A lightweight design method for DD lay-up composite material rotary body based on MIGA-SQP hybrid algorithm

Abstract

The application discloses a lightweight design method for a DD layup composite material rotary body based on a MIGA-SQP hybrid algorithm, relates to the technical field of composite material structure optimization, and comprises the following steps: dividing a composite material rotary body structure into a plurality of design regions, establishing continuous design variables based on a DD layup mode, and establishing a parameterized finite element model template file; establishing a two-stage hybrid optimization strategy; iteratively optimizing the parameterized finite element model template file and the two-stage hybrid optimization strategy; substituting an optimal design scheme into a finite element model for verification; and after verification, generating a manufacturing guidance file according to optimal layup parameters; and the method is used for solving the problems of low optimization efficiency and insufficient utilization of the continuous variable characteristics of the DD layup.

Description

Technical Field

[0001] This invention relates to the field of composite material structure optimization technology, specifically to a lightweight design method for DD layup composite rotating bodies based on the MIGA-SQP hybrid algorithm. Background Technology

[0002] Composite materials, due to their superior specific strength, specific modulus, and excellent designability, have become key materials for weight reduction and efficiency improvement in the aerospace field. In launch vehicle structures, the skirt, as a load-bearing structure connecting sections of different diameters, directly affects the rocket's launch efficiency in terms of its lightweighting level. Traditional composite skirt designs are mostly based on classical laminate theory, employing a layup sequence with fixed angles (such as 0°, ±45°, 90°). When facing complex multi-condition loads, this discrete angle layup method often relies on increasing the number of layups or adopting a conservative design to meet performance requirements, resulting in significant design redundancy in the structure and limiting the space for further lightweighting.

[0003] To overcome the limitations of traditional discrete ply designs, an innovative design concept called Double-Double (DD) ply has attracted attention. The core of DD ply lies in the fact that its basic ply unit consists of two pairs of symmetrical angles, i.e. The entire laminated plate structure is constructed by repeating this basic unit n times, denoted as . The advantages of this design concept include: ply angle and Instead of fixed discrete values, they are design variables that can be optimized in a continuous space according to the load conditions, giving the structure unprecedented design flexibility; the ply structure itself has balanced symmetry, which can effectively suppress warping deformation caused by coupling effect; theoretically, higher material utilization can be achieved.

[0004] Despite the significant theoretical advantages of DD layup, its application in practical engineering structures, particularly in the lightweight design of complex components, still faces numerous technical challenges. Existing solutions largely focus on theoretical verification and exploration of simple components: at the theoretical verification level, existing research has explored stiffness equivalence and layup substitution methods for DD laminates by introducing the principle of Chua's modulus invariance; at the simple component optimization level, existing research has applied the DD layup method to composite C-beams, cap-shaped stiffened structures, etc., combining parametric modeling and optimization algorithms for weight reduction design, achieving some success.

[0005] However, a comprehensive review of existing technical solutions reveals significant limitations, failing to fully leverage the core advantages of DD layup: First, current research largely focuses on simple, regular components such as beams, reinforcement bars, and slabs, lacking sufficient validation of their applicability and effectiveness for real-world engineering products with complex geometry and subjected to multiple loads (such as composite skirts for launch vehicles). Second, in terms of optimization methods, most solutions still rely on heuristic global optimization algorithms such as genetic algorithms and particle swarm optimization. While these algorithms possess global search capabilities, they suffer from slow convergence speeds, high computational costs, and fail to fully utilize the design variables of DD layup. The key mathematical property of continuous differentiability lacks a high-precision optimization framework deeply coupled with this property. Thirdly, from the perspective of the design process, existing solutions often fail to achieve a fully automated closed loop. Manual intervention is still required in steps such as parameter updates, simulation-driven processes, and result extraction, which affects the reliability and repeatability of the design.

