value of each response, adopting the penalty function method to add each constraint to the objective function to calculate the modified response value of the objective function; and calculating the comprehensive score value of each sample point to obtain the optimal result of the inner layer iteration;

[0014] Step five, judging whether the optimization target meets the optimal result condition of the inner layer iteration, if yes, returning the optimal result of the inner layer iteration and the comprehensive score value of the optimal result to the orthogonal table of the variable of the outer layer iteration, if not, moving the design space of each variable and returning to step three;

[0015] Step six, obtaining all the response values returned by the inner layer iteration for the orthogonal table of the outer layer iteration; judging whether the optimization target meets the optimal result condition of the outer layer iteration, if yes, obtaining the final result of the optimization target, if not, moving the design space of the variable number and returning to step two.

[0016] Further, the optimization model comprises:

[0017] The multi-objective multi-constraint optimization model is as follows:

[0018]

[0019] Wherein, n represents the number of optimization objects; X represents the variable vector, x1, x2, …, x n Respectively represent the 1st, 2nd, …, nth optimization object; N1, N2, …, N n Respectively represent the 1st, 2nd, …, nth optimization object variable number; the alternative value of the i-th variable of the 1st optimization object is x 1i ∈{α1,α2,…,α m}, i = 1, 2, ..., N1, m represents the number of candidate values ​​for the variable itself; the candidate value of the j-th variable of the second optimization object is x. 2j ∈{β1,β2,…,β p}, j = 1, 2, ..., N2, p represents the number of candidate values ​​for the variable itself; the candidate value of the k-th variable of the n-th optimization object is x. nk ∈{γ1,γ2,…,γ k}, k = 1, 2, ..., N n k represents the number of alternative values ​​for the variable itself; F1(X) and F2(X) are objective functions, G1(X) and G2(X) are constraint functions, [G1(X)] and [G2(X)] represent the boundary values ​​of the constraint functions, and the number of objective functions and constraint functions can be determined according to the specific problem;

[0020] Furthermore, the outer iteration variable is the number of variables for each optimization object; the inner iteration variable is the variables of different optimization objects themselves.

[0021] Furthermore, optimize objects x1, x2, ..., x n The number of variables are N1, N2, ..., N n The number of variables further determines the variables themselves for each optimization object, and then specific values ​​are selected from the candidate set of variables themselves; the number of variables for each optimization object and the variables themselves constitute the design space;

[0022] Furthermore, step four includes:

[0023] 1) Calculate the response value F(X) of the objective function within the inner iteration step;

[0024] 2) Using the penalty function method, each constraint G(X) is added to the objective function to obtain the corrected response value F'(X) of the objective function:

[0025]

[0026] Where s represents the penalty factor, G τ (X) represents the function value of the τth constraint;

[0027] 3) Obtain the comprehensive score Y for each scheme using the weighted scoring method:

[0028]

[0029] Where u represents the test number, v represents the index sequence number, w represents the number of indicators, and Y represents the number of indicators. u b represents the overall score of the u-th trial. v F represents the weight of the v-th indicator. 0v F represents the corrected response value of the v-th index in the initial design.uv This represents the corrected response value of the v-th index in the u-th experiment.

[0030] Furthermore, in step six, the orthogonal array of the outer iteration obtains all response values ​​returned by the inner iteration; it is determined whether the optimization objective satisfies the optimal result condition of the outer iteration. If it does, the final result of the optimization objective is obtained; if not, the design space of the number of variables is shifted and the process returns to step two, including:

[0031] The orthogonal array of the outer iteration obtains the optimal result returned by the inner iteration and the comprehensive score value of the optimal result. The outer iteration compares the comprehensive score values ​​of each scheme, selects the scheme with the maximum comprehensive score value, and obtains the optimal level of this outer iteration.

[0032] Determine whether the maximum comprehensive score obtained from the screening meets the convergence condition: if it does, then take the maximum comprehensive score as the maximum score of the entire outer iteration, and take the optimal level corresponding to the maximum comprehensive score as the final optimization result; if it does not meet the condition, then take the optimal result of the orthogonal design of the outer iteration as the initial value of the next orthogonal design, and return, until the convergence condition is met, and obtain the number of variables that meet the condition.

[0033] The final optimization result includes the number of variables that meet the conditions and the values ​​of the design variables themselves.

