A design method, system, device, and medium for stiffened conical shell structures based on Bayesian optimization algorithm.

Abstract

The purpose of this invention is to address the challenges of existing stiffened conical shell structure optimization design methods in terms of computational efficiency, resource usage, and development time. It proposes a design method, system, device, and medium for stiffened conical shell structures based on a Bayesian optimization algorithm to reduce the number of finite element simulation evaluations, thereby shortening the design cycle and reducing resource consumption. The technical solution of this invention is: a design method for stiffened conical shell structures based on a Bayesian optimization algorithm. The specific implementation steps of the method are: defining design variables, determining the optimization objective, constructing a training sample set, creating N sets of weight vectors, randomly selecting a set of weight vectors, parallel optimization, updating the sample set, and determining the termination condition. This invention significantly reduces the number of simulation evaluations and lowers computational costs; it efficiently handles high-dimensional multi-objective optimization problems in engineering, improving design accuracy; it effectively avoids getting trapped in local optima through a Pareto optimization strategy; and it shortens the design cycle.

Description

Technical Field

[0001] This invention relates to the field of ship structural engineering optimization design, and in particular to a design method, system, equipment and medium for stiffened conical shell structures based on Bayesian optimization algorithms. Background Technology

[0002] With the continuous advancement of marine engineering and shipbuilding technology, the demand for lightweight and high-performance ship structures is increasing. While ensuring structural strength and stiffness, how to reduce the weight of ship structures and improve their load-bearing capacity and fatigue resistance has become a key issue in ship design. As a supporting structure for the shafting system, the vibration characteristics of stiffened conical shells play a decisive role in the stability and acoustic performance of the entire underwater structure. Due to the special nature of the underwater environment, the vibrations generated by the rotation of the shafting system not only affect the structural safety of the structure itself but may also interfere with acoustic performance such as sonar detection and communication. Therefore, reducing the structural response of stiffened conical shell structures under shafting rotation excitation is of great significance for improving the overall performance and concealment of underwater structures. Stiffened conical shell structures, with their unique geometry and structural characteristics, have been widely used in key parts of ships such as hulls and decks.

[0003] The design of stiffened conical shells requires comprehensive consideration of multiple factors, including material properties, geometric parameters, and load conditions, to achieve structural optimization. Currently, in the design phase of stiffened conical shells, numerical simulation of the mechanical properties of the structure is typically performed using finite element analysis. While this method provides relatively accurate simulation results, the computational process is complex and consumes significant time and computational resources. Furthermore, multi-objective optimization problems often require trade-offs between multiple performance indicators, further increasing the complexity of the design process. Traditional multi-objective optimization algorithms, such as Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Genetic Algorithm (MOGA), while finding a feasible solution in some cases, often require extensive simulation evaluation when dealing with high-dimensional, nonlinear, and multimodal optimization problems, and are prone to getting trapped in local optima. Summary of the Invention

[0004] To address the challenges of computational efficiency, resource utilization, and development time in existing methods for optimizing the design of stiffened conical shell structures, this invention proposes a design method, system, device, and medium for stiffened conical shell structures based on a Bayesian optimization algorithm. This reduces the number of finite element simulation evaluations, thereby shortening the design cycle and reducing resource consumption.

[0005] The technical solution adopted in this invention is:

[0006] A design method for a stiffened conical shell structure based on Bayesian optimization algorithm includes the following steps:

[0007] Step S1: Define design variable d, which includes the dimensional parameters of the stiffener and the plate thickness; the set of design variables is represented as follows. The i-th design variable d i The range of values ​​is ,in ;

[0008] Step S2: Determine the optimization objective; the optimization objective includes maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight ;

[0009] Step S3: Based on the value range of the design variable d mentioned in step S1, the design variable d is sampled in the design space using a sampling scheme. The finite element software is used to perform simulation calculations on each sampling point. The sampling points and optimization target results are organized to construct a training sample set.

[0010] Step S4: Create N sets of weight vectors The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective;

[0011] Step S5: Randomly select from the N sets of weight vectors created in step S4 The weight vector described in the group;

[0012] Step S6: Use the Bayesian optimization algorithm to optimize the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point;

[0013] Step S7: The product generated in step S6... Each candidate sample point is added to the training sample set constructed in step S3, the training sample set is updated, and the simulation design continues.

[0014] Step S8: Determine whether the design of the reinforced conical shell structure has reached the preset maximum number of iterations. If so, output the current Pareto optimal solution; otherwise, repeat steps S5 to S8.

