CPU parallel parameterization level set topological optimization method for CAD model in any shape

Through the CPU parallel parameterized horizontal set topology optimization method, combined with implicit identification model and parallel finite element analysis, the problems of low computational efficiency of existing topology optimization methods in complex structures and poor compatibility of CAD models are solved, achieving high efficiency and precision The topological optimization design is suitable for engineering applications of large-scale complex structures.

CN118964038AActive Publication Date: 2024-11-15江淮前沿技术协同创新中心 +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411158811.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-15
Estimated Expiration
2044-08-22

AI Technical Summary

Technical Problem

The existing topology optimization methods face problems such as low computing efficiency, poor compatibility with CAD models, and limited computing scale when dealing with complex structures, which are difficult to meet the industry's demand for large-scale complex structure optimization.

Method used

The CPU parallel parameterized horizontal set topology optimization method for arbitrary shape CAD model is adopted. Through implicit identification model, CPU parallel finite element analysis and parameterized horizontal set method, efficient topology optimization of the structure is achieved, and the CAD and CAE are solved. Problems of insufficient communication, poor grid generation quality, and low computing efficiency.

Benefits of technology

It significantly improves the topological optimization efficiency and accuracy of complex structures, provides an efficient and practical optimization design tool that can handle complex CAD models, solves the insufficient communication between CAD and CAE and poor grid generation quality in the existing technology. , the problem of low computing efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118964038A_ABST
    Figure CN118964038A_ABST
Patent Text Reader

Abstract

The invention discloses a CPU (central processing unit) parallel parameterization level set topological optimization method for a CAD (computer aided design) model in any shape, belongs to the technical field of high-performance design, and solves the problems that CAD and CAE (computer aided engineering) are split mutually, the grid generation quality is poor and time is consumed, and the large-scale grid calculation efficiency is low. By introducing a parameterization level set method, a parallel finite element analysis technology and seamless integration with a CAD model, topological optimization efficiency and precision of a complex structure are remarkably improved, and the method has a wide engineering application prospect; not only are many problems of a traditional topological optimization method solved, but also an efficient and practical optimization design tool is provided for the field of engineering design. According to the method, when a complex CAD model is processed, the efficiency and precision of topological optimization can be remarkably improved, the problems of insufficient communication between CAD and CAE, poor grid generation quality, low calculation efficiency and the like in the prior art are solved, and the method has wide engineering application prospects.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of high-performance design of high-precision and advanced engineering structures, and relates to a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes. Background Art

[0002] With the rapid development of modern industry, the requirements for structures in the field of engineering design have become increasingly complex and diverse, and traditional structural topology optimization methods are unable to cope with these complex requirements. Structural topology optimization is an optimization method that seeks the best material distribution within a given design space to meet specific performance indicators. However, with the increase in design requirements, especially in high-demand fields such as aerospace, automobile manufacturing, and construction engineering, the limitations of traditional topology optimization methods have gradually emerged, making it difficult to meet the high-precision, high-efficiency, and high-complexity requirements of actual engineering applications.

[0003] Most of the emerging topology optimization methods currently focus on effectiveness research in the two-dimensional design domain, while ignoring the solution efficiency and engineering practicality in the three-dimensional design domain. Existing methods often require large computing resources and time costs when dealing with three-dimensional structures. In particular, traditional methods show obvious performance bottlenecks when dealing with complex geometric structures and large-scale grids. This lack of computational efficiency not only limits the application of topology optimization in complex engineering, but also leads to delays in the optimization process, affecting the timeliness and economy of the design.

[0004] Secondly, many newly proposed topology optimization methods are mostly implemented based on advanced computing software platforms such as MATLAB. These platforms have certain advantages in algorithm verification and small-scale calculations, but in actual engineering applications, due to the limitation of computing power, they perform poorly when processing large-scale three-dimensional grids. Usually, these methods can only process hundreds of thousands to millions of grid cells, and the computational efficiency is low, which cannot meet the needs of the industry for large-scale complex structure optimization. This lack of performance has greatly restricted these methods in actual engineering applications and made it difficult to achieve their due effects.

[0005] In addition, due to the different basis functions between CAD (computer-aided design) and CAE (computer-aided engineering), existing topology optimization methods often cannot interact directly with CAD models. This obstacle leads to a large gap between the optimized design and the actual engineering model, increasing the workload and time cost in the early stage of design. In the traditional design process, CAD models usually need to be manually adjusted and regenerated many times to meet the requirements of CAE analysis. This not only increases the complexity of the design, but may also cause the optimization results to be inconsistent with the actual engineering requirements, thus affecting the performance and quality of the final product.