[0034] Another object of the present invention is to provide a system for determining a two-level iterative optimization scheme with a variable number of variables, which implements the method for determining a two-level iterative optimization scheme with a variable number of variables. The system for determining a two-level iterative optimization scheme with a variable number of variables includes:

[0035] The parameter setting module is used to determine the optimization objective and constraints and build the optimization model; determine the optimization object, the variables of each optimization object, the number of each variable, and the design space;

[0036] The initialization module is used to determine the variables of the outer iteration. It uses a three-level orthogonal experimental design method to establish the level table and orthogonal table of the outer iteration variables and determine the number of variables for each optimization object. The module is also used to determine the variables of the inner iteration. Based on the determined number of variables of the outer iteration, it uses a three-level orthogonal experimental design method to establish the level table and orthogonal table of the inner iteration variables.

[0037] The iterative experiment module is used to perform inner and outer layer iterations based on the level table and orthogonal table of the outer layer iteration variables and the level table and orthogonal table of the inner layer iteration variables.

[0038] The results output module is used to output the optimal results obtained from the inner and outer layer iterations as the optimization scheme.

[0039] Another object of the present invention is to provide a computer device, the computer device including a memory and a processor, the memory storing a computer program, the computer program being executed by the processor causing the processor to perform the steps of the method for determining a two-level iterative optimization scheme with a variable number of variables.

[0040] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method for determining a two-level iterative optimization scheme with a variable number of variables.

[0041] Another objective of this invention is to provide an information data processing terminal for implementing the two-layer iterative optimization scheme determination system with a variable number of variables.

[0042] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0043] This invention defines the number of variables for each optimization object that significantly affects the target as external variables. Based on the Taguchi method of orthogonal experiments, it establishes level tables and orthogonal tables for the external variables to determine the number of variables for each optimization object. This invention also defines the variables for each optimization object themselves as internal variables. Based on the Taguchi method of orthogonal experiments, it establishes level tables and orthogonal tables for the internal variables, calculates the comprehensive score value for each scheme regarding the internal variables, and obtains the optimal combination of levels for the internal variables.

[0044] This invention utilizes the optimal combination of levels in the internal table and the corresponding comprehensive score to obtain the optimal level of the external variable. This invention highlights a solution for complex problems such as "variable number of variables," and can systematically solve complex optimization problems with a variable number of design variables. That is, the optimization result is jointly determined by the number of variables and the variables themselves, and the obtained result satisfies multi-variable combination and multi-criteria decision-making.

[0045] This invention proposes a two-level iterative method to solve optimization problems with a variable number of variables. Furthermore, by utilizing orthogonal array design from the Taguchi method, this invention effectively reduces the number of trials, minimizing the cost associated with multiple iterations. This is a practical optimization method for improving engineering efficiency.

[0046] This invention considers two optimization metrics: the number of variables and the variables themselves, and can be well applied to optimization problems in related engineering problems.

[0047] Does the technical solution of this invention solve a long-standing but unsolved technical problem? In engineering design, there is a widespread need for optimization algorithms where the number of design variables and the variables themselves jointly determine the optimization result. Examples include composite material layup design: the number of layups and the layup angle determine the material's performance; satellite orbit arrangement in space: the number of satellite orbits and the satellite's orbital angle; and wind turbine arrangement in wind farms: the number of turbines and the type of turbine determine the power output. These problems use the number of variables as a crucial indicator in the optimization design, influencing the design result. There is an urgent need for an efficient optimization algorithm that can directly handle optimization problems with a variable number of design variables. While many existing optimization algorithms exist, such as gradient descent, genetic algorithms, and ant colony optimization, they are commonly used to solve optimization problems with a fixed number of design variables. These algorithms are limited in their application to complex optimization problems with a variable number of design variables. To address this problem, this invention proposes a two-layer iterative method to solve optimization problems with a variable number of variables. Meanwhile, this paper utilizes orthogonal array design in the Taguchi method to reasonably reduce the number of experiments and decrease the cost of multiple iterations, which is a feasible optimization method to improve engineering efficiency. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of a method for determining a two-level iterative optimization scheme with a variable number of variables, provided in an embodiment of the present invention.

[0049] Figure 2 This is a schematic diagram of the method for determining a two-layer iterative optimization scheme with a variable number of variables provided in an embodiment of the present invention.

[0050] Figure 3 This is a flowchart of a method for determining a two-level iterative optimization scheme with a variable number of variables, provided in an embodiment of the present invention.

[0051] Figure 4 This is a schematic diagram of the car door structure provided in an embodiment of the present invention;

[0052] Figure 5 This is a schematic diagram of composite material layup provided in an embodiment of the present invention.

[0053] Figure 6 The maximum comprehensive score value Z' of the external variable number of layup layers provided in the embodiments of the present invention. max A line graph showing the number of iterations.

[0054] Figure 7 This is a comparison chart of the performance metrics of the patented algorithm provided in this invention and the traditional genetic algorithm. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0056] To enable those skilled in the art to fully understand how the present invention is specifically implemented, this section provides an explanatory description of the embodiments that expand upon the technical solutions of the claims.