[0015] Furthermore, in the parallel optimization of step S6, simultaneous generation... When there are 10 candidate sample points, the process for generating each candidate sample point is the same, specifically including the following steps:

[0016] Step S61: Use a weighted sum method to transform the multiple optimization objectives from step S2 into a single scalar value;

[0017] Step S62: Construct a Gaussian process model based on the training sample set created in step S3 and the scalar value obtained in step S61;

[0018] Step S63: Randomly select from the n design variables d in step S1 Individual design variables, and based on Each of the aforementioned sub-design variables generates a new subspace design variable sample point;

[0019] Step S64: Use the crossover and mutation operator of the Bayesian optimization algorithm to calculate the new subspace design variable sample points generated in step S63, and generate new design variable sample points; for each design variable sample point, use the Gaussian process model constructed in step S62 to calculate the EI value corresponding to each design variable sample point;

[0020] Step S65: Determine whether the Bayesian optimization algorithm has reached the maximum number of iterations. If so, select the design variable sample point with the largest EI value as the first iteration. If the candidate sample points of the optimization process are not selected, then steps S64 to S65 are re-executed.

[0021] Furthermore, in step S61, the formula for converting multiple optimization objectives into a single scalar value using a weighted sum method is as follows:

[0022] ,

[0023] in, Indicates the first The value of the optimization objective;

[0024] Each component Indicates the first Weights of each optimization objective

[0025] To optimize the number of targets.

[0026] Furthermore, in step S63, p ≥ n / 2.

[0027] Furthermore, the sampling scheme used in step S3 is the Latin hypercube sampling scheme.

[0028] Furthermore, in step S3, the number of sampling points is twice or more than twice the number of design variables d.

[0029] Furthermore, the preset maximum number of iterations in step S8 is 40 to 200 times.

[0030] Based on the same inventive concept, this invention also provides a design system for a stiffened conical shell structure based on a Bayesian optimization algorithm, implementing the aforementioned design method for a stiffened conical shell structure based on a Bayesian optimization algorithm, including:

[0031] The definition module is used to define design variables d, which include the dimensional parameters of the stiffeners and the plate thickness; the set of design variables is represented as follows. The i-th design variable d i The range of values ​​is ,in ;

[0032] The determination module is used to determine the optimization objective; the optimization objective includes maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight .

[0033] The sampling and simulation module is used to sample the design variable d in the design space according to the value range of the design variable d in step S1, and to perform simulation calculations on each sampling point using finite element software; at the same time, it organizes the sampling points and optimization target results to construct a training sample set.

[0034] The creation module is used to create N sets of weight vectors. The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective.

[0035] The selection module is used to randomly select from the N sets of weight vectors created in step S4. The weight vector described in the group.

[0036] The optimization module is used to apply a Bayesian optimization algorithm to the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point;

[0037] The update module is used to update the module generated in step S6. Each candidate sample point is added to the training sample set constructed in step S3, the training sample set is updated, and the simulation design continues.

[0038] The judgment module is used to determine whether the design of the stiffened conical shell structure has reached the preset maximum number of iterations. If so, the current Pareto optimal solution is output; otherwise, the selection module, the optimization module, and the update module are reused to perform the corresponding operations, and the judgment module makes a new judgment.

[0039] Based on the same inventive concept, the present invention also provides a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the aforementioned design method for a stiffened conical shell structure based on a Bayesian optimization algorithm.

[0040] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing computer instructions, which are used to cause a computer to execute the aforementioned design method for a stiffened conical shell structure based on a Bayesian optimization algorithm.

[0041] The beneficial effects of this invention are:

[0042] 1. Compared with the prior art, the present invention significantly reduces the number of simulation evaluations and lowers the computational cost.

[0043] 2. This invention, through parallel optimization, can generate simultaneously The system uses a number of candidate sample points, thereby shortening the design cycle, efficiently handling high-dimensional multi-objective optimization problems in engineering, and improving design accuracy.

[0044] 3. This invention effectively avoids getting trapped in local optima by using a Pareto optimization strategy, thus accelerating the transformation of products from concept to market. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart illustrating a design method for a reinforced conical shell structure based on a Bayesian optimization algorithm, according to the present invention.

