Optimization method and system for MBB beam

Abstract

The invention discloses an optimization method and system for MBB beams, and is applied to the technical field of topological optimization, and the method comprises the steps: building an initial configuration library according to a random point distribution and random link method; optimizing the initial configuration library based on a mobile variable component method, a homogenization method, a structure optimization theory and finite element analysis to obtain optimized unit cells, and forming a unit cell database; training and predicting the component parameters and the performance indexes in the unit cell database through a neural network model to obtain performance index predicted values; and deploying the trained neural network model to practical application for monitoring and maintenance.

Description

Technical Field

[0001] The present invention relates to the technical field of topology optimization, and in particular to an optimization method and system for MBB beams. Background Art

[0002] Materials are usually composed of multiple component materials. By adjusting the geometric shape and size of the component materials, the properties and distribution of the component materials, the macroscopic properties of the composite material can be changed to meet the special requirements of the structure for the material properties. In this way, the designability of the composite material and its advantages such as high specific strength, specific stiffness, and good physical properties are achieved, making it easier and more likely to design high-performance structures. Because it not only maintains the advantages of the properties of each component material, but also obtains comprehensive performance that cannot be achieved by a single component material through the complementarity and association of the properties of each component. However, the design of composite material structures involves different design levels at the macro and micro levels, and the synthesis is complex and difficult to complete. In 1988, Bendsoe and Kikuchi proposed a topology optimization method based on homogenization theory and introduced the concept of microstructure, which greatly promoted the development of continuum topology optimization. Since then, a large number of effective topology optimization framework methods have emerged, such as the variable density method SIMP and the level set method Levelset.

[0003] In the prior art, Guo et al. proposed a new topology optimization method called mobile deformable component framework, which is completely different from the existing optimization methods in the field of optimization. This method introduces a set of deformable components as the basic components for optimization. By optimizing the shape, length, width, direction and layout of these basic components, the optimal topology structure can be obtained. And because the mobile deformable component method uses parameters to explicitly describe curves / surfaces, which is the basic operation of computer graphics, this method has a natural advantage in docking with CAD systems. At present, the same view is adopted by Norato et al., who are based on the topology optimization of continuous structures based on the SIMP framework. The research on structural topology optimization based on the Level set framework is a hot topic that has developed in the past decade. The Level set method is a numerical method proposed by Osher and Sethian in 1988, which was originally used to study dynamic interface tracking. The central idea of ​​this method is to use the zero equipotential surface of the higher-dimensional level set function to implicitly express the surface under study, and realize the evolution of the surface (line) under a specific velocity field. The Level set method was originally used by Wang et al. and Allaire et al. to solve topology optimization problems. Based on the sensitivity information of the response functional, they constructed a velocity field that reduces the target functional value, and updated the level set function by solving the hyperbolic conservation equation to achieve the purpose of topological evolution.

[0004] At present, although the Level set method has achieved great success in the field of topology optimization, the evolution of the level set function requires solving hyperbolic conservation equations, so the time step of the evolution is limited by the CFL condition, which makes the convergence speed of this method lower than that of other topology optimization methods. In addition, since the Level set method uses the level set function to implicitly describe the structural boundary, which is different from the explicit expression of the model under the CAD system, it also makes it difficult to obtain explicit geometric information.

[0005] In order to overcome these defects, the present application proposes an optimization method and system for MBB beams. Summary of the invention

[0006] The purpose of this application is to provide an optimization method and system for MBB beams, aiming to solve the problem that the prior art has too many implicitly expressed design variables and cannot directly handle new types of sets.

[0007] To achieve the above objectives, this application provides the following technical solutions:

[0008] The present application provides an optimization method for MBB beams, including:

[0009] Establish an initial configuration library based on random point distribution and random linking methods;

[0010] The initial configuration library is optimized based on a mobile variable component method, a homogenization method, a structural optimization theory and a finite element analysis to obtain an optimized unit cell and form a unit cell database;

[0011] The component parameters and performance indicators in the unit cell database are trained and predicted by a neural network model to obtain predicted values ​​of the performance indicators;

[0012] Deploy the trained neural network model to actual application for monitoring and maintenance.