[0057] like Figures 1 to 3 As shown, the method for determining a two-level iterative optimization scheme with a variable number of variables provided in this embodiment of the invention includes the following steps:

[0058] S101, Determine the optimization objective and constraints and construct the optimization model; determine the optimization object, the variables of each optimization object, the number of each variable, and the design space;

[0059] S102, determine the variables of the outer iteration, and use the three-level orthogonal experiment method to establish the level table and orthogonal table of the outer iteration variables, and determine the number of variables for each optimization object;

[0060] S103, determine the variables of the inner iteration, and establish the level table and orthogonal table of the inner iteration variables based on the determined number of variables of the outer iteration using the three-level orthogonal experimental design method;

[0061] S104. Perform experimental calculations on each sample point of the inner iteration under each scheme to obtain the experimental value and comprehensive score value of each response. Use the penalty function method to attach each constraint to the objective function and calculate the corrected response value of the objective function. Calculate the comprehensive score value of each sample point to obtain the optimal result of this inner iteration.

[0062] S105, determine whether the optimization objective satisfies the condition for the optimal result of the inner iteration. If it does, return the optimal result obtained from the inner iteration and the comprehensive score value of the optimal result to the orthogonal table of the outer iteration variables. If it does not satisfy the condition, move the design space of each variable and return to step S103.

[0063] S106, the orthogonal array of the outer iteration obtains all response values ​​returned by the inner iteration; determine whether the optimization objective satisfies the optimal result condition of the outer iteration. If it does, obtain the final result of the optimization objective. If it does not, move the design space of the number of variables and return to step S102.

[0064] Taking a two-objective, two-constraint optimization problem with n optimization objects as an example, the optimization mathematical model provided in this embodiment of the invention is as follows:

[0065]

[0066] Where n represents the number of optimization objects; X represents the variable vector, x1, x2, ..., x n These represent the 1st, 2nd, ..., nth optimization objects, respectively. N1,N2,…,N n Let x represent the number of variables in the 1st, 2nd, ..., nth optimization objects, respectively; x is the candidate value of the i-th variable in the 1st optimization object. 1i ∈{α1,α2,…,α m}, i = 1, 2, ..., N1, m represents the number of candidate values ​​for the variable itself; the candidate value of the j-th variable of the second optimization object is x. 2j ∈{β1,β2,…,β p}, j = 1, 2, ..., N2, p represents the number of candidate values ​​for the variable itself; the candidate value of the k-th variable of the n-th optimization object is x. nk ∈{γ1,γ2,…,γ k}, k = 1, 2, ..., N n k represents the number of alternative values ​​for the variable itself; F1(X) and F2(X) are objective functions, G1(X) and G2(X) are constraint functions, and [G1(X)] and [G2(X)] represent the boundary values ​​of the constraint functions. Note that the number of objective functions and constraint functions can be determined according to the specific problem.

[0067] The method for determining a two-level iterative optimization scheme with a variable number of variables provided in this embodiment of the invention specifically includes the following steps:

[0068] (1) Determine the optimization objective and constraints. In optimization model (II), the optimization objective is to find the minimum values ​​of F1(X) and F2(X), and the functions G1(X) and G2(X) are constraint functions about X.

[0069] (2) Select experimental objects that have a significant impact on the objective as the optimization objects of the optimization problem, and determine the variables of each optimization object, the number of each variable, and the design space. For example, in model (II), the optimization objects are x1, x2, ..., x n The number of elements are N1, N2, ..., N n That is, variables x1, x2, ..., x n The design spaces are N1, N2, ..., N n dimension.

[0070] (3) The number of variables for each optimization object is determined as the variables of the outer iteration. The three-level orthogonal experimental design method is adopted to establish the level table and orthogonal table of the outer iteration variables and determine the number of variables for each optimization object.

[0071] In model (II), three optimization objects X = [x1, x2, x3] are considered. T For example, the number of variables for optimization objects x1, x2, and x3 are N1, N2, and N3, respectively. When using a three-level orthogonal experimental design, level 2 is the initial design, and levels 1 and 3 are adjacent alternative values ​​of the initial design. Assume that the initial design level 2 values ​​for all three optimization objects are 4, and the level 1 and level 3 values ​​for all three optimization objects are 3 and 5, respectively. The number of variables N1, N2, and N3 for optimization objects x1, x2, and x3 are determined as the variables for the outer iteration. The table of the number of variables in the first outer iteration is shown in Table 1. Nine sets of data are generated using the Taguchi orthogonal method. The orthogonal table of the number of variables in the first outer iteration is shown in Table 2. For example, in experiment number 1 in the table, the number of variables N1, N2, and N3 for optimization objects x1, x2, and x3 are all at level 1, i.e., 3.