[0006] Therefore, we propose a method that can deeply integrate the advantages of DD layup theory and improve optimization efficiency. Summary of the Invention

[0007] The purpose of this invention is to provide a lightweight design method for DD ply composite rotating bodies based on the MIGA-SQP hybrid algorithm, which is used to solve the problems of low optimization efficiency and insufficient utilization of the continuous variable characteristics of DD ply.

[0008] This invention is achieved through the following technical solution:

[0009] A lightweight design method for DD ply composite material rotating bodies based on the MIGA-SQP hybrid algorithm, specifically including: S1. Divide the composite material rotating body structure into multiple design regions, and based on the double-double layup method, determine the layup angle for each design region. , and the number of repetitions of the ply unit Establish continuous design variables and create a parametric finite element model template file; S2. Establish a two-stage hybrid optimization strategy, namely, using the Multi-Island Genetic Algorithm (MIGA) for global exploration and the Sequential Quadratic Programming (SQP) method for local refinement; S3. Iterative optimization is performed based on a parametric finite element model template file and a two-stage hybrid optimization strategy; The iteration continues until the multi-island genetic algorithm outputs an optimized initial point within the global optimal solution region. Subsequently, the sequential quadratic programming method continues iterating from this initial point until it converges to obtain the optimal design scheme that minimizes the total structural mass while satisfying all constraints. This design scheme includes the optimal ply angles for each design region. , and the optimal number of ply unit repetitions The optimal layup parameters are formed; S4. Substitute the optimal design scheme into the finite element model for verification. After successful verification, generate manufacturing guidance documents based on the optimal layup parameters.

[0010] Furthermore, the composite skirt structure is divided into multiple design regions along its height or according to load distribution characteristics.

[0011] Furthermore, the design variables and The value ranges from 0° to 90°, and the number of repetitions of the layup unit is... It can be an integer or a continuous variable greater than or equal to 1.

[0012] Furthermore, the parametric finite element model template file is an Abaqus INP file, which contains layup definition keywords corresponding to each design region. These keyword parameters are related to the design variables. , and Related.

[0013] Furthermore, the execution process of the multi-island genetic algorithm includes: The overall population is divided into multiple subpopulation islands; Each island independently performs selection, crossover, and mutation operations, and migrates individuals between islands according to a set migration rate; The process continues until a preset termination condition is reached, and the optimal individual is output as the initial point of optimization.

[0014] Furthermore, the sequential quadratic programming method utilizes the objective function and constraint functions to apply to continuous design variables. , and The gradient information is used to construct and solve a quadratic programming subproblem to determine the search direction and step size, and the design variables are iteratively updated until convergence.

[0015] Furthermore, each iteration in step S3 includes: S31. Generate a set of design variable values ​​using the optimization algorithm at the current stage; S32. Through automated scripts, update the ply parameters in the parametric finite element model template file according to the design variable values, submit it to the finite element solver for calculation, and extract the total structural mass, displacement and buckling load factor as the response values ​​of the optimization objective and constraint conditions; S33. Feedback the response value to the two-stage hybrid optimization strategy to determine convergence or generate the next generation of design variables.

[0016] Furthermore, the automation script is a Python script that updates the layup keyword parameters in the INP file by reading the parameter file output by the optimization algorithm, calls the finite element solver to perform calculations through system commands, and extracts the simulation response values ​​through the result database interface.

[0017] Furthermore, the constraints include: under specified working conditions, the maximum displacement of the composite skirt structure is less than the allowable value, the first-order linear buckling load factor is greater than the safety threshold, and the maximum stress is less than the allowable stress.

[0018] Furthermore, the optimization process is built within an integrated optimization software platform, and the automated scripts are encapsulated and scheduled through components of the optimization software platform to achieve an automatic closed loop of optimization iteration and simulation analysis.

