Modular structure optimization design method and device

Abstract

The invention discloses a modular structure optimization design method and device, and relates to the technical field of modular structure modeling and modular structured algorithms. The method comprises the following steps: constructing a modular structure system based on a target structure, and decomposing the modular structure system into a plurality of isomorphic subsystems and residual systems; finite element analysis is carried out on the isomorphic subsystems, and a super-element matrix is constructed; the isomorphic subsystem converts physical coordinates into modal coordinates, and constructs a modular virtual framework based on a super-unit matrix; initializing design variables of the modularized virtual framework, and introducing a Bayesian optimization method to construct a Bayesian agent model; training the Bayesian agent model until a preset stop condition is met to obtain an optimization parameter; and updating the design variables according to the optimization parameters to obtain an optimized modular structure system. According to the method, efficient optimization design of a multi-dimensional modular structure is realized, and meanwhile, the accuracy of a calculation result can be ensured.

Description

Technical Field

[0001] The present application relates to the technical field of modular structure modeling and modular structured algorithm, and in particular to a modular structure optimization design method and device. Background Art

[0002] As the complexity of space systems continues to increase, modular structures have also had higher demands in reconfiguration design. In space systems, traditional modular designs include components such as antennas, solar arrays, and optical reflectors, which are usually composed of two-dimensional modules, while existing three-dimensional modules are mostly composed of cylindrical modules and cubic connection nodes. Modular structures, with their advantages such as high redundancy, robustness, and maintainability, have shown significant potential in meeting the performance and economic requirements of aircraft structures in complex space environments.

[0003] In the field of aerospace, the parametric reconfiguration modeling method of modular structures is used to quickly design and manufacture complex products such as aircraft and rockets. The existing optimization design of modular structures mainly relies on a single model dimensionality reduction technology or an efficient optimization algorithm, introducing parameterization at the finite element level, and optimizing the topological structure of the standard module and the joint configuration of the interconnected modules at the same time. Although traditional methods can be used for structural optimization, they have certain limitations in optimization effects and application scenarios, bringing challenges to the accurate evaluation of system dynamic parameters, as well as the problem of a sharp increase in computational costs due to the increase in the number of modules in the optimization design of modular reconfigurable structures. How to maximize its efficiency and robustness in the optimization design of high-dimensional modular structures is still under exploration. In addition, traditional optimization methods have a single means, a long cycle, and it is difficult to guarantee the accuracy of the calculation results. Summary of the invention

[0004] The embodiments of the present application provide a modular structure optimization design method and device, which solves the problem that the modular structure of the existing modular structure optimization design method is complex, and the computational cost increases dramatically due to the increase in the number of modules. Through the component modal synthesis method, a virtual framework at the modular structure matrix level is established, and the parametric rapid reorganization of the multi-dimensional modular structure is realized. The Gaussian process and / or random forest algorithm is used as a proxy model for Bayesian optimization, and a material-free calculation model for empty positions and a hyperparameter dimensionality reduction method are introduced, which improves the computational efficiency and adaptability of the system in solving high-dimensional problems.

[0005] In a first aspect, an embodiment of the present application provides a modular structure optimization design method, comprising: constructing a modular structure system based on a target structure, and decomposing it into multiple isomorphic subsystems and a remaining system; performing finite element analysis on the isomorphic subsystems and constructing a super-unit matrix; converting the isomorphic subsystems from physical coordinates to modal coordinates, and constructing a modular virtual framework based on the super-unit matrix; initializing the design variables of the modular virtual framework, and introducing a Bayesian optimization method to construct a Bayesian proxy model; training the Bayesian proxy model until a preset stop condition is met to obtain optimization parameters; updating the design variables according to the optimization parameters to obtain an optimized modular structure system.

[0006] In a possible implementation, the finite element analysis of the isomorphic subsystems and the construction of the super-unit matrix include: constructing a finite element model of the isomorphic subsystem; performing finite element analysis on the finite element model of the isomorphic subsystem and extracting a local characteristic matrix of the finite element model of the isomorphic subsystem; wherein the local characteristic matrix includes a stiffness matrix and a mass matrix; setting the degrees of freedom of the nodes in the finite element model of the isomorphic subsystem, and reducing the dimension of the local characteristic matrix based on the degrees of freedom to obtain a reduced local matrix; and assembling the reduced local matrices of the finite element models of each of the isomorphic subsystems to obtain the super-unit matrix.