[0072] Table 1. Level table of the number of variables in the first outer iteration.

[0073]

[0074] Table 2. Horizontal orthogonal table of the number of variables in the first outer iteration.

[0075]

[0076]

[0077] Table 2 shows the experimental schemes for each experiment number, corresponding to different numbers of variables for different optimization objects.

[0078] (4) The variables of each optimization object are determined as the variables of the inner iteration. For the number of variables determined in the outer iteration, a three-level orthogonal experimental design method is adopted to establish the level table and orthogonal table of the inner iteration variables.

[0079] In model (II), three optimization objects X = [x1, x2, x3] are considered. T For example, according to step (3), take the first scheme of the outer iteration, that is, the number of variables of the three optimization objects x1, x2, x3 N1=N2=N3=3: x1=[x 11 ,x 12 ,x 13 ] T , x2=[x 21 ,x 22 ,x 23 ] T x3=[x 31 ,x 32 ,x 33 ] T Then the variable of the optimization object itself is x.11 ,x 12 ,x 13 ,x 21 ,x 22 ,x 23 ,x 31 ,x 32 ,x 33 When using a three-level orthogonal experimental design, level 2 is the initial design, and levels 1 and 3 are adjacent alternative values ​​of the initial design.

[0080] Assume x 1i x 2j x 3k The selected values ​​(i = j = k = 1, 2, 3) are {α1, α2, α3, α4, α5}, {β1, β2, β3, β4, β5}, and {γ1, γ2, γ3, γ4, γ5}, respectively. The initial design level 2 values ​​of the variables of the optimization object are α3, β3, and γ3, respectively. Then, the adjacent candidate level 1 and level 3 are α2, β2, and γ2, and α4, β4, and γ4, respectively. The level table of the inner layer variables in the first iteration under the first outer iteration scheme is shown in Table 3.

[0081] Generate information about x using Taguchi orthogonality. 1i x 2j x 3k The 27 sets of data (i = j = k = 1, 2, 3) were used as representative level combinations for the experiment, and the iterative orthogonal array is shown in Table 4.

[0082] Table 3. Level table of variables in the first iteration under the first outer iteration scheme (N1=N2=N3=3).

[0083]

[0084] Table 3.1 Level table of variables in the second inner iteration under the first outer iteration scheme (N1=N2=N3=3)

[0085]

[0086]

[0087] Table 4. Horizontal orthogonal table of variables for the first inner-layer iteration under the first outer-layer iteration scheme (N1=N2=N3=3)

[0088]

[0089]

[0090] (5) Calculate the comprehensive score value Y within the inner iteration step under this scheme. The specific calculation process is as follows:

[0091] 1) Calculate the response value F(X) of the objective function within the iteration step.

[0092] 2) Using the penalty function method, each constraint G(X) is added to the objective function to obtain the corrected response value F'(X) of the objective function. The calculation method is as follows:

[0093]

[0094] In the formula, s is the penalty factor, and G τ (X) is the function value of the τth constraint.

[0095] 3) Obtain the comprehensive score Y for each scheme using a weighted scoring method. To minimize the corrected response value, the comprehensive score is calculated as follows:

[0096]

[0097] Where u represents the test number, v represents the index sequence number, w represents the number of indicators, and Y represents the number of indicators. u b represents the overall score of the u-th trial. v F represents the weight of the v-th indicator. 0v F represents the corrected response value of the v-th index in the initial design. uv This represents the corrected response value of the v-th index in the u-th experiment.

[0098] In Model (II), taking three optimization objects as an example, assuming that the number of variables for each of the three optimization objects is 3, according to the three-level orthogonal experimental design rules, 27 experimental numbers for the variables of the optimization objects themselves are generated, and the comprehensive score values ​​of each scheme are shown in the last column of Table 4.

[0099] (6) Based on the representative level combinations and their comprehensive scores within the iteration step, the analysis of means (ANOM) method is used to calculate the comprehensive score Y corresponding to the level combinations of variables after the full experiment of all inner-layer iterations, and the maximum comprehensive score Y' within the iteration step is selected. max .

[0100] (7) Determine the maximum comprehensive score value Y' within the outer iteration step. max Does it satisfy the condition that there is no increase after 5 consecutive iterations? If so, then change Y' max Y is the maximum comprehensive score value of the inner iteration under this scheme. max The corresponding level combination is the optimal level combination of the inner iterative variables under the scheme, and proceed to step (8); if it is not satisfied, the optimal result of the nth orthogonal design of the variable is used as the initial value of the (n+1)th orthogonal design, and return to step (4).