[0019] The technical solution of the present invention has at least the following advantages and beneficial effects: This invention discloses a lightweight design method for DD ply composite material rotating bodies based on the MIGA-SQP hybrid algorithm. First, it utilizes the powerful global exploration capability of the multi-island genetic algorithm to efficiently explore within a broad DD ply continuous design space, effectively avoiding premature convergence to local optima during the optimization process and locating high-quality initial regions for subsequent optimization. Then, it employs a sequential quadratic programming method for local refinement, utilizing the gradient information of the objective and constraint functions to achieve fast and accurate convergence near the optimal solution. This "coarse-to-fine" strategy significantly reduces the total number of simulation iterations required while ensuring the discovery of a globally optimal solution.

[0020] Furthermore, through the deep integration of parametric finite element model templates and automated scripts, the entire process from design variable updates and finite element analysis to key performance index extraction is automated, so that each iteration does not require manual intervention, eliminating human error and ensuring the consistency and repeatability of the design process. Moreover, this closed-loop system can greatly improve the design efficiency and engineering response speed of complex composite material structures, making the design of real engineering products more precise and faster. Attached Figure Description

[0021] Figure 1This is a schematic diagram of the process for a lightweight design method of DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to the present invention; Figure 2 This is a schematic diagram of the iterative optimization method of the present invention; Figure 3 This is a schematic diagram of a lightweight design system for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0023] Example 1 like Figures 1-2 This paper presents a lightweight design method for DD-layup composite material rotational bodies based on the MIGA-SQP hybrid algorithm. In this embodiment, the composite material rotational body structure is specifically a composite material rotational body with flanges, i.e., a composite skirt structure. The method specifically includes: S1. Divide the composite material rotating body structure into multiple design regions, and based on the double-double layup method, determine the layup angle for each design region. , and the number of repetitions of the ply unit Establish continuous design variables and create a parametric finite element model template file; Specifically, the composite material rotating body structure is divided into multiple design regions along its height or according to load distribution characteristics; and the design variables are... and The value ranges from 0° to 90°, and the number of repetitions of the layup unit is... It can be an integer or a continuous variable greater than or equal to 1.

[0024] Specifically, this step can be broken down as follows: Establish an accurate geometric model of the composite skirt and perform appropriate mesh generation; Based on the load distribution and structural characteristics of the composite skirt, it is divided into multiple design regions, and an independent geometric set is created for each region; the layup parameters for each region, i.e., fiber angles... , and number of ply units , is defined as a continuous design variable for subsequent optimization; Assign composite material properties to each region set and define them using the DD layup method. The initial values ​​of the layup parameters can be set to any reasonable values. Based on actual service conditions, apply accurate boundary conditions and establish multiple analysis steps to simulate different load conditions, such as static and buckling. After completing the above settings, export the .inp file containing all the model information; this file will serve as a parametric data template, and its key feature is that the model data section includes ply definition keywords associated with the design region set (such as...). SHELLSECTION ORIENTATION), these keyword parameters will be recognized and modified by the script program in subsequent steps.

[0025] The core principle of this step in the overall technical solution is to transform the engineering design problem into a highly structured, parameter-driven digital experimental device that seamlessly integrates with subsequent automated processes; this is achieved by "creating an independent geometric set for each region" and "defining the layup keywords (...)". SHELL SECTION "ORIENTATION" is a strong correlation. Step S1 establishes precise anchor points between design variables and physical properties in the model file. This allows subsequent scripts to accurately and error-free change the mechanical properties of the model by modifying the parameter values ​​of these keyword lines, without rebuilding the geometry or mesh, thus ensuring the purity and efficiency of the "input-output" mapping.

[0026] The optimization of traditional discrete layups (0° / 45° / 90°) is a combinatorial problem, while DD layups... and It is a continuous variable; step S1 explicitly states that... , , Defined as "continuous design variables," an interface for continuous assignment is reserved for them in the model template. This allows optimization algorithms (especially SQP) to fine-tune the angle like adjusting a knob (e.g., from 75.2° to 75.3°) and receive immediate feedback in the simulation.