[0007] In a possible implementation, constructing a modular virtual framework based on the super-element matrix includes: introducing interface displacement coordination and force balance conditions in the isomorphic subsystems under physical coordinates; determining the finite element model function and boundary of each of the isomorphic subsystems, and defining them as independent modules; establishing connections between the independent modules to construct the modular virtual framework.

[0008] In a possible implementation, the design variables include generalized density variables, module rotation variables, and module type variables.

[0009] In a possible implementation, after the design variables, it also includes: constructing a comprehensive generalized density design variable; wherein the comprehensive generalized density design variable includes the generalized density variable and the module type variable of the selected position; and using the comprehensive generalized density design variable to reduce the dimension of the design variable.

[0010] In a possible implementation, the using the comprehensive generalized density design variable to reduce the dimension of the design variable includes: , , ; In the formula, represents the search for design variables after dimensionality reduction, represents the comprehensive generalized density design variable, represents the module rotation variable, represents the transposed matrix, Represents a module type variable, represents the objective function, represents the number of positions to be selected, represents the total number of randomly placed candidate modules, represents the number constraint of the modules to be selected, express Scale constraints in the axis direction, Indicates The number of modules to be selected on the layer is constrained. and denote the upper and lower constraints of the comprehensive generalized density design variable, respectively. Indicates comprehensive generalized density design variables, Indicates A rotating variable, and They respectively represent the upper and lower limit constraints of the module rotation angle.

[0011] In a possible implementation, the Bayesian proxy model includes: a Bayesian proxy model based on Gaussian process and a Bayesian proxy model based on random forest.

[0012] In the second aspect, an embodiment of the present application provides a device for modular structural optimization design, including: a decomposition module, used to construct a modular structural system based on a target structure, and decompose it into multiple isomorphic subsystems and a remaining system; a reorganization module, used to perform finite element analysis on the isomorphic subsystems and construct a super-unit matrix; convert the isomorphic subsystems from physical coordinates to modal coordinates, and construct a modular virtual framework based on the super-unit matrix; a structural optimization module, used to initialize the design variables of the modular virtual framework, introduce a Bayesian optimization method to construct a Bayesian proxy model; train the Bayesian proxy model until a preset stop condition is met to obtain optimization parameters; a reconstruction module, used to update the design variables according to the optimization parameters to obtain an optimized modular structural system.

[0013] In the third aspect, an embodiment of the present application provides a device for executing a modular structure optimization design method, the device comprising: a processor; a memory for storing processor executable instructions; when the processor executes the executable instructions, it implements the method described in the first aspect or any possible implementation method of the first aspect.

[0014] In a fourth aspect, an embodiment of the present application provides a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium includes a medium for storing a computer program or instructions, which, when executed, enables the method described in the first aspect or any possible implementation method of the first aspect to be implemented.