[0006] The development of computer technology, especially the progress of multi-core processors and parallel computing technology, has provided a new way to solve the above problems. Parallel computing technology can greatly improve computing efficiency by processing multiple computing tasks at the same time. Especially in large-scale finite element analysis and topology optimization, the application prospects of parallel computing technology are very broad. The use of parallel computing technology can significantly reduce the calculation time in the topology optimization process and improve the accuracy and reliability of the optimization results. However, in current research and applications, how to effectively integrate parallel computing technology into the topology optimization design process is still a problem that needs to be solved.

[0007] In summary, existing topology optimization methods face problems such as low computational efficiency, poor compatibility with CAD models, and limited computational scale when dealing with complex structures. These problems not only limit the application of topology optimization methods in practical engineering, but also affect the quality and performance of design results. Summary of the invention

[0008] The technical solution of the present invention is used to solve the problems of the separation between CAD and CAE, poor quality and time-consuming mesh generation, and low efficiency of large-scale mesh calculation.

[0009] The present invention solves the above technical problems through the following technical solutions:

[0010] A CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes, comprising:

[0011] Implicit recognition model: scaling and translating the geometric information of the CAD model stored in the STL file format; after completing the scaling and translation of the model, calculating the value of the signed distance field for each node in the background grid; after the signed distance field is generated, the level set function is initialized;

[0012] CPU parallel finite element analysis: The structural model is divided into multiple sub-areas, and the concepts of local domain and ghost domain are introduced by defining private areas and shared areas, and parallel finite element analysis is performed in each sub-area;

[0013] Parameterized level set topology optimization: It combines parameterization with the level set method to generate different topological configurations by adjusting parameters. The level set method is used to describe the boundaries of the structure and optimize the topology of the structure by gradually evolving these boundaries. By incorporating multiple working conditions into the topology optimization process, the structure is optimized simultaneously under multiple working conditions, thereby designing a structural solution with higher stability and reliability.

[0014] Furthermore, the method of the scaling operation is as follows: by traversing the coordinate information of each vertex in the STL file, the maximum and minimum values ​​of the model in the three directions of x, y, and z are found, thereby calculating the size of the model in each normal direction; the size of the background grid in each direction is divided by the size of the model in the corresponding direction to obtain three scaling factors; the smallest scaling factor is selected and applied to all coordinate values ​​of the model to complete the scaling operation of the model.

[0015] Furthermore, the translation operation method is as follows: traverse all vertices of the model again, obtain their minimum x, y, z coordinate information, and then subtract the corresponding minimum values ​​from the x, y, z coordinates of all vertices to complete the translation operation of the model.

[0016] Furthermore, the method for calculating the value of the signed distance field is as follows: traverse each node in the background grid to determine its position relative to the model; if the node is outside the model, the value of its signed distance field is negative; if the node is inside the model, its value is positive; the absolute value of the signed distance field is equal to the minimum distance from the node to the model surface.

[0017] Furthermore, the level set function value of each node is calculated through the value of the signed distance field and the compactly supported radial basis function; the zero isosurface of the level set function is intercepted, and the initial model described by the level set function is obtained.

[0018] Furthermore, when the radius of the compact support region is equal to one unit length, the shared area is defined as the set of all cells within one unit length from the partition boundary, and the private area is the cell that does not belong to the shared area.

[0019] Furthermore, the local domain is composed of the private area and the shared area of ​​the current processor core, including all grid cells that the processor core is responsible for and their corresponding CSRBF tight supports and internal cells, ensuring that the processor has all necessary node information when performing CSRBF calculations.

[0020] Furthermore, the ghost domain is the shared area information obtained by the current processor core from the adjacent processor core, providing the adjacent node data required by the nodes located at the boundary of the local domain.

[0021] An electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above-mentioned CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes, and the processor is configured to execute the program stored in the memory.

[0022] A storage medium stores a computer program, which, when executed by a processor, executes the steps of the above-mentioned CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes.