[0027] Furthermore, compared to uniform layup, the zoned variable thickness design can precisely match the load distribution of the structure, further eliminate redundant material, and divide the structure into multiple design zones, defining the number of repetitions of the layup unit independently for each zone. This allows the optimization algorithm to optimize not only the global fiber angle, but also the material distribution along space (which areas should be thick and which areas should be thin).

[0028] In addition, the output parametric finite element model template file is the starting point for the automation process; this template file is structured text, containing keywords (such as...). The SHELL SECTION and its associated set name (such as Set-1) provide a precise "address map" for Python scripts. Scripts can reliably locate and modify the ply parameters of specified regions, much like a program reads and writes to a database, thus enabling unattended batch simulations.

[0029] S2. Establish a two-stage hybrid optimization strategy, namely, use a multi-island genetic algorithm for global exploration and a sequential quadratic programming method for local refinement. This is to address the potential multidimensionality and nonlinearity of the DD ply design space, as well as the sensitivity of gradient algorithms to initial values. By establishing a two-stage hybrid optimization strategy, we can balance global optimization capability with local convergence efficiency, thereby efficiently optimizing the continuous variables of DD ply.

[0030] Specifically, the multi-island genetic algorithm is configured in Isight, setting parameters such as the number of islands, population size per island, maximum number of generations, crossover rate, mutation rate, and migration rate. The problem definition of the multi-island genetic algorithm is: with the objective of minimizing the structural quality, and with each region ( , , The variable is a continuous design variable, and its value range is defined. The constraints of this algorithm include: maximum displacement < allowable value, first-order buckling load factor > safety threshold. Iterative optimization of the multi-island genetic algorithm: The algorithm initializes by generating a random population and assigning individuals to various islands. In each generation, each island independently performs selection, crossover, and mutation operations; individuals are periodically migrated between islands to exchange information. For each individual (i.e., a set of design variables), an automated simulation process is used for evaluation in the second step. The algorithm continues to evolve until it reaches the preset maximum number of generations, and outputs the currently found global optimum as a high-quality initial point.

[0031] Multi-island genetic algorithm, as a first-stage global optimization method, improves the global exploration capability and convergence efficiency of the optimization process through its population structure and migration mechanism, providing high-quality and reliable initial design points for subsequent gradient-based local refinement. Its advantages are mainly reflected in the following aspects: 1) Enhanced global search capabilities to avoid premature convergence. Traditional single-population genetic algorithms (GA) are prone to getting trapped in local optima in complex, multi-dimensional design spaces. Multi-island genetic algorithms divide the total population into multiple subpopulations ("islands") and perform independent genetic evolution (selection, crossover, mutation) on each island, effectively maintaining population diversity. After each island evolves independently for a period, superior individuals are periodically exchanged according to a set migration rate (migration). This mechanism simulates gene flow in biological evolution, spreading superior genes (design schemes) found on different islands and continuously introducing new diversity, thus reducing the risk of the algorithm converging to a local optimum early on.

[0032] 2) Efficient exploration of complex, non-convex design spaces Design variables for DD layup ( , , Although continuous and differentiable, the mapping relationship between the target function and the final structural performance (mass, displacement, buckling factor) is highly nonlinear, and the constraints (strength, stiffness, stability) constitute a complex design feasible region. As a gradient-free optimization method, the multi-island genetic algorithm does not rely on the gradient of the objective function and constraints, and can directly handle non-convex, discontinuous, or derivative-difficult problems. Through multi-island sampling, the algorithm can extensively explore the entire design variable space, making it more likely to locate the region where the global optimum is located.