[0013] Furthermore, in the step of establishing the initial configuration library according to the random point arrangement and random linking method, the following steps are specifically included:

[0014] Define basic parameters, including but not limited to: spatial range, number of points, and link generation rules;

[0015] Use the rand function to randomly generate the coordinates of points in the specified space;

[0016] Store a linked list of each point based on an empty cell array;

[0017] Traverse each point, and if the distance is within the maximum link distance, randomly decide whether to establish a link;

[0018] Use MATLAB's plotting capabilities to visualize the points and links.

[0019] Furthermore, the step of optimizing the initial configuration library based on the mobile variable component method, the homogenization method, the structural optimization theory and the finite element analysis to obtain the optimized unit cell and form the unit cell database specifically includes the following steps:

[0020] Construct the topological description function corresponding to each component and obtain the material properties by adjusting the design variables of the component;

[0021] Select structural flexibility as the objective function and volume upper limit as the constraint function for optimization;

[0022] Three-dimensional finite element analysis and sensitivity analysis are performed using a substitute material model algorithm.

[0023] Furthermore, the expression of the topology description function is:

[0024]

[0025] Where x, y, z are global coordinate systems, x′, y′, z′ are local coordinate systems; p is a positive integer; L1 is the semi-axis length of the component in the x direction;

[0026]

[0027] and

[0028]

[0029] where s a = sinα,s b = sinβ,s t = sinθ,

[0030] Furthermore, the expression of the design variable of the component is:

[0031] D=(x0,y0,z0,L1,s a ,s b ,s t ,h 1 ,h 2 ,h 3 , t 1 , t 2 , t 3 ) T

[0032] Where x0, y0, z0 are the center coordinates of the component; h and t are parameter vectors.

[0033] Furthermore, in the step of using the substitute material model algorithm to perform three-dimensional finite element analysis and sensitivity analysis, the following steps are specifically included:

[0034] The stiffness matrix of the finite element numbered e is:

[0035]

[0036] Where Ω e is the area occupied by the finite element; B is the strain matrix; D0 is the elastic matrix of the solid material, ρ e For the region Ω e The volume ratio of the area occupied by the internal solid material to the entire unit area;

[0037] Sensitivity analysis is performed and the variation of the flexibility objective functional pair is as follows:

[0038]

[0039] where δ is the Dirac delta function and δ i (x s )=min(δ(χ i ),δ(χ s )),i=1,…,n.

[0040] Furthermore, in the step of training and predicting the component parameters and performance indicators in the unit cell database by using a neural network model to obtain the predicted value of the performance indicator, the following steps are specifically included:

[0041] Select a convolutional neural network, determine the number of network layers, the number of neurons in each layer, the activation function, set the number of input layer nodes to match the number of component parameters, and set the number of output layer nodes to match the number of predicted performance indicators;

[0042] Split the dataset into training, validation, and test sets;

[0043] Train the neural network based on the training set data and adjust the network weights and biases through the back-propagation algorithm;

[0044] During the training process, a validation set is used to monitor the performance of the neural network model, and the network structure and hyperparameters are adjusted according to the performance of the training set;

[0045] The trained neural network model is evaluated based on the test set, and the accuracy and mean square error are calculated; the new component parameters are input into the trained neural network model to obtain the corresponding performance indicator prediction values.

[0046] This application proposes an optimization system for MBB beams, including:

[0047] Establishing module: Establishing the initial configuration library based on random point distribution and random linking methods;

[0048] Optimization module: optimizing the initial configuration library based on the mobile variable component method, homogenization method, structural optimization theory and finite element analysis to obtain optimized unit cells and form a unit cell database;

[0049] Training module: training and predicting the component parameters and performance indicators in the unit cell database through a neural network model to obtain predicted values ​​of the performance indicators;

[0050] Deployment module: deploy the trained neural network model to actual applications for monitoring and maintenance.

[0051] The present application provides a device, which includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing an optimization method for MBB beams; and the processor is used to execute the program instructions stored in the memory to implement an optimization method for MBB beams.

[0052] The present application provides a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute an optimization method for an MBB beam.