[0023] The advantages of the present invention are:

[0024] The parallel computing design framework for topology optimization proposed in the present invention significantly improves the efficiency and accuracy of topology optimization of complex structures by introducing a parameterized level set method, parallel finite element analysis technology, and seamless integration with CAD models, and has broad prospects for engineering application. The framework not only solves many problems of traditional topology optimization methods, but also provides an efficient and practical optimization design tool for the field of engineering design. The method proposed in the present invention can significantly improve the efficiency and accuracy of topology optimization when processing complex CAD models, solve the problems of insufficient communication between CAD and CAE, poor mesh generation quality, low computational efficiency, etc. in the prior art, and has broad prospects for engineering application. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a technical solution flow chart of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0026] Figure 2 It is a flow chart of an implicit recognition model of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0027] Figure 3 A schematic diagram of model scaling of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0028] Figure 4 A schematic diagram of model translation of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0029] Figure 5 This is a flow chart of distance field calculation of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0030] Figure 6 A schematic diagram of Boolean operations of a signed distance field of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0031] Figure 7 A shared area and private area component graph of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0032] Figure 8 It is a definition diagram of the local domain and the "shared domain" of the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0033] Fig. 9 This is a graph showing the variation trend of the computational efficiency of the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes with the number of cores according to the first embodiment of the present invention;

[0034] Fig.10 The topology optimization result of a single working condition of a jet engine bracket using the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention;

[0035] Fig.11 The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention is applied to the multi-operating condition topology optimization results of a jet engine;

[0036] Fig.12 This is the result of applying the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to the first embodiment of the present invention to the topology optimization of VR glasses parts. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0039] Embodiment 1

[0040] like Figure 1 The figure shows a flow chart of a CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes provided by this embodiment. The present invention can realize efficient, high-performance, and high-precision topology optimization design of complex feature structures by combining implicit recognition models, CPU parallel finite element analysis, and parameterized level set topology optimization.

[0041] 1. Implicit recognition model

[0042] In the design process of the present invention, the implicit recognition model plays a vital role in accurately mapping the geometric shape of the CAD model to the background mesh and laying the foundation for subsequent topology optimization. Specifically, the model includes three main steps: scaling and translation of the model, generation of the signed distance field, and initialization of the level set function, such as Figure 2 shown.

[0043] (1) Scaling and translation of the model

[0044] In the implementation of the implicit recognition model, the geometric information of the CAD model (stored in STL file format) needs to be scaled and translated first to ensure that it better fits the background mesh. The purpose of this step is to enable more cells in the background mesh to participate in subsequent analysis and topology optimization, thereby improving computational efficiency and optimization accuracy.

[0045] Specifically, first, by traversing the coordinate information of each vertex in the STL file, find the maximum and minimum values ​​of the model in the three directions of x, y, and z, and then calculate the size of the model in each normal direction. Then, divide the size of the background grid in each direction by the size of the model in the corresponding direction to obtain three scaling factors. Select the smallest scaling factor and apply it to all coordinate values ​​of the model to complete the scaling operation of the model, such as Figure 3 shown.

[0046] Next, the scaled model needs to be translated. Traverse all the vertices of the model again, obtain their minimum x, y, and z coordinate information, and then subtract the corresponding minimum value from the x, y, and z coordinates of all vertices to complete the translation operation of the model, such as Figure 4 This process ensures the reasonable positioning of the model in the background grid, laying the foundation for the subsequent steps.

[0047] (2) Generation of signed distance field

[0048] After completing the scaling and translation of the model, the next step is to calculate the value of the signed distance field for each node in the background mesh. The signed distance field is a key data structure that describes the geometry of the model. Its value represents the distance from the background mesh node to the model surface, and distinguishes the inside and outside positions of the node relative to the model by its sign.

[0049] In the specific calculation process, each node in the background grid is first traversed to determine its position relative to the model. If the node is outside the model, the value of its signed distance field is negative; if the node is inside the model, its value is positive. The absolute value of the signed distance field is equal to the minimum distance from the node to the model surface, such as Figure 5 shown.

[0050] This process generates a three-dimensional signed distance field in the background grid, which is used to describe the distribution of the model in the background grid. The accurate generation of the signed distance field is crucial for the subsequent level set function calculation. It not only directly affects the initialization accuracy of the level set function, but also has a profound impact on the final topology optimization results.

[0051] In order to conveniently apply boundary conditions, different STL files need to be input to form different signed distance fields, and a series of Boolean calculations need to be performed for different signed distance fields. The specific implementation process is as follows: Figure 6 shown.