[0015] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: The embodiment of the present application adopts a modular structure optimization design method and device, and constructs a modular virtual framework to characterize its modular structure by performing finite element analysis on the modular structure system, thereby completing multi-dimensional rapid module reorganization. Based on the Bayesian optimization method, the design variables are initialized, and the hyperparameter space dimension reduction is performed to reduce the difficulty of building the Bayesian proxy model. Through the hyperparameter dimension reduction method, the Bayesian proxy model is trained, so that the optimized modular aerospace structure system after the reconstruction design can complete the modular design of large structures more efficiently and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flowchart of a modular structure optimization design method provided in an embodiment of the present application; Figure 2 A schematic diagram of a two-dimensional modular virtual framework reconstruction provided in an embodiment of the present application; Figure 3 A schematic diagram of an optimization design space of a Bayesian proxy model based on a Gaussian process provided in an embodiment of the present application; Figure 4 A schematic diagram of the initial layout of the Bayesian proxy model based on Gaussian process provided in an embodiment of the present application; Figure 5 A schematic diagram of an optimized layout of a Bayesian proxy model based on a Gaussian process provided in an embodiment of the present application; Figure 6 A schematic diagram of a three-dimensional linear optimization design space of a Bayesian surrogate model based on random forests provided in an embodiment of the present application; Figure 7 A schematic diagram of a three-dimensional linear initial layout of a Bayesian proxy model based on random forests provided in an embodiment of the present application; Figure 8A schematic diagram of a three-dimensional linear optimization layout of a Bayesian proxy model based on random forests provided in an embodiment of the present application; Fig. 9 A schematic diagram of a three-dimensional nonlinear optimization design space of a Bayesian surrogate model based on random forests provided in an embodiment of the present application; Fig.10 A schematic diagram of a three-dimensional nonlinear initial layout of a Bayesian proxy model based on random forests provided in an embodiment of the present application; Fig.11 A schematic diagram of a three-dimensional nonlinear optimization layout of a Bayesian proxy model based on random forests provided in an embodiment of the present application; Fig.12 A schematic diagram of the optimization results of the Bayesian proxy model based on Gaussian process provided in the embodiment of the present application; Fig.13 A schematic diagram of a three-dimensional linear optimization result of a Bayesian proxy model based on random forest provided in an embodiment of the present application; Fig.14 A schematic diagram of a three-dimensional nonlinear optimization result of a Bayesian proxy model based on random forest provided in an embodiment of the present application; Fig.15 A schematic diagram of the structure of a modular structure optimization design device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] The following describes some of the techniques involved in the embodiments of the present application to facilitate understanding, and they should be considered as merely exemplary. Therefore, it should be appreciated by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, some descriptions of well-known functions and structures are omitted in the following description.

[0020] Figure 1 It is a flowchart of a modular structure optimization design method provided in an embodiment of the present application, including steps 101 to 106. Figure 1 This is only an execution order shown in the embodiment of the present application, and does not represent the only execution order of the modular structure optimization design method. In the case where the final result can be achieved, Figure 1 The steps shown may be performed in parallel or in reverse, as follows.

[0021] Step 101: construct a modular structural system based on the target structure and decompose it into multiple isomorphic subsystems and residual systems. In the embodiment of the present application, the target structure involved is a large modular aerospace structure, and the density of all materials is defined as , Young's modulus is defined as 71.7 GPa, and Poisson's ratio is defined as 0.33. A modular aerospace structure system is constructed by combining modular design and parametric modeling methods. The modular aerospace structure system is decomposed into multiple modules with specific functions using finite element analysis methods, and these modules are quickly designed and reorganized using parametric modeling technology, thereby achieving rapid customization and optimization design of the target aerospace structure, as follows.

[0022] ; in, Represents a modular aerospace structure system. represents a homogeneous subsystem, represents the remaining system, Represents the total number of homogeneous subsystems and the remaining systems.

[0023] Specifically, the modular aerospace structure system is defined as , according to the system The characteristics of the system are decomposed The internal components of each single module are highly related and can fully execute specific functions or logic. Since the coupling between modules is low, the independence of modules and the system can be improved. flexibility.

[0024] Step 102: Perform finite element analysis on the homogeneous subsystems and construct a super-element matrix. In the embodiment of the present application, a finite element model of the homogeneous subsystem is constructed based on the component modal synthesis method (CMS). By performing finite element analysis on the finite element model, feature extraction is performed in the homogeneous subsystem to obtain a local feature matrix of each finite element model, and then the entire system is analyzed. ; Among them, the local characteristic matrix includes the stiffness matrix and the mass matrix.

[0025] Specifically, finite element analysis can be processed using finite element software such as Abaqus, Nastran and Ansys Mechanical by extracting large sparse matrices to reduce data storage space. These software provide a rich unit library and material model library to simulate various complex engineering problems. At the same time, they also provide powerful pre- and post-processing functions to facilitate users to conduct subsequent model building, result analysis and optimization design.

[0026] System-based The finite element model of the homogeneous subsystem is constructed by using a finite number of polyhedrons to simulate the actual modular structure and reflect the approximate situation of the real physical system. After determining the super-element and boundary nodes, perform modal coordinate transformation on the boundary nodes of the homogeneous subsystem, perform structural degree of freedom condensation on the local characteristic matrix to obtain a reduced local matrix, assemble the reduced local matrices of the finite element models of each homogeneous subsystem, and obtain the super-element matrix. This process mainly condenses the degrees of freedom of the internal component nodes to the boundary nodes, thereby reducing the size of the local matrix. While reducing the amount of calculation, it can also maintain sufficient accuracy and improve calculation efficiency. After constructing the super-element matrix, it can be solved using finite element solvers such as Optistruct, FreeFEM and Elmer.