[0033] 3) Provides an excellent initial point for the second-stage gradient optimization. Sequential Quadratic Programming (SQP) and other gradient-based local optimization algorithms heavily rely on the initial point setting for their convergence speed and final result. An initial point near the global optimum ensures that SQP converges quickly and accurately to a high-quality solution. The multi-island genetic algorithm, through its global exploration capabilities, finds a "current optimal solution" (e.g., (88.32°, 75.46°) in the example that is extremely close to the global optimum. Using this solution as the starting point for SQP effectively avoids the problem of SQP getting trapped in suboptimal local solutions or failing to converge due to improper initial point selection, ensuring the reliability of the two-stage optimization process.

[0034] 4) Adaptability to multidimensional design variable problems When the composite skirt is divided into multiple design areas (such as the 10 areas in the example), the number of design variable dimensions increases significantly (each area has...). , , (Three variables). Multi-island genetic algorithms, through their population mechanism, can effectively handle this type of medium-dimensional continuous variable optimization problem. The independent evolution of multiple islands facilitates simultaneous exploration in different subspaces, and the integration of information through migration makes them better suited to multidimensional problems than single evolutionary strategies, finding possible design directions with reasonable computational costs.

[0035] Furthermore, based on local refinement using sequential quadratic programming, Initialization: The optimal solution obtained in the first stage is used as the initial design point of the sequential quadratic programming method.

[0036] Gradient optimization: due to DD layup variables ( , , Only by ensuring the continuity of the objective function and constraint functions can their gradient information be effectively estimated. The SQP algorithm uses this gradient information to construct and solve a series of quadratic programming subproblems under the condition of satisfying the constraints, in order to determine the optimal search direction and step size, and thus iteratively update the design variables.

[0037] Convergence criterion: Optimization stops when the gradient norm of the objective function is less than the set tolerance, the constraint violation is sufficiently small, or the maximum number of iterations is reached, and the final optimal design scheme is output. .

[0038] S3. Based on a parametric finite element model template file and a two-stage hybrid optimization strategy, an iterative optimization was performed, constructing a closed-loop iterative system of "perception-decision-execution" consisting of intelligent decision-making by the optimization algorithm and precise execution by automated scripts. This achieved a complete optimization iteration cycle, where each iteration includes: S31. Generate a set of design variable values ​​using the optimization algorithm at the current stage; S32. Through automated scripts, update the ply parameters in the parametric finite element model template file according to the design variable values, submit it to the finite element solver for calculation, and extract the total structural mass, displacement and buckling load factor as the response values ​​of the optimization objective and constraint conditions; Specifically, the automation script is a Python script that updates the layup keyword parameters in the INP file by reading the parameter file output by the optimization algorithm, calls the finite element solver to perform calculations through system commands, and extracts the simulation response values ​​through the result database interface.

[0039] The constraints are those of the multi-island genetic algorithm, namely, under the specified working conditions, the maximum displacement of the composite skirt structure is less than the allowable value, the first-order linear buckling load factor is greater than the safety threshold, and the maximum stress is less than the allowable stress.

[0040] S33. Feed the response value back to the two-stage hybrid optimization strategy to determine convergence or generate the next generation of design variables; In addition, the optimization process is built in an integrated optimization software platform, and the automated script is encapsulated and scheduled through the components of the optimization software platform to achieve an automatic closed loop of optimization iteration and simulation analysis.

[0041] This closed loop is seamlessly connected and automated through the scheduling of integrated optimization software platforms (such as Isight). Its unique feature is that it transforms the "simulation analysis" process, which is usually driven, interpreted, and decided manually by engineers, into an automated process that is controlled by algorithms and can run 24 hours a day. This perfectly combines human wisdom with the tireless and absolutely precise execution of machines.