[0052] (3) Initialization of the level set function

[0053] After the signed distance field is generated, the next step is to initialize the level set function. The level set function is an implicit function used to describe the boundaries of geometric shapes. In the present invention, the level set function value of each node in the background grid can be calculated through the value of the signed distance field and the compactly supported radial basis function (CSRBFs).

[0054] First, the level set function value of each node is calculated by using the value of the signed distance field and CSRBFs. CSRBFs has the characteristic of local support, which can improve the smoothness and accuracy of the level set function while maintaining computational efficiency. Next, the zero isosurface of the level set function is intercepted to obtain the initial model described by the level set function. This model will be used as the initial design in the topology optimization process and directly applied to subsequent optimization calculations.

[0055] Through the above steps, the implicit recognition model realizes the seamless connection between the geometric information of the CAD model and the background mesh, providing accurate initial data for the topology optimization process. This method improves the accuracy of topology optimization and is an indispensable and important part of the present invention.

[0056] Traditional topology optimization methods usually require multiple data conversions and manual adjustments between CAD models and optimization models, which increases the complexity and time cost of the design. To solve this problem, the present invention proposes a design framework that can directly interact with the CAD model, so that the optimization results can be seamlessly applied to actual engineering design.

[0057] The framework enables seamless two-way interaction between the CAD model and the optimization model by directly generating a parameterized level set function corresponding to the optimization model in the CAD system. During the optimization process, the designer can directly adjust the geometric parameters of the CAD model according to the needs, and the optimization framework will automatically generate the level set function and perform optimization calculations. The final optimization results can be exported to the CAD model format, and thanks to the smooth characteristics of the level set modeling, the frequent smoothing and model reconstruction in the traditional method is avoided, which greatly improves the design efficiency and accuracy.

[0058] 2. CPU parallel finite element analysis

[0059] The present invention adopts a parallel finite element analysis method to improve the computational efficiency of topology optimization. In parallel computing, the task load is balanced by dividing the grid into multiple parts and distributing them to different processor cores, avoiding queuing of tasks on a single core, thereby achieving efficient parallel processing. However, due to the characteristics of CSRBF, simple grid division will lead to some problems, so the present invention proposes an improved grid division and processing strategy to ensure the accuracy and efficiency of parallel computing.

[0060] In the process of topology optimization, finite element analysis (FEA) is a key step in calculating structural response and evaluating structural performance. However, when faced with large-scale three-dimensional grids and complex geometric structures, the computational efficiency of traditional finite element analysis is often difficult to meet actual engineering needs. To this end, the present invention adopts parallel computing technology, which greatly improves the efficiency of finite element analysis by executing multiple computing tasks simultaneously through multi-core processors.

[0061] Specifically, the present invention divides the structural model into multiple sub-regions and performs parallel finite element analysis in each sub-region. This method can not only fully utilize the computing power of multi-core processors, but also significantly reduce the overall computing time while ensuring the calculation accuracy. In addition, through the reasonable scheduling of parallel computing tasks, the design framework of the present invention can efficiently process large-scale three-dimensional grids, ensuring the efficiency and reliability of the optimization process.

[0062] (1) Overview of Parallel Strategy

[0063] In traditional parallel computing strategies, the entire computational domain is usually divided into several sub-regions, and then these sub-regions are assigned to different processor cores for calculation. The core idea of ​​this method is to achieve uniform distribution of computing tasks among multiple processor cores through task decomposition and allocation, thereby avoiding the waste of computing resources. However, the characteristics of CSRBF (compactly supported radial basis function) bring challenges to this simple grid division.

[0064] A notable feature of CSRBF is that the value of the level set function of a given node depends not only on the coordinate information of the node itself, but also on the information of the nodes around it. This means that if the grid is only divided into multiple non-overlapping subdomains, the nodes located at the subdomain boundary will not be able to access the relevant information belonging to other subdomains, resulting in inaccurate calculation results. Therefore, a meshing method that takes into account the interdependence between nodes must be adopted to ensure that each processor core can obtain the accurate data required for calculation, thereby optimizing the parallel processing efficiency of topology optimization based on CSRBF.

[0065] (2) The concepts of private and shared areas in grid division

[0066] In order to meet the above challenges, Figure 7 As shown, this embodiment proposes the concept of private area and shared area, and introduces the division strategy of local domain and shadow domain on this basis, such as Figure 8 This method aims to effectively manage the information exchange between subdomains and ensure the calculation accuracy of CSRBF.