[0047] Figure 2 This is a schematic diagram of the finite element method for constructing a reinforced conical shell structure according to the present invention;

[0048] Figure 3 This is a hypervolume iterative curve diagram for the optimization of the dimensions of the reinforced conical shell according to the present invention. Detailed Implementation

[0049] To address the challenges of computational efficiency, resource utilization, and development time in existing methods for optimizing the design of stiffened conical shell structures, this invention proposes a design method, system, device, and medium for stiffened conical shell structures based on a Bayesian optimization algorithm. This reduces the number of finite element simulation evaluations, thereby shortening the design cycle and reducing resource consumption.

[0050] The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0051] The invention will now be further described with reference to the accompanying drawings.

[0052] Reference Figure 1 As shown, this embodiment is based on a design method for a stiffened conical shell structure using a Bayesian optimization algorithm, including the following steps:

[0053] Step S1: Define design variable d, which includes the dimensional parameters of the stiffener and the plate thickness; the set of design variables is represented as follows. The i-th design variable d i The range of values ​​is ,in .

[0054] In this embodiment, the design variables of the stiffened conical shell structure include the dimensions of the stiffeners and the plate thickness. The radius of the large end and the radius of the small end of the stiffened conical shell are 5000 mm and 3000 mm, respectively. The conical shell is divided by 11 stiffeners spaced 1000 mm apart. These 11 stiffeners have different dimensions, resulting in 12 plate segments with different thicknesses. The elastic modulus of the stiffened conical shell material is... Poisson's ratio is 0.3, and density is The design variables can be defined as the dimensional parameters of 11 stiffeners (each stiffener has 4 variables) and 12 plate thicknesses. The lower bound of the 4 variables for each stiffener is [200, 10, 100, 10] mm, and the upper bound is [340, 24, 240, 24] mm. The value range of each plate thickness is [50, 64] mm.

[0055] Step S2: Determine the optimization objective; In this embodiment, there are 5 performance parameters for the optimization objective, including maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight The performance parameters of the optimization target were obtained through simulation analysis of the stiffened conical shell structure using the commercial software ANSYS. Figure 2 A schematic diagram of the finite element simulation is shown.

[0056] Step S3: Based on the value range of the design variable d described in Step S1, the design variable d is sampled in the design space using the Latin hypercube sampling scheme. The finite element software is used to perform simulation calculations on each sampling point. The sampling points and optimization target results are organized to construct a training sample set.

[0057] Since the number of design variables d in the optimization problem is 56 (the size parameters of the 11 stiffeners in step S1 * the size parameters of each stiffener have 4 variables + 12 plate thicknesses), it belongs to a high-dimensional optimization problem. In order to obtain better optimization results, the number of sampling points in this embodiment is 112.

[0058] Step S4: Create N sets of weight vectors The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective.

[0059] In this example, the number of weight vectors is set to 20. The number of weight vectors should be set according to the optimization objective and the characteristics of the problem structure.

[0060] Step S5: Randomly select from the N sets of weight vectors created in step S4 The weight vectors described herein are used to prepare for the subsequent parallel optimization process.

[0061] In this embodiment, 5 sets of weight vectors are randomly selected from the N sets of weight vectors created in step S4, namely... =5, which means that 5 simulation calculations can be run simultaneously. This parameter represents the setting for this embodiment only, and the specific setting value can be adjusted appropriately according to the simulation conditions.

[0062] Step S6: In this embodiment, the Bayesian optimization algorithm is used to optimize the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point.

[0063] In this embodiment, the parallel optimization process can generate 5 candidate sample points simultaneously, thereby shortening the design cycle. To further improve optimization efficiency, the number of sub-design variables is set to 35 (the number of sub-design variables is greater than half the number of design variables). Meanwhile, for... There are 10 candidate sample points, and the process for generating each candidate sample point is the same, specifically including the following steps:

[0064] Step S61: Use a weighted sum method to transform the multiple optimization objectives from step S2 into a single scalar value; the calculation is as follows:

[0065] ,

[0066] in, Indicates the first The value of the optimization objective;

[0067] Each component Indicates the first Weights of each optimization objective

[0068] To optimize the number of targets.

[0069] In this embodiment, the design variable with the smallest scaling value is stored in In it, its smallest scale value is stored in middle.

[0070] Step S62: Construct a Gaussian process model based on the training sample set created in step S3 and the scalar values ​​obtained in step S61.

[0071] Step S63: Randomly select from the n design variables d in step S1 p sub-design variables are used to generate new subspace design variable sample points based on the p sub-design variables.

[0072] Step S64: Use the crossover and mutation operator of the Bayesian optimization algorithm to calculate the new subspace design variable sample points generated in step S63, and generate new design variable sample points; for each design variable sample point, use the Gaussian process model constructed in step S62 to calculate the EI value corresponding to each design variable sample point.