[0101] Assume the first iteration x 11 =x 12 =x 13 =α2,x 21 =x 22 =x 23 =β1,x 31 =x 32 =x 33 Taking γ3 as the optimal result as an example, we treat it as level 2. We then shift the adjacent candidate values ​​(levels 1 and 3) accordingly. The resulting table of inner-layer second-stage iteration variables under the first outer-layer iteration scheme is shown in Table 3.1. Next, based on the Taguchi orthogonal design method, we obtain an experimental scheme table similar to Table 4. Repeating the above calculation steps achieves iterative optimization.

[0102] The movement of the level table in subsequent iterations follows the same pattern. Continue iterating through the inner table until the maximum overall score Y is obtained. max .

[0103] (8) Calculate the optimal level combination of the inner iteration under each scheme in the outer iteration step and the corresponding comprehensive score value Y. max Return to Table 2, compare the comprehensive score Z of each scheme, and select the scheme with the highest comprehensive score Z'. max This allows us to obtain the optimal level for the outer iteration in that step.

[0104] (9) Determine the maximum comprehensive score Z' within the outer iteration step. max Does the result satisfy the condition that five consecutive optimal design results do not increase? If so, then Z' max Z is the maximum score value of the entire outer iteration. max And take the corresponding optimal level as the final design result, that is, get the number of variables that meet the conditions, i.e. the variable values; if not, take the optimal result of the nth orthogonal design of the outer layer as the initial value of the (n+1)th orthogonal design, and return to step (3) until the outer surface design target meets the condition that the five consecutive optimal design results have not increased, and get the number of variables that meet the conditions.

[0105] To demonstrate the inventiveness and technical value of the technical solution of this invention, this section provides specific product or related technology application examples of the technical solution claimed.

[0106] The technical solution of the present invention will be further described below with reference to specific embodiments.

[0107] Example 1:

[0108] A car's weight directly affects its fuel consumption. Achieving lightweight construction while maintaining overall vehicle performance is a pressing issue. Car doors are a crucial component of a car. They primarily consist of an inner panel, outer panel, anti-collision beam, mounting brackets, window frame reinforcement plates, door lock reinforcement plates, rubber strip guides, and glass guides. A schematic diagram of the car door structure is shown below. Figure 4 As shown. Among them, the inner panel and the anti-collision bar are important components that determine the performance of the car door.

[0109] The method for determining a two-layer iterative optimization scheme with a variable number of variables provided in this embodiment of the invention is applied to the process of determining a lightweight optimization scheme for automobiles. The composite material layup of the inner panel of the car door and the anti-collision bar is used as the optimization variable. Taking the car door as an example, the weight of the car door structure is reduced as much as possible while meeting the requirements of stiffness and crashworthiness.

[0110] The method for optimizing the layup design of composite materials for car door inner panels with a variable number of variables provided in this invention includes:

[0111] (1) Considering the requirements of stiffness, modality and lightweighting of the car door, the optimization objective F(X) is determined to be the minimum value of the mass M and the maximum value of the stiffness R of the car door; the upper torsional stiffness G1(X), lower torsional stiffness G2(X) and first natural frequency f(X) of the car door are used as constraints.

[0112] (2) The inner door panel is one of the important components of a car door. The inner door panel, which significantly affects the quality and stiffness of the car door, is selected as the optimization object to determine the design space for the number of composite material layups and the layup angles of the inner door panel. The mathematical model for door optimization using the inner door panel as the research object is as follows:

[0113]

[0114] In the formula, x1 is the optimization object, i.e., the inner panel of the car door; the number of plies of the optimization object x1 is the number of variables, and its design space is N1-dimensional; the variable of the optimization object x1 itself is the ply angle x of the i-th layer of the composite material. 1i The optimization objective F(X) is to minimize the mass M(X) and maximize the stiffness R(X) of the door; the constraints are the upper torsional stiffness G1(X), lower torsional stiffness G2(X), and first natural frequency f(X) of the door.

[0115] (3) The number of variables x1, i.e., the number of layers N1, is determined as the variable for the outer layer iteration. A three-level orthogonal experimental design method is adopted, with level 2 as the initial design and levels 1 and 3 as adjacent alternative values ​​of the initial design. It is assumed that the initial design value of N1 at level 2 is 3 layers, and the values ​​of level 1 and level 3 are 2 layers and 4 layers, respectively. The table of the number of variables of the inner door panel in the first outer layer iteration is shown in Table 5; through Taguchi orthogonality, the scheme for determining the outer variables of the inner door panel in the first iteration is shown in Table 6, and the determined number of variables x1 is obtained.

[0116] Table 5. Level table of external variables of the inner door panel in the first iteration.

[0117]

[0118] Table 6. Horizontal orthogonal table of external variables of the car door inner panel in the first iteration.