[0067] Specifically, when meshing, the shared region is first defined as the set of elements within a certain range from the partition boundary. For example, when the radius of the compact support domain is equal to one unit length, the shared region is defined as the set of all elements within one unit length from the partition boundary. The private region includes all other elements, that is, those that do not belong to the shared region.

[0068] In parallel computing, each processor core is responsible for the computing tasks of the subdomain to which it is assigned. The private area represents the area that the processor core is solely responsible for, while the shared area is the area shared with adjacent subdomains. In order to ensure the accuracy of the calculation, the information in the shared area needs to be exchanged between adjacent processor cores.

[0069] (3) Division and application of local domain and ghost domain

[0070] In the actual computing process, in order to better organize and manage the data exchange between the sub-domains, the present invention proposes the concept of dividing the local domain and the ghost domain.

[0071] Local domain: The local domain consists of the private area and shared area of ​​the current processor core. It includes all the grid cells that the core is responsible for and their corresponding CSRBF tight supports and internal cells, ensuring that the processor has all the necessary node information when performing CSRBF calculations.

[0072] Ghost domain: The ghost domain is the shared area information obtained by the current processor core from the adjacent processor core. This information is crucial for the CSRBF calculation of the current core because it provides the adjacent node data required by the nodes at the boundary of the local domain.

[0073] In each iteration step of parallel computing, each processor core first calculates the CSRBF in its local domain. Then, the processor core copies the information of its shared area to the adjacent cores to form the ghost domain of these cores. This information exchange process ensures that each core can accurately consider the influence of surrounding nodes when calculating the CSRBF of nodes in its local domain, thereby improving the calculation accuracy.

[0074] (4) Optimizing parallel computing efficiency

[0075] Through the above-mentioned grid division and information exchange strategy, the present invention effectively solves the challenges brought by node interdependence in CSRBF calculation. Specifically, by defining private areas and shared areas, and introducing the concepts of local domain and ghost domain, each processor core can accurately obtain all necessary data in parallel calculation, avoiding calculation errors caused by lack of boundary node information.

[0076] In addition, this strategy also significantly optimizes the efficiency of parallel computing. In traditional methods, since partition boundary nodes cannot access all necessary information, additional calculations or repeated processing are usually required to make up for these deficiencies, resulting in a waste of computing resources. However, through the method of the present invention, each processor core can complete all necessary CSRBF calculations in one calculation without repeated processing, thereby improving the overall computing efficiency. The specific computing efficiency changes with the number of cores as follows Fig. 9 shown.

[0077] In summary, the CPU parallel finite element analysis method proposed in this invention successfully overcomes the challenges in CSRBF parallel computing by introducing concepts such as private area, shared area, local domain and ghost domain, and realizes efficient and accurate topology optimization calculation. This method not only improves the calculation speed, but also ensures the accuracy of the calculation results, providing strong technical support for the topology optimization of complex engineering structures.

[0078] 3. Parameterized Level Set Topology Optimization

[0079] Parametric level set topology optimization is an innovative design method proposed in this paper, which is mainly used to achieve topology optimization of complex structures. By combining parametric design with the level set method, this method can not only improve the accuracy of topology optimization, but also effectively meet the needs of multi-condition design.

[0080] The level set method is a very effective mathematical tool for dealing with complex topological changes such as splitting and merging. It represents the structure in the design space as an implicit function so that the boundary of the structure can be defined by the zero level set of the level set function. Compared with the traditional voxel method, the level set method has higher accuracy when dealing with highly complex structures. However, when applied to topology optimization, the existing level set method still has the problems of low computational efficiency and difficulty in direct interaction between the optimization process and the CAD model.

[0081] In order to solve these problems, the present invention introduces parameterization processing on the basis of the traditional level set method. By parameterizing the level set function, the controllability and accuracy of the design variables in the optimization process are further enhanced. This parameterized level set method can not only effectively represent the geometric characteristics of the structure, but also flexibly adjust the optimization objectives and constraints according to the design requirements, thereby achieving efficient optimization of complex structures.

[0082] (1) Parameterization: Parameterization refers to the generation of different topological configurations by adjusting parameters during the design process. This method can overcome the problems of the traditional level set method, such as the inability to generate voids, unstable iterations, and slow convergence. In the present invention, parameterization is combined with the level set method to make the topology optimization process more flexible and efficient.