[0053] The present application provides an optimization method and system for MBB beams, which have the following beneficial effects:

[0054] (1) Based on the node-driven mobile deformable component method, the unit cell topology description is realized, and the designed unit cell structure has displayed geometric information and smooth and clear structural boundaries. It can be seamlessly integrated with the CAD / CAE system and the manufacturing constraints can be controlled. At the same time, there are only a few design variables;

[0055] (2) Improve the accuracy of computational simulations and reduce the R&D cycle from discovery to application of new materials;

[0056] (3) Through three-dimensional modeling, an optimal structure with the best structure, minimal material consumption and maximum bearing capacity is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a flow chart of an optimization method for MBB beams according to Example 1 of the present application;

[0058] Figure 2 This is a schematic diagram of the initial distribution of components of Example 1 of the present application;

[0059] Figure 3 This is a schematic diagram of the optimal topological structure of Example 1 of the present application;

[0060] Figure 4This is a schematic diagram of the initial design under the encrypted grid of Example 1 of the present application;

[0061] Figure 5 This is a schematic diagram of the optimal topological structure obtained under the encrypted grid of Example 1 of the present application;

[0062] Figure 6 A schematic diagram of using two initial component distributions as comparative examples in Example 1 of the present application;

[0063] Figure 7 This is a schematic structural diagram of an optimization system for MBB beams according to Embodiment 2 of the present application;

[0064] Figure 8 This is a schematic diagram of the device structure of Example 3 of the present application;

[0065] Fig. 9 This is a schematic diagram of the storage medium structure of Example 4 of the present application. DETAILED DESCRIPTION

[0066] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0067] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0068] Example 1

[0069] See also Figure 1 , is a flow chart of an optimization method for MBB beams according to Example 1 of the present application; the steps include:

[0070] S1: Establish an initial configuration library based on random placement and random linking methods.

[0071] In this embodiment, basic parameters are defined, including but not limited to: spatial range, number of points, and link generation rules; the coordinates of points are randomly generated in the specified space using the rand function; a matrix or list for storing link information is initialized according to the number of points, and the link information is stored using an adjacency matrix or adjacency list; each point is traversed, and if the distance is within the maximum link distance, it is randomly decided whether to establish a link; and the drawing function of MATLAB is used to visualize the points and links.

[0072] S2: Optimizing the initial configuration library based on the mobile variable component method, the homogenization method, the structural optimization theory and the finite element analysis to obtain the optimized unit cell and form a unit cell database.

[0073] In this embodiment, the step of optimizing the initial configuration library includes:

[0074] Construct the topological description function corresponding to each component, and obtain the material properties by adjusting the design variables of the component. Construct the topological description function of the 3D component similar to a prism, and the expression is:

[0075]

[0076] Where x, y, z are global coordinate systems, x′, y′, z′ are local coordinate systems; p is a positive integer; L1 is the semi-axis length of the component in the x direction;

[0077] where g(x′) and f(x′, y′) are expressed as:

[0078]

[0079] and

[0080]

[0081] where s a = sinα,s b = sinβ,s t = sinθ,

[0082] The expressions of the design variables of the components are:

[0083] D=(x0,y0,z0,L1,s a ,s b ,s t ,h 1 ,h 2 ,h 3 , t 1 , t 2 , t 3 ) T

[0084] Where x0, y0, z0 are the center coordinates of the component; h and t are parameter vectors.

[0085] Structural flexibility is selected as the objective function and volume upper limit is selected as the constraint function for optimization. For ease of understanding and simplified calculations, only a single material is considered in the optimization problem. Structural flexibility is an indicator that reflects the degree of deformation of a structure when subjected to external forces. During the optimization process, the goal is usually to minimize structural flexibility, that is, to maximize the stiffness of the structure, so that the deformation of the structure under the same load is minimized. At the same time, in order to control cost, weight or material usage, a volume upper limit is usually set as a constraint, which means that during the optimization process, the total volume of the structure cannot exceed this preset upper limit value.

[0086] In the optimization with structural flexibility as the objective function, the objective function is usually expressed as the integral of the strain energy or compliance of the structure under a specific load. In addition to the volume upper limit, other constraints such as stress constraints, displacement constraints, etc. can be added as needed. These constraints will ensure that the optimized structure meets specific performance requirements.

[0087] Three-dimensional finite element analysis and sensitivity analysis are performed using a substitute material model algorithm.

[0088] The stiffness matrix of the finite element numbered e is:

[0089]

[0090] Where Ω e is the area occupied by the finite element; B is the strain matrix; D0 is the elastic matrix of the solid material, ρ e For the region Ω e The volume ratio of the area occupied by the internal solid material to the entire unit area; and by setting ρ e =1×10 -6 , if To avoid the occurrence of singularity in the overall stiffness matrix. The value of at each node of the unit e can be calculated, and this method is sufficiently precise and effective.