[0119]

[0120] (4) The layup angle x of the i-th layer of the composite material for the inner door panel 1i The inner-layer iteration variables are determined. For different schemes with a determined number of variables N1, a three-level orthogonal experimental design method is used to establish the level table and orthogonal table for the inner-layer iteration. Specifically, when the number of variables of the optimization object x1 is N1 = 2, the variable of the optimization object itself is x. 11 ,x 12 When using a three-level orthogonal experimental design, level 2 is the initial design, and levels 1 and 3 are adjacent alternative values ​​of the initial design. Assume x... 1i The initial design had level 2 values ​​of 20°, and level 1 and level 3 values ​​of 15° and 25° respectively. The level table of the inner layer iteration variables for the car door inner panel is shown in Table 7. Through Taguchi orthogonalization, a value related to x is generated. 1i Nine sets of data (i = 1, 2) were used as representative level combinations for the experiment. The orthogonal table of the variables of the inner door panel itself is shown in Table 8.

[0121] Table 7. Level table of inner panel surface variables for the first iteration (N1=2)

[0122]

[0123]

[0124] Table 8. Horizontal orthogonal array of variables for the inner panel of the car door during the first iteration (N1=2)

[0125]

[0126] (5) Based on the three-level orthogonal experimental design rules, nine experimental numbers were generated for the variables related to the optimization object itself. The comprehensive score value Y for each scheme regarding the variable of the inner surface of the car door inner panel was calculated, as shown in the last column of Table 8. The specific calculation process is as follows:

[0127] 1) Calculate the response values ​​M(X) and R(X) of the objective function for each scheme;

[0128] 2) Using the penalty function method, each constraint G(X) is added to the objective function to obtain the corrected response value F'(X) of the objective function. The calculation method is as follows:

[0129]

[0130] In the formula, s is the penalty factor.

[0131] 3) Obtain the comprehensive score Y for each scheme using a weighted scoring method. To minimize the corrected response value, the comprehensive score is calculated as follows:

[0132]

[0133] Where u represents the test number, v represents the index sequence number, w represents the number of indicators, and Y represents the number of indicators. u b represents the overall score of the u-th trial. v F represents the weight of the v-th indicator. 0v F represents the corrected response value of the v-th index in the initial design. uv This represents the corrected response value of the v-th index in the u-th experiment.

[0134] (6) Based on the representative level combinations and their comprehensive scores within the iteration step, the analysis of means (ANOM) method is used to calculate the comprehensive score Y corresponding to the level combinations after the full experiment of all internal table variables, and the maximum comprehensive score within the iteration step is selected.

[0135] Let Y(x) 111 Y(x) represents the overall score corresponding to level 1 of the first variable of the inner door panel. 112 Y(x) represents the overall score corresponding to level 2 of the first variable of the inner door panel. 113 Y(x) represents the overall score corresponding to level 3 of the first variable of the door inner panel; 121 Let be the overall score value corresponding to level 1 of the second variable of the door inner panel; ...; and so on. According to the ANOM method, then:

[0136]

[0137] …

[0138] The maximum comprehensive score within the iteration step is:

[0139] Y' max =max[Y(x 111 ), Y(x 112 ), Y(x 113 ), Y(x 121 ), Y(x 122 ), Y(x 123 )]

[0140] (7) Determine the maximum comprehensive score value Y' of the variables in the iterative step table. max Does it satisfy the condition that there is no increase after 5 consecutive iterations? If so, then change Y' max The maximum comprehensive score value Y, which is an internal table variable. max The corresponding horizontal combination is the optimal horizontal combination of the ply angles, and proceed to step (8); if not satisfied, the optimal result of the nth orthogonal design of the ply angle of the inner table variable is used as the initial value of the (n+1)th orthogonal design, and return to step (4) to continue the inner table iteration.

[0141] (8) Combine the optimal level of the internal table with the corresponding comprehensive score value Y max Returning to Table 6, calculate the comprehensive score Z for each scheme based on the number of layers in the external variable, and select the scheme with the highest comprehensive score Z'. max To obtain the optimal level of composite material layup number.

[0142] (9) Determine the maximum comprehensive score Z' of the external variables in the iteration step. max Does the result satisfy the condition that five consecutive optimal design results do not increase? If so, then Z' max Z, the maximum score value throughout the entire iteration max And take the corresponding optimal level as the final design result, that is, get the number of plies that meet the conditions; if not, take the optimal result of the nth orthogonal design of the number of plies of the composite material of the external variable as the initial value of the (n+1)th orthogonal design, and return to step (3) until the external design target meets the condition that the number of plies of the composite material of the inner panel of the door does not increase, and get the number of plies of the composite material of the inner panel of the door that meet the conditions and the ply angle of each ply.