[0073] In this embodiment, the crossover and mutation operator of the genetic algorithm is used to generate new design variable sample points and calculate the EI value for each sample. The EI is calculated as follows:

[0074] ,

[0075] and These are the predicted values ​​and standard deviations of the Gaussian process model, respectively. and Let represent the probability density function and cumulative distribution function of the standard normal distribution, respectively. This represents the best scaling value for the current optimization process. When calculating the EI value, other sub-variables besides the selected sub-dimension should be used... Replace the corresponding positional variable.

[0076] Step S65: Determine whether the Bayesian optimization algorithm has reached the preset maximum number of iterations, such as 30 times; if it has reached 30 times, then select the design variable sample point with the largest EI value as the first iteration. If the candidate sample points of the optimization process are not selected, then steps S64 to S65 are re-executed.

[0077] Step S7: Generate the parallel optimization solution from step S6... Each candidate sample point is added to the training sample set described in step S3, the training sample set is updated, and the simulation design continues to be executed to prepare for the next step.

[0078] Step S8: Determine whether the design of the stiffened conical shell structure has reached the preset maximum number of iterations, such as 40 times; if so, output the current Pareto optimal solution; otherwise, repeat steps S5 to S8.

[0079] The hypervolume is the volume of the region in the target space enclosed by the non-dominated solution set obtained by the Bayesian optimization algorithm and the reference point; a larger hypervolume value indicates better overall algorithm performance. This embodiment uses the hypervolume value to analyze the optimization results. Figure 3The iterative curve of the hypervolume is shown. It can be seen that the parallel multi-objective Bayesian optimization design method can find a relatively optimal solution set in about 17 iterations. Compared with evolutionary algorithms that require tens of thousands of calculations, this greatly shortens the design cycle and can achieve better optimization results with fewer simulation optimizations.

[0080] This embodiment is based on a design method for stiffened conical shell structures using a Bayesian optimization algorithm. A design system for stiffened conical shell structures based on a Bayesian optimization algorithm is designed, including:

[0081] The definition module is used to define design variables d, which include the dimensional parameters of the stiffeners and the plate thickness; the set of design variables is represented as follows. The i-th design variable d i The range of values ​​is ,in .

[0082] The determination module is used to determine the optimization objective; the optimization objective includes maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight .

[0083] The sampling and simulation module is used to sample the design variable d in the design space according to the value range of the design variable d in step S1, and to perform simulation calculations on each sampling point using finite element software; at the same time, it organizes the sampling points and optimization target results to construct a training sample set.

[0084] The creation module is used to create N sets of weight vectors. The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective.

[0085] The selection module is used to randomly select from the N sets of weight vectors created in step S4. The weight vector described in the group.

[0086] The optimization module is used to apply a Bayesian optimization algorithm to the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point.

[0087] The update module is used to update the module generated in step S6. Each candidate sample point is added to the training sample set constructed in step S3, the training sample set is updated, and the simulation design continues.

[0088] The judgment module is used to determine whether the design of the stiffened conical shell structure has reached the preset maximum number of iterations. If so, the current Pareto optimal solution is output; otherwise, the selection module, the optimization module, and the update module are reused to perform the corresponding operations, and the judgment module makes a new judgment.

[0089] This embodiment is based on a design method for a stiffened conical shell structure based on a Bayesian optimization algorithm. A computer device is designed, including a memory and a processor. The memory and the processor are interconnected. The memory stores computer instructions. The processor executes the computer instructions to perform the design method for a stiffened conical shell structure based on a Bayesian optimization algorithm in this embodiment.

[0090] This embodiment is based on a design method for a stiffened conical shell structure based on a Bayesian optimization algorithm. A computer-readable storage medium is designed, on which computer instructions are stored. The computer instructions are used to enable the computer to execute the design method for a stiffened conical shell structure based on a Bayesian optimization algorithm in this embodiment.