[0091] Sensitivity analysis is performed and the variation of the flexibility objective functional pair is as follows:

[0092]

[0093] where δ is the Dirac delta function and δ i (x s )=min(δ(X i ),δ(χ s )),i=1,…,n. If and only if When δ i (X s (x))≠0.

[0094] Furthermore, the three-dimensional finite element analysis specifically includes: clarifying the specific object of analysis, the purpose of analysis and the required accuracy. Obtaining relevant physical and mechanical properties data of substitute materials, such as elastic modulus, Poisson's ratio, yield strength, etc. Using CAD software or based on physical measurement data, a three-dimensional geometric model of the substitute material is established. The geometric model is then divided into a finite number of small units, which will be used for subsequent calculations and analysis. Input the physical and mechanical properties data of the substitute material into the finite element model to ensure that each unit has the correct material properties. According to the purpose of the analysis, appropriate boundary conditions and loads are applied to the model, such as fixed constraints, forces, temperatures, etc. Finally, the solver will perform numerical analysis on the model based on the input material properties, boundary conditions and loads. Start the solver to calculate and obtain the stress, strain, displacement and other response data of the model. Then post-process the structure for easy understanding and analysis.

[0095] Sensitivity analysis is used to evaluate the impact of model parameter changes on output results. Commonly used sensitivity indicators include local sensitivity indicators and global sensitivity indicators. Select the appropriate sensitivity indicator according to the analysis requirements. Set a reasonable range of changes for the parameters to be analyzed and calculate the impact of parameter changes on the output results. Based on the analysis results, identify the parameters that have the greatest impact on model performance and provide guidance for optimal design.

[0096] S3: Train and predict the component parameters and performance indicators in the unit cell database through a neural network model to obtain performance indicator prediction values.

[0097] In this embodiment, component parameters and corresponding performance indicator data are collected from a single cell database. These data should cover a wide range of scenarios and conditions to ensure the generalization ability of the model. Preprocess the data: clean data, select features, standardize / normalize data, and partition data. Split the data set into a training set, a validation set, and a test set.

[0098] The neural network is trained based on the training set data, and the network weights and biases are adjusted through the back-propagation algorithm; during the training process, the validation set is used to monitor the performance of the neural network model, and the network structure and hyperparameters are adjusted according to the performance of the training set; the trained neural network model is evaluated based on the test set, and the accuracy and mean square error are calculated; the new component parameters are input into the trained neural network model to obtain the corresponding performance indicator prediction values.

[0099] Furthermore, according to the nature of the problem and the characteristics of the data, a suitable neural network type is selected, such as a fully connected network (MLP), a convolutional neural network (CNN), or a recurrent neural network (RNN). For the relationship modeling between component parameters and performance indicators, if there are many input parameters and the relationship is complex, MLP can be selected.

[0100] S4: Deploy the trained neural network model to actual applications for monitoring and maintenance.

[0101] In this embodiment, the trained neural network model can be deployed locally, in the cloud, or on a mobile device, and converted into a format suitable for deployment. The model is then integrated into the target application or system to ensure that the model can be loaded and run smoothly. An interface suitable for practical applications is designed so that users or systems can call the model for reasoning through network requests. Service configuration is set, including concurrent processing, load balancing, error handling, etc., to ensure the stability and reliability of the service.

[0102] When monitoring, keep track of the performance of the model in the production environment, including key indicators such as response time, accuracy, and throughput. Use monitoring tools to build a monitoring system to implement real-time monitoring and alert mechanisms. You can also set up user feedback channels to collect and analyze user feedback on model prediction results, such as error reports and performance evaluations. Use automatic reporting tools to collect and classify user issues, and respond and handle them in a timely manner. Based on monitoring results and user feedback, regularly update model parameters or complete models to improve their performance or add new features. Use continuous integration / continuous deployment (CI / CD) pipelines to automatically update models to ensure that new models can be quickly deployed and replace old models.

[0103] See also Figure 2 , which is a schematic diagram of the initial distribution of components of Example 1 of the present application. Due to the symmetry of the boundary conditions, loads and geometric shape of the design domain of the structure, one quarter of the structure is selected for optimization design. This optimization problem is to make the structure volume meet a certain upper limit value to ensure the minimum structural flexibility. The upper left corner of the design domain is subjected to a unit downward concentrated force load, and the lower right corner point is simply supported. The design domain size of the MBB beam example is 3×0.5×1, and the upper limit of the volume fraction constraint is The design domain is uniformly discretized by 60×10×20 8-node 24-DOF solid elements.