[0083] (2) Level set method: The level set method is a numerical technique based on implicit function representation of geometric shapes. In the present invention, the level set method is used to describe the boundaries of the structure and optimize the topology of the structure by gradually evolving these boundaries.

[0084] (3) Multi-condition topology optimization: Multi-condition topology optimization is another feature of the present invention. In practical engineering applications, structures often need to maintain stability and performance under multiple conditions. The present invention can optimize the structure under multiple conditions simultaneously by incorporating multiple conditions into the topology optimization process, thereby designing a structural solution with higher stability and reliability.

[0085] In complex engineering applications, structures often need to meet performance requirements under multiple working conditions. For example, building structures may need to consider the effects of static loads and wind loads at the same time, and aerospace structures may need to cope with load changes under different flight conditions at the same time. Therefore, structural topology optimization under multiple working conditions becomes the key to improving the overall performance of the structure.

[0086] The present invention first defines and analyzes different working conditions. In multi-condition topology optimization, working conditions generally refer to the stress state of a structure under different use conditions or environments. To this end, the present invention calculates and evaluates the structural response under each working condition through finite element analysis. When performing the analysis, the present invention also takes into account the correlation and mutual influence between different working conditions to ensure that the final optimization result can show good performance under multiple working conditions. The present invention introduces a setting and adjustment mechanism for working condition weights to achieve comprehensive optimization under multiple working conditions. Specifically, each working condition is assigned a different weight value according to its importance. During the optimization process, the working condition weight will affect the calculation and optimization direction of the objective function. When the weight of a certain working condition is higher, the optimization process will be more inclined to meet the performance requirements under that working condition. Through this flexible weight setting and adjustment mechanism, the present invention can balance the contradictions between multiple working conditions and generate an optimized design that performs well under various working conditions.

[0087] In summary, the CAD parallel computing design framework proposed in the present invention significantly improves the efficiency and accuracy of topological optimization of complex structures by introducing parameterized level set methods, parallel finite element analysis technology and seamless integration with CAD models, and has broad prospects for engineering applications. This framework not only solves many problems of traditional topological optimization methods, but also provides an efficient and practical optimization design tool for the field of engineering design. Through the above-mentioned technical means, the method proposed in the present invention can significantly improve the efficiency and accuracy of topological optimization when processing complex CAD models, solve the problems of insufficient communication between CAD and CAE, poor mesh generation quality, low computational efficiency, etc. in the prior art, and has broad prospects for engineering applications.

[0088] The following is the specific application of the method of the present invention:

[0089] Application 1: Jet engine bracket

[0090] Taking the engineering structure of a jet engine bracket as an example, the single working condition topology optimization and multi-working condition topology optimization effects of the present invention are tested respectively. For a single working condition, four single working conditions are used for topology optimization. In these four numerical design cases, the volume fraction of the material in the optimization is defined as 20%. The optimization design results of the four design cases are Fig.10 The first design case is presented in the upper left area, and the lower left area shows the design results of the second design case. The optimized solution for the third design case is presented in Fig.10 The upper right part of Fig.10 The lower right part of the figure provides the design topology of the fourth design case. In addition, in order to more clearly show the optimization solutions in four different cases, Fig.10 The optimized topology is provided in six different perspectives, including isometric view, front view, back view, left view, top view and bottom view.

[0091] For multiple working conditions, the multiple load conditions of the jet engine bracket, that is, the above four different load conditions are considered in the optimization. The corresponding optimization design results are shown in Fig.11 As shown, 6 different views of the topological structure are also included. It can be seen that in the current design, the jagged boundary design problem of the density method can be successfully eliminated, which is derived from the use of the level set function to build an implicit identification model. In general, the optimization results of the jet engine bracket under several different boundary conditions can strongly demonstrate the effectiveness of the invention.

[0092] Application 2: VR glasses parts

[0093] In order to further demonstrate the powerful ability of the proposed CAD-oriented parallel computing design framework on extremely complex structures, the engineering design case of VR glasses is taken as an example. Figure 2 (a). It can be easily seen that the corresponding structural geometry is much more complex than the above design case, where a large number of thin and special structural features exist within the design domain. Relatively speaking, the parallel design of such structures is difficult to solve.