[0091] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A design method for a stiffened conical shell structure based on Bayesian optimization algorithm, characterized in that, Includes the following steps: Step S1: Define design variable d, which includes the dimensional parameters of the stiffener and the plate thickness; The set of design variables is represented as follows: The i-th design variable The range of values ​​is ,in ; Step S2: Determine the optimization objective; the optimization objective includes maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight ; Step S3: Based on the value range of the design variable d mentioned in step S1, the design variable d is sampled in the design space using a sampling scheme. The finite element software is used to perform simulation calculations on each sampling point. The sampling points and optimization target results are organized to construct a training sample set. Step S4: Create N sets of weight vectors The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective; Step S5: Randomly select from the N sets of weight vectors created in step S4 The weight vector described in the group; Step S6: Use the Bayesian optimization algorithm to optimize the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point; Step S7: The product generated in step S6... Each candidate sample point is added to the training sample set constructed in step S3, the training sample set is updated, and the simulation design continues. Step S8: Determine whether the design of the reinforced conical shell structure has reached the preset maximum number of iterations. If so, output the current Pareto optimal solution; otherwise, repeat steps S5 to S8.

2. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 1, characterized in that, In the parallel optimization of step S6, the following steps are performed simultaneously: When there are 10 candidate sample points, the process for generating each candidate sample point is the same, specifically including the following steps: Step S61: Use a weighted sum method to transform the multiple optimization objectives from step S2 into a single scalar value; Step S62: Construct a Gaussian process model based on the training sample set created in step S3 and the scalar value obtained in step S61; Step S63: Randomly select from the n design variables d in step S1 Individual design variables, and based on Each of the aforementioned sub-design variables generates a new subspace design variable sample point; Step S64: Use the crossover and mutation operator of the Bayesian optimization algorithm to calculate the new subspace design variable sample points generated in step S63, and generate new design variable sample points; for each design variable sample point, use the Gaussian process model constructed in step S62 to calculate the EI value corresponding to each design variable sample point; Step S65: Determine whether the Bayesian optimization algorithm has reached the maximum number of iterations. If so, select the design variable sample point with the largest EI value as the first iteration. If the candidate sample points of the optimization process are not selected, then steps S64 to S65 are re-executed.

3. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 2, characterized in that, The formula for converting multiple optimization objectives into a single scalar value using the weighted sum method in step S61 is as follows: ， in, Indicates the first The value of the optimization objective; Each component Indicates the first Weights of each optimization objective To optimize the number of targets.

4. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 2, characterized in that, In step S63, p ≥ n / 2.

5. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 1, characterized in that, The sampling scheme used in step S3 is the Latin hypercube sampling scheme.

6. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 1, characterized in that, The number of sampling points in step S3 is twice or more than twice the number of design variables d.

7. The design method for a stiffened conical shell structure based on Bayesian optimization algorithm according to claim 1, characterized in that, The preset maximum number of iterations in step S8 is 40 to 200.

8. A design system for a stiffened conical shell structure based on a Bayesian optimization algorithm, implementing the design method for a stiffened conical shell structure based on a Bayesian optimization algorithm as described in any one of claims 1 to 7, characterized in that, include: A definition module is used to define design variable d, which includes the dimensional parameters of the stiffener and the plate thickness. The set of design variables is represented as follows: The i-th design variable d i The range of values ​​is ,in ; The determination module is used to determine the optimization objective; the optimization objective includes maximizing the overall buckling stress of the structure. Minimize the circumferential stress on the mid-surface of the conical shell. Minimize the longitudinal stress on the inner surface of the conical shell. Minimize the internal stress of the stiffeners and minimizing the total structural weight ; The sampling and simulation module is used to sample the design variable d in the design space according to the value range of the design variable d in step S1, and to perform simulation calculations on each sampling point using finite element software. Simultaneously, the sampling points and optimization results are organized to construct a training sample set; The creation module is used to create N sets of weight vectors. The weight vector is used to subsequently integrate the optimization objectives of multiple reinforced conical shell structures into a single optimization objective; Defined in a normalized space; where each component Indicates the first The weights of each optimization objective; The selection module is used to randomly select from the N sets of weight vectors created in step S4. The weight vector described in the group; The optimization module is used to apply a Bayesian optimization algorithm to the randomly selected step S5. The weight vectors described in the group are optimized in parallel, and at the same time, the following is generated: 1 candidate sample point; The update module is used to update the module generated in step S6. Each candidate sample point is added to the training sample set constructed in step S3, the training sample set is updated, and the simulation design continues. The judgment module is used to determine whether the design of the stiffened conical shell structure has reached the preset maximum number of iterations. If so, the current Pareto optimal solution is output; otherwise, the selection module, the optimization module, and the update module are reused to perform the corresponding operations, and the judgment module makes a new judgment.

9. A computer device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform a design method for a stiffened conical shell structure based on a Bayesian optimization algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute a design method for a stiffened conical shell structure based on a Bayesian optimization algorithm as described in any one of claims 1 to 7.

Images