[0104] See also Figure 3 , which is a schematic diagram of the optimal topological structure of Example 1 of the present application. It should be emphasized here that the boundary of the optimal topological structure is smooth. Figure 3 The structural boundary in b does not look smooth, which is because a low-resolution method is used to describe the corresponding topological description function, namely TDFS(xs). Figure 3 As shown in (c), if a higher image resolution is used, the optimal structure with smooth boundaries is obtained.

[0105] See also Figure 4 and Figure 5, respectively, are the initial design schematic diagram and the optimal topology structure schematic diagram under the encrypted grid of Example 1 of the present application. In order to verify the stability of the method proposed in the present application, a different finite element grid (i.e., 150×10×50) is used to solve the same MBB problem, and the upper limit value of the volume constraint remains unchanged.

[0106] See also Figure 6 , which is a schematic diagram of the application of two initial component distributions as comparative examples in Example 1 of the present application. In order to verify the effectiveness of the proposed method, two common initial component distributions are used as comparative examples using the three-dimensional MMC method. The first is an unfavorable distribution method with an insufficient number of components. There are 6 components arranged orthogonally at 45 degrees in the design domain horizontally to the xoz surface. The second is a better specific component arrangement method. There are 24 components distributed in the design domain. Every 4 components cross in space to form a unit cell-like structure. A total of 6 unit cells evenly fill the entire design domain. This better initial component distribution forms a complete force transmission path at the load point and the constraint point.

[0107] The first initial component arrangement has a small number of components, so the structure is formed in the initial iterations of the optimization process, which makes it impossible for the leftmost component to form a better structural form. The objective function value of the optimization result is 42.822. The second initial component arrangement has a large number of components and a more uniform distribution of components, so a better optimization result is obtained, and its objective function value is 37.033.

[0108] In summary, this application aims to solve the problems of traditional topology optimization methods that implicitly express many design variables and have difficulty in dealing with new geometries; by combining the homogenization method, structural optimization theory and finite element technology, an initial configuration library is constructed, and the MBB beam structure is optimized using the MMC optimization algorithm. Finally, high-quality unit cells are collected and performance prediction is performed using neural networks, so that the designed unit cell structure has explicit geometric information and smooth and clear structural boundaries, thereby improving the accuracy of computational simulation and R&D efficiency.

[0109] Example 2

[0110] See also Figure 7 , is a structural schematic diagram of an optimization system for MBB beams according to Example 2 of the present application; the specific contents include:

[0111] Establishing module: Establishing the initial configuration library based on random point distribution and random linking methods;

[0112] Optimization module: optimizing the initial configuration library based on the mobile variable component method, homogenization method, structural optimization theory and finite element analysis to obtain optimized unit cells and form a unit cell database;

[0113] Training module: training and predicting the component parameters and performance indicators in the unit cell database through a neural network model to obtain predicted values ​​of the performance indicators;

[0114] Deployment module: deploy the trained neural network model to actual applications for monitoring and maintenance.

[0115] Example 3

[0116] See also Figure 8 , is a schematic diagram of the device structure of Embodiment 3 of the present application. The device 50 includes a processor 51 and a memory 52 coupled to the processor 51 .

[0117] The memory 52 stores program instructions for implementing the above-mentioned optimization method for MBB beams.

[0118] The processor 51 is used to execute program instructions stored in the memory 52 to achieve an optimization for the MBB beam.

[0119] The processor 51 may also be referred to as a CPU (Central Processing Unit).

[0120] The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0121] Example 4

[0122] See also Fig. 9 , which is a schematic diagram of the structure of the storage medium of Example 4 of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods, wherein the program file 61 can be stored in the above storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or a computer, a server, a mobile phone, a tablet, and other devices.

[0123] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0124] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0125] Although the embodiments of the present application have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the appended claims and their equivalents.

[0126] Of course, the present invention may have many other implementations. Based on this implementation, other implementations obtained by ordinary technicians in this field without any creative work are all within the scope of protection of the present invention.