[0143] Example 1 shows that the present invention can use a systematic algorithm to determine the ply angle of each layer of the inner panel of the car door, which is the single optimization object, under the condition of changing the number of ply layers of the inner panel composite material, so as to minimize the mass and maximize the stiffness of the inner panel. At the same time, the optimization result satisfies multiple constraints, namely, it is within the specific torsional stiffness and natural frequency of the car door.

[0144] Example 2

[0145] The variable-number-of-variables two-level iterative optimization scheme determination method provided in this invention is applied to the process of determining a lightweight automotive optimization scheme. Since there are two optimization objects, namely the inner panel x1 and the anti-collision bar x2, the number of variable objects, i.e., the external variables, is two: N1 and N2. The internal variable is x. 1i x 2j (i = 1, 2…N1; j = 1, 2…N2). The remaining objective functions and constraints are the same as in Example 1.

[0146] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0147] The embodiments of the present invention have achieved some positive results during the research and development or use process, and have indeed great advantages compared with the prior art. The following content describes them in conjunction with the data, charts and other information of the experimental process.

[0148] Create a door system model in ABAQUS software, such as Figure 4 As shown, the inner door panel is designed as a composite material structure, with T300 as the material. The specific layup method is as follows: Figure 5 As shown. The design variable is the number of ply layers N1∈{2,3,4,5,6,7,8,9,10} of the inner door panel, and the angle of each ply is the design variable x itself. 1i ∈{0°,5°,10°,…,180°}, the initial iteration N1 has levels 1, 2, and 3 as layers 3, 4, and 5 respectively, x 1i The initial levels 1, 2, and 3 were set to 10°, 15°, and 20°, respectively. The proposed algorithm was implemented using MATLAB programming and converged after 14 iterations, as shown in the following results. Figure 6 As shown. To compare with the traditional genetic algorithm, this paper adopts a method of fixing the number of plies N1 and optimizing the ply angle. Finally, the optimal results of different ply numbers are compared, and the maximum value is taken as the final result.

[0149] from Figure 6 As can be seen from this, after the 9th iteration, the algorithm proposed in this invention achieves 5 consecutive Z'... max Since it did not increase, it is considered that the 9th convergence occurred, and the largest Z' was obtained. max The value is set to 14.5, recording intermediate values ​​during the iteration process: including the optimal number of layers, the ply angle for each layer, and the number of ABAQUS calculations called in each iteration. Traditional genetic algorithms optimize each door system with a fixed number of ply layers separately, gradually increasing the number of generations by keeping the population at 30, until Z'... max Convergence: Record the convergence value and the corresponding ply angle. Include performance metrics in... Figure 7 As can be seen from the figure, the algorithm proposed in this paper achieves optimal results similar to those of the traditional method with only 15% of the computational cost.

[0150] 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, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for determining a two-layer iterative optimization scheme with a variable number of variables for optimizing the layup of composite materials in automotive door inner panels, characterized in that, The method for determining a two-layer iterative optimization scheme with a variable number of variables applied to the optimization of composite material layups for automotive door inner panels includes the following steps: Step 1: Determine the optimization objective as finding the minimum mass and maximum stiffness of the car door, and construct an optimization model with the upper torsional stiffness, lower torsional stiffness, and first natural frequency of the car door as constraints; determine the inner panel of the car door as the optimization object, the variable for each optimization object is the layup angle of the i-th layer of the composite material, and the number of auxiliary layers for each optimization object is the number of variables; determine the design space for the number of auxiliary layers of the inner panel composite material and the angle of the auxiliary layers of the inner panel composite material. Step 2: Determine the number of auxiliary layers as variables for outer layer iteration, and use a three-level orthogonal experiment method to establish a level table and orthogonal table of outer layer iteration variables, and determine the number of variables for each optimization object. Step 3: Determine the layup angle of the i-th layer of the composite material as the variable for inner layer iteration. Based on the determined number of variables for outer layer iteration, use a three-level orthogonal experimental design method to establish the level table and orthogonal table of inner layer iteration variables. Step 4: Perform experimental calculations on each sample point of the inner iteration under each scheme to obtain the experimental value and comprehensive score value of each response. Use the penalty function method to attach each constraint to the objective function and calculate the corrected response value of the objective function. Calculate the comprehensive score value of each sample point to obtain the optimal result of this inner iteration. Step 5: Determine whether the optimization objective satisfies the condition for the optimal result of the inner iteration. If it does, return the optimal result obtained from the inner iteration and the comprehensive score of the optimal result to the orthogonal table of the outer iteration variables. If it does not satisfy the condition, take the optimal result of the nth orthogonal design of the inner table variable ply angle as the initial value of the (n+1)th orthogonal design and return to Step 3 to continue the inner iteration. Step 6: Obtain all response values ​​returned by the inner iteration from the orthogonal array of the outer iteration; determine whether the optimization objective satisfies the optimal result condition of the outer iteration. If it does, obtain the final result of the optimization objective; otherwise, shift the design space of each variable and return to step 2. In step six, the orthogonal array of the outer iteration obtains all response values ​​returned by the inner iteration; it is determined whether the optimization objective satisfies the optimal result condition of the outer iteration. If it does, the final result of the optimization objective is obtained; otherwise, the design space of the number of variables is shifted and the process returns to step two, including: The orthogonal array of the outer iteration obtains the optimal result returned by the inner iteration and the comprehensive score value of the optimal result. The outer iteration compares the comprehensive score values ​​of each scheme, selects the scheme with the maximum comprehensive score value, and obtains the optimal level of this outer iteration. Determine whether the maximum comprehensive score obtained from the screening meets the convergence condition: if it does, then take the maximum comprehensive score as the maximum score of the entire outer iteration, and take the optimal level corresponding to the maximum comprehensive score as the final optimization result; if it does not meet the condition, then take the optimal result of the orthogonal design of the outer iteration as the initial value of the next orthogonal design, and return, until the convergence condition is met, and obtain the number of variables that meet the condition. The final optimization results include the number of composite material layups and the layup angle of each layer for the door inner panel that meet the conditions.