[0042] The iteration continues until the multi-island genetic algorithm outputs the initial optimization point within the global optimal solution region. Then, the sequential quadratic programming method continues iterating from this initial point until it converges to obtain the optimal design scheme that minimizes the total structural mass while satisfying all constraints, i.e., the optimal ply angle for each design region. , and the optimal number of ply unit repetitions ; During the MIGA phase, it can stably handle concurrent or sequential evaluations of hundreds of random individuals. Once MIGA converges, the system automatically and error-free transfers the optimal individual parameters to SQP as the initial point and switches to the iterative logic of SQP to continue running. The entire process is seamless, avoiding errors or interruptions that might occur from manual data transfer, ensuring the precise realization of the hybrid strategy's design intent. Furthermore, this step, through fully automated scripts, reduces the manual operation time for a single simulation to near zero. More importantly, its tight integration with the optimization algorithm ensures that each simulation is "targeted," part of the algorithm's proactive exploration, avoiding ineffective computation. Although the total number of simulations remains high, the "information gain" (contribution to finding the optimal solution) per unit of computation is maximized, making it possible to complete complex optimizations based on high-fidelity models within acceptable total time and computational resources.

[0043] S4. Substitute the optimal design scheme into the finite element model for verification. After successful verification, generate manufacturing guidance documents based on the optimal layup parameters, where the optimal layup parameters refer to... .

[0044] Example 2 For a carbon fiber / epoxy resin composite skirt, its geometry is cylindrical, with a diameter at the upper end... 1386mm, lower diameter 1393mm in length, 657mm in height, and 5.46mm in nominal skin thickness; Design objective: To minimize the mass of the composite skirt structure while satisfying all strength, stiffness, and stability constraints.

[0045] Design variables were defined: DD ply was used for design. For finer optimization, the composite skirt height direction was divided into multiple regions along the negative Y-axis. Each region had corresponding continuous variables defined for ply angle and ply number.

[0046] Set constraints: Strength: Under all operating conditions, the stress of all units of the skirt sleeve skin is <160 MPa.

[0047] Stiffness: Under a 50kg load, the maximum radial displacement of the upper mounting surface of the skirt sleeve is <1.1 mm.

[0048] Stability: Under axial compression, the first-order linear buckling load factor of the structure is >58000.

[0049] Setting loads and operating conditions: Condition 1 (Static): A 50kg gravity load (approximately 500,000 N) is applied to the upper flange mounting surface.

[0050] Condition 2 (Axial Compression): An axial pressure of 10N is applied to the mounting surface of the upper flange for buckling analysis.

[0051] Boundary conditions: The lower end face is fixed.

[0052] Designed using this invention: First, finite element models of the two composite skirts are built using Abaqus / CAE 2022. The models are meshed using S4R shell elements. In the Property module, three independent geometric sets (Set-1, Set-2, ... Set-10) are created for each region (1, 2, ... 10). A composite ply section is created for each set, with its ply angle and number of layers defined parametrically. The initial values ​​can be set to (0, 90, 4), representing... , , Number of times the layer unit is repeated.

[0053] In the Step module, create corresponding load analysis steps for each model: Static and General for static analysis (Case 1), and Buckle for buckling analysis (Case 2). After applying the appropriate loads and boundary conditions, submit the analysis to generate two complete files: composite_static.inp and composite_buckle.inp. Check the inp files with a text editor to confirm that they contain the required information. SHELL SECTION, COMPOSITE and Keywords such as ORIENTATION, and their parameters are associated with the aforementioned set.

[0054] Then, build and optimize the process in Isight 2022.

[0055] Optimization components: Set the objectives and constraints, and use the "Simcode" component to encapsulate the above Python script to automate the simulation. Optimization is performed using the "Multi-Island Genetic Algorithm" and "Sequence Quadratic Programming (SQP)" components.

[0056] Phase 1: Global Exploration (Multi-Island Genetic Algorithm) Algorithm parameters: Number of islands: 5; Population size per island: 10; Maximum number of generations: 10; Migration rate: 10%. Fast global optimization is achieved by controlling these parameters.

[0057] Design variables: 3 variables, all defined as continuous variables, and with reasonable ranges set. .