Claims

1. An optimization method for MBB beams, characterized in that: include: Establish an initial configuration library based on random point distribution and random linking methods; The initial configuration library is optimized based on a mobile variable component method, a homogenization method, a structural optimization theory and a finite element analysis to obtain an optimized unit cell and form a unit cell database; The component parameters and performance indicators in the unit cell database are trained and predicted by a neural network model to obtain predicted values ​​of the performance indicators; Deploy the trained neural network model to actual application for monitoring and maintenance.

2. The optimization method for MBB beams according to claim 1, characterized in that: The steps of establishing the initial configuration library according to the random point distribution and random linking method specifically include the following steps: Define basic parameters, including but not limited to: spatial range, number of points, and link generation rules; Use the rand function to randomly generate the coordinates of points in the specified space; Store a linked list of each point based on an empty cell array; Traverse each point, and if the distance is within the maximum link distance, randomly decide whether to establish a link; Use MATLAB's plotting capabilities to visualize the points and links.

3. The optimization method for MBB beams according to claim 1, characterized in that: The step of optimizing the initial configuration library based on the mobile variable component method, the homogenization method, the structural optimization theory and the finite element analysis to obtain the optimized unit cell and form the unit cell database specifically includes the following steps: Construct the topological description function corresponding to each component and obtain the material properties by adjusting the design variables of the component; Select structural flexibility as the objective function and volume upper limit as the constraint function for optimization; Three-dimensional finite element analysis and sensitivity analysis are performed using a substitute material model algorithm.

4. The optimization method for MBB beams according to claim 3, characterized in that: The expression of the topology description function is: Where x, y, z are global coordinate systems, x′, y′, z′ are local coordinate systems; p is a positive integer; L1 is the semi-axis length of the component in the x direction; and among them a =sine, s b =sinβ,s t =sinθ, 5. The optimization method for MBB beams according to claim 3, characterized in that: The expressions of the design variables of the components are: D=(x0,y0,z0,L1,s a ,s b ,s t ,h 1 ,h 2 ,h 3 ,t 1 ,t 2 ,t 3 ) T Where x0, y0, z0 are the center coordinates of the component; h and t are parameter vectors.

6. The optimization method for MBB beams according to claim 3, characterized in that: The steps of using the substitute material model algorithm to perform three-dimensional finite element analysis and sensitivity analysis specifically include the following steps: The stiffness matrix of the finite element numbered e is: Where Ω e is the area occupied by the finite element; B is the strain matrix; D0 is the elastic matrix of the solid material, ρ e For the region Ω e The volume ratio of the area occupied by the internal solid material to the entire unit area; Sensitivity analysis is performed and the variation of the flexibility objective functional pair is as follows: where δ is the Dirac delta function and δ i (x s )=min(δ(χ i ),δ(χ s )),i=1,…,n.

7. The optimization method for MBB beams according to claim 1, characterized in that: The step of training and predicting the component parameters and performance indicators in the unit cell database by using a neural network model to obtain the predicted value of the performance indicator specifically includes the following steps: Select a convolutional neural network, determine the number of network layers, the number of neurons in each layer, the activation function, set the number of input layer nodes to match the number of component parameters, and set the number of output layer nodes to match the number of predicted performance indicators; Split the dataset into training, validation, and test sets; Train the neural network based on the training set data and adjust the network weights and biases through the back-propagation algorithm; During the training process, a validation set is used to monitor the performance of the neural network model, and the network structure and hyperparameters are adjusted according to the performance of the training set; The trained neural network model is evaluated based on the test set, and the accuracy and mean square error are calculated; the new component parameters are input into the trained neural network model to obtain the corresponding performance indicator prediction values.

8. A system for an optimization method for MBB beams according to any one of claims 1 to 7, characterized in that: include: Establishing module: Establishing the initial configuration library based on random point distribution and random linking methods; Optimization module: optimizing the initial configuration library based on the mobile variable component method, homogenization method, structural optimization theory and finite element analysis to obtain optimized unit cells and form a unit cell database; Training module: training and predicting the component parameters and performance indicators in the unit cell database through a neural network model to obtain predicted values ​​of the performance indicators; Deployment module: deploy the trained neural network model to actual applications for monitoring and maintenance.

9. A device, characterized in that: The device includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing an optimization method for an MBB beam as described in any one of claims 1 to 7; and the processor is used to execute the program instructions stored in the memory to implement an optimization method for an MBB beam.

10. A storage medium, characterized in that: Program instructions executable by a processor are stored, and the program instructions are used to execute the optimization method for MBB beams according to any one of claims 1 to 7.

Images