2. The method for determining a two-layer iterative optimization scheme with a variable number of variables applied to the layup optimization of composite materials for automotive door inner panels as described in claim 1, characterized in that, The optimization model includes: The multi-objective, multi-constraint optimization model is as follows: In the formula, x1 is the optimization object, i.e., the inner panel of the car door; the number of plies of the optimization object x1 is the number of variables, and its design space is N1-dimensional; the variable of the optimization object x1 itself is the ply angle x of the i-th layer of the composite material. 1i The optimization objective F(X) is the minimum value of the door mass M(X) and the maximum value of the stiffness R(X); the constraints are the upper torsional stiffness G1(X), the lower torsional stiffness G2(X) and the first natural frequency f(X) of the door, where [G1(X)] and [G2(X)] represent the boundary values ​​of the constraint functions.

3. The method for determining a two-layer iterative optimization scheme with a variable number of variables applied to the layup optimization of composite materials for automotive door inner panels as described in claim 1, characterized in that, Step four includes: 1) Calculate the response value F(X) of the objective function within each iteration step; 2) Using the penalty function method, each constraint G(X) is added to the objective function to obtain the corrected response value F'(X) of the objective function: Where s represents the penalty factor, G τ (X) represents the function value of the τth constraint; 3) Obtain the comprehensive score Y for each scheme using the weighted scoring method: Where u represents the test number, v represents the index sequence number, w represents the number of indicators, and Y represents the number of indicators. u b represents the overall score of the u-th trial. v Let y represent the weight of the v-th indicator. 0v y represents the corrected response value of the v-th index in the initial design. uv This represents the corrected response value of the v-th index in the u-th experiment.

4. A system for determining a two-layer iterative optimization scheme with a variable number of variables for optimizing the layup of composite materials for automotive door inner panels, implementing the method for determining a two-layer iterative optimization scheme with a variable number of variables for optimizing the layup of composite materials for automotive door inner panels as described in any one of claims 1-3, characterized in that, The system for determining the two-level iterative optimization scheme with a variable number of variables includes: The parameter setting module is used to determine the optimization objective and constraints and build the optimization model; determine the optimization object, the variables of each optimization object, the number of each variable, and the design space; The initialization module is used to determine the variables of the outer iteration. It uses a three-level orthogonal experimental design method to establish the level table and orthogonal table of the outer iteration variables and determine the number of variables for each optimization object. The module is also used to determine the variables of the inner iteration. Based on the determined number of variables of the outer iteration, it uses a three-level orthogonal experimental design method to establish the level table and orthogonal table of the inner iteration variables. The iterative experiment module is used to perform inner and outer layer iterations based on the level table and orthogonal table of the outer layer iteration variables and the level table and orthogonal table of the inner layer iteration variables. The results output module is used to output the optimal results obtained from the inner and outer layer iterations as the optimization scheme.

5. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor performs the steps of the method for determining a two-layer iterative optimization scheme with a variable number of variables applied to the optimization of composite material layup for automotive door inner panels as described in any one of claims 1-3.

6. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method for determining a two-layer iterative optimization scheme with a variable number of variables applied to the optimization of composite material layup for automotive door inner panels as described in any one of claims 1-3.

7. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the two-layer iterative optimization scheme determination system with a variable number of variables for the optimization of composite material layup of automotive door inner panel as described in claim 4.

Images