[0094] The topology of the project structure after optimization is as follows Fig.12 As shown. We can easily see that even though the components of the VR glasses have complex geometric shapes, several thin strips and geometric features can be generated in the final optimized topology. The proposed CAD-oriented parallel computing design framework uses implicit identification models to optimize engineering structures, showing its superior ability. At the same time, the final design also has the characteristics of smooth boundaries and clear interfaces inside the solid cavity, which can provide more benefits for later manufacturing. In summary, the optimization results of the above VR glasses can clearly show the powerful ability of the proposed CAD-oriented parallel computing design framework in solving the optimization problems of complex geometric structures in engineering applications. Compared with existing engineering or industrial software, the proposed framework considers both parallel optimization design with computational cost savings and the effectiveness of accurate implicit identification models, and has considerable engineering potential. Therefore, the proposed CAD-oriented parallel computing design framework has considerable potential in future engineering applications, which can not only promote the development and expansion of parameterized level set topology optimization methods in practical engineering fields, but also effectively and fully connect modern industrial CAD modeling software to topology optimization through the developed implicit representation model.

[0095] An electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes in Example 1, and the processor is configured to execute the program stored in the memory.

[0096] A storage medium stores a computer program, which, when executed by a processor, executes the steps of the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes in embodiment 1.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes, characterized in that: include: Implicit recognition model: scaling and translation operations are performed on the geometric information of CAD models stored in STL file format; After completing the scaling and translation of the model, the value of the signed distance field is calculated for each node in the background grid; after the signed distance field is generated, the level set function is initialized; CPU parallel finite element analysis: The structural model is divided into multiple sub-areas, and the concepts of local domain and ghost domain are introduced by defining private areas and shared areas, and parallel finite element analysis is performed in each sub-area; Parameterized level set topology optimization: It combines parameterization with the level set method to generate different topological configurations by adjusting parameters. The level set method is used to describe the boundaries of the structure and optimize the topology of the structure by gradually evolving these boundaries. By incorporating multiple working conditions into the topology optimization process, the structure is optimized simultaneously under multiple working conditions, thereby designing a structural solution with higher stability and reliability.

2. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 1, characterized in that: The scaling operation method is as follows: by traversing the coordinate information of each vertex in the STL file, the maximum and minimum values ​​of the model in the three directions of x, y, and z are found, thereby calculating the size of the model in each normal direction; the size of the background grid in each direction is divided by the size of the model in the corresponding direction to obtain three scaling factors; the smallest scaling factor is selected and applied to all coordinate values ​​of the model to complete the scaling operation of the model.

3. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 2, characterized in that: The method of the translation operation is as follows: traverse all vertices of the model again, obtain their minimum x, y, z coordinate information, and then subtract the corresponding minimum value from the x, y, z coordinates of all vertices to complete the translation operation of the model.

4. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 1, characterized in that: The method for calculating the value of the signed distance field is as follows: traverse each node in the background grid to determine its position relative to the model; if the node is outside the model, the value of its signed distance field is negative; if the node is inside the model, its value is positive; the absolute value of the signed distance field is equal to the minimum distance from the node to the model surface.

5. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 4, characterized in that: The level set function value of each node is calculated by the value of the signed distance field and the compactly supported radial basis function; the zero isosurface of the level set function is intercepted to obtain the initial model described by the level set function.

6. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 1, characterized in that: When the radius of the compactly supported domain is equal to one unit length, the shared area is defined as the set of all cells within one unit length from the partition boundary, and the private area is the cell that does not belong to the shared area.

7. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 1, characterized in that: The local domain is composed of the private area and the shared area of ​​the current processor core, including all grid cells that the processor core is responsible for and their corresponding CSRBF tight supports and inner cells, ensuring that the processor has all necessary node information when performing CSRBF calculations.

8. The CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes according to claim 1, characterized in that: The ghost domain is the shared area information obtained by the current processor core from the adjacent processor core, providing the adjacent node data required by the nodes located at the boundary of the local domain.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes as described in any one of claims 1 to 8, and the processor is configured to execute the program stored in the memory.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the CPU parallel parameterized level set topology optimization method for CAD models of arbitrary shapes as described in any one of claims 1 to 8 are executed.

Citation Information

Patent Citations

  • Structural topology optimization design method based on stress punishment and self-adaptive volume

    CN107273613A

  • Large-scale parallel mesh generation system and method for finite element analysis

    CN111125949A

  • Topological optimization method and system for linear elastic structure

    CN116611213A

  • Double-scale connectable topological optimization method based on material field series expansion

    CN116861559A

  • Parameterized level set topological optimization method and system based on unstructured grid

    CN117744291A