[0058] Run: Start optimization. Isight automatically drives hundreds of simulations, continuously evolving the population. After 10 generations, the algorithm converges to a globally optimal solution region. Record the optimal angle parameters (88.32°, 75.46°) and the number of repetitions (1) at this point as the starting point for the next stage.

[0059] Phase Two: Local Refinement (Sequential Quadratic Programming, SQP) Algorithm parameters: Maximum number of iterations: 30; Gradient tolerance: 1e-4; Constraint tolerance: 1e-3.

[0060] Initial point: Set to the optimal solution obtained by the multi-island genetic algorithm.

[0061] Operation: The SQP algorithm starts from this point and performs a local fine-grained search using the gradient information of the objective function and constraint functions. After 10 iterations, the convergence condition is met, and the final optimal solution (90°, 74.4°) is obtained.

[0062] Data output: The extracted mass (target), maximum displacement, and buckling factor are written to the results file output.txt for Isight to read and use for optimization iteration.

[0063] Next, the Python script auto_simulation.py was written, which contains the following core functionalities: Parameter Reading and Model Update: The script reads the new ply scheme from the parameter file (input.txt, containing the current values ​​of three variables) passed from Isight. The script locates the ply definition keyword lines corresponding to Set-1, Set-2, ... Set-10 in the inp file and updates them using the new parameters ( , , Replace the original values ​​and generate new files composite_static.inp and composite_buckle.inp. The key logic of the replacement is based on two fundamental perspectives. Generate a balanced minimum ply unit that includes its positive and negative symmetry angles: [+ , - , + , - ], and the unit can be repeated.

[0064] Call the solver: Use system commands to call the Abaqus Standard solver to execute the newly generated inp file: abaqus job= composite_static.inp, composite_buckle.inp.

[0065] Result extraction: After the solution is completed, without opening the CAE interface, directly read the result database composite_static.odb and composite_buckle.odb through the odbAccess module.

[0066] Finally, the final solution was verified: the optimal parameter combination obtained from the optimization, as shown in Table 1, was substituted into the simulation model to perform a complete and high-precision finite element analysis. The results show: The maximum radial displacement is 1.02 mm (<1.1 mm), the first-order buckling load factor is 60261 (>58000), and all constraints are satisfied.

[0067] Layup scheme export: Generate manufacturing guidance documents based on optimal parameters. For example, for region 1, the layup scheme can be described as: "Lay up 4 layers, with fiber orientation angles of each layer according to..." Alternating orientation.

[0068] Table 1: Comparison of Schemes Before and After Optimization

[0069] Conclusion: This embodiment demonstrates that the parametric modeling, automated simulation, and two-stage hybrid optimization method for DD layups provided by this invention can effectively achieve lightweight design of composite material structures. Under strict constraints on strength, stiffness, and stability, the mass of the composite skirt structure was successfully reduced by 14.1%, significantly improving material utilization efficiency and structural performance. Furthermore, the entire process is highly automated and has strong engineering applicability.

[0070] Example 3 like Figure 3 An optimization system for lightweight design of DD ply composite rotating bodies, as shown, includes: The parametric modeling module is used to divide the composite material body-of-rotation structure into multiple design regions, and based on a double-double layup, sets the layup angle for each design region. , and the number of repetitions of the ply unit Establish continuous design variables and generate a parametric finite element model template file containing the association information of the design variables; An optimization strategy management module is used to establish and manage a two-stage hybrid optimization strategy, which includes a multi-island genetic algorithm stage for global exploration and a sequential quadratic programming stage for local refinement. The automated simulation and optimization execution module is used for iterative optimization based on parametric finite element model template files and a two-stage hybrid optimization strategy; it includes a simulation driving unit and an optimization logic unit. The simulation driving unit is configured to update the model template file according to the current design variable values ​​in each iteration, drive the finite element solver to perform calculations, and extract the structural performance response values. The optimization logic unit is configured to run the multi-island genetic algorithm and the sequential quadratic programming method, generate new design variables or determine convergence based on the performance response value, and automatically pass the optimization initial point output by the multi-island genetic algorithm stage to the sequential quadratic programming method stage after the multi-island genetic algorithm stage ends. The result verification and output module is used to substitute the final design scheme output by the automated simulation and optimization execution module into the finite element model for independent verification, and generate manufacturing guidance documents based on the scheme after the verification is passed.

[0071] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A lightweight design method for DD ply composite material rotating bodies based on the MIGA-SQP hybrid algorithm, characterized in that, Specifically, it includes: S1. Divide the composite material rotating body structure into multiple design regions. Based on the Double-Double layup method, establish continuous design variables for the layup angle and the number of repetitions of the layup unit for each design region, and establish a parametric finite element model template file. S2. Establish a two-stage hybrid optimization strategy, namely, use a multi-island genetic algorithm for global exploration and a sequential quadratic programming method for local refinement; S3. Iterative optimization is performed based on the parametric finite element model template file and a two-stage hybrid optimization strategy to obtain the optimal design scheme; S4. Substitute the optimal design scheme into the finite element model for verification. After successful verification, generate manufacturing guidance documents based on the optimal layup parameters in the optimal design scheme.

2. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: The composite rotating body structure is divided into multiple design regions along the height direction or according to the load distribution characteristics.

3. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: The design variables and The value ranges from 0° to 90°, and the number of repetitions of the layup unit is... It can be an integer or a continuous variable greater than or equal to 1.

4. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: The parametric finite element model template file is an Abaqus INP file, which contains layup definition keywords corresponding to each design region. These keyword parameters are related to design variables. , and Related.

5. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: The execution process of the multi-island genetic algorithm includes: The overall population is divided into multiple subpopulation islands; Each island independently performs selection, crossover, and mutation operations, and migrates individuals between islands according to a set migration rate; The process continues until a preset termination condition is reached, and the optimal individual is output as the initial point of optimization.

6. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: The sequential quadratic programming method utilizes objective functions and constraint functions to apply continuous design variables. , and The gradient information is used to construct and solve a quadratic programming subproblem to determine the search direction and step size, and the design variables are iteratively updated until convergence.

7. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 1, characterized in that: Each iteration in step S3 includes: S31. Generate a set of design variable values ​​using the optimization algorithm at the current stage; S32. Through automated scripts, update the ply parameters in the parametric finite element model template file according to the design variable values, submit it to the finite element solver for calculation, and extract the total structural mass, displacement and buckling load factor as the response values ​​of the optimization objective and constraint conditions; S33. Feed the response value back to the two-stage hybrid optimization strategy to determine convergence or generate the next generation of design variables; The iteration continues until the multi-island genetic algorithm outputs an optimized initial point within the global optimal solution region. Subsequently, the sequential quadratic programming method continues iterating from this initial point until it converges to obtain the optimal design scheme that minimizes the total structural mass while satisfying all constraints. This design scheme includes the optimal ply angles for each design region. , and the optimal number of ply unit repetitions The optimal layup parameters are determined.

8. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 7, characterized in that: The automated script is a Python script that updates the layup keyword parameters in the INP file by reading the parameter file output by the optimization algorithm, calls the finite element solver to perform calculations through system commands, and extracts the simulation response values ​​through the result database interface.

9. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 7, characterized in that: The constraints include: under specified working conditions, the maximum displacement of the composite skirt structure is less than the allowable value, the first-order linear buckling load factor is greater than the safety threshold, and the maximum stress is less than the allowable stress.

10. The lightweight design method for DD layup composite material rotating bodies based on the MIGA-SQP hybrid algorithm according to claim 7, characterized in that: The optimization process is built in an integrated optimization software platform. The automated scripts are encapsulated and scheduled through the components of the optimization software platform to achieve an automatic closed loop of optimization iteration and simulation analysis.

Images