[0027] According to the super-element matrix, the undamped discretized branch main modes of the homogeneous subsystem are obtained to The system is analyzed as follows.

[0028] ; In the formula, represents the stiffness matrix, represents the eigenvalues ​​of the dynamical equations of the isomorphic subsystem, represents the mass matrix, represents the main mode of the undamped discretized branch.

[0029] Among them, the main mode of the undamped discretized branch is the system Inherent vibration characteristics, independent of external excitation. Modal analysis can fully describe the system The overall dynamic characteristics of the system. The space formed by each order modal vector is defined as the main space, and its corresponding modal coordinates are the main coordinates. In modal analysis, the continuous system Discretization into isomorphic subsystems and residual systems with finite degrees of freedom retains the system In order to perform numerical calculations, each degree of freedom in each discretized system corresponds to a mode, which is arranged from low to high in frequency, and each order of main modes has orthogonality in its N-dimensional main space, that is, the vibrations between different modes are independent of each other, ensuring the system Low coupling.

[0030] Step 103: Convert the isomorphic subsystem from physical coordinates to modal coordinates, and construct a modular virtual framework based on the super-element matrix. Exemplarily, constructing the modular virtual framework includes: converting the physical coordinates of the boundary nodes of the isomorphic subsystem into modal coordinates, which is a modal coordinate transformation, as follows.

[0031] ; In the formula, represents the physical coordinates of the homogeneous subsystem, represents the main mode of the undamped discretized branch, Represents the modal coordinates of the homogeneous subsystem.

[0032] According to the physical coordinate design interface displacement coordination and force balance conditions, the finite element model functions and boundaries of each isomorphic subsystem are determined, and they are defined as independent modules to obtain the dynamic equations in the global coordinate system, as shown below.

[0033] , ; In the formula, and Representing homogeneous subsystems and Interface degrees of freedom for modal coordinate transformation, and Representing homogeneous subsystems and Interface forces for modal coordinate transformation.

[0034] Exemplarily, the dynamic equations of isomorphic subsystem 1 in the global coordinate system of isomorphic subsystem 2 are as follows.

[0035] ; In the formula, represents the mass matrix of homogeneous subsystem 1, represents the stiffness matrix of homogeneous subsystem 1, represents the modal coordinates, represents the displacement of isomorphic subsystem 2.

[0036] The stiffness matrix and mass matrix composed of modal coordinates constitute a modular virtual framework. Specifically, according to the super-unit node relationship of the isomorphic subsystem, the independent modules of the isomorphic subsystem are connected through matrix transformation to obtain the isomorphic subsystem matrix. The system is realized by matrix transformation. The secondary layout of the reorganized modular aerospace structure system is obtained, that is, the new system According to the assembly of the isomorphic subsystem matrices, the reconstructed dynamic equations of the isomorphic subsystems are expressed by independent modal coordinates as follows.

[0037] ; in, represents the mass matrix, represents the stiffness matrix, represents the reconstructed modal coordinates, represents independent modal coordinates.

[0038] When the homogeneous subsystem performs translational motion, its stiffness and mass matrix do not change. In order to improve the new system Design level flexibility, introducing coordinate transformation matrix It is rotated, and the stiffness and mass matrices after rotation are processed as follows.

[0039] , ; In the formula, With expression The stiffness matrix and mass matrix after rotation respectively, represents the coordinate transformation matrix, and Representation System The original stiffness matrix and mass matrix of Indicates the rotation angle.

[0040] Among them, Figure 2 As shown, it is a schematic diagram of the reconstruction of a two-dimensional modular virtual framework. Indicates The mass matrix of the modal coordinates, Indicates The stiffness matrix of each modal coordinate is: At the same time, compared with the prior art, the present application can extend the two-dimensional space to the three-dimensional linear and nonlinear design space through the super-element method of component modal synthesis, and realize the rapid reassembly of the new system within the virtual framework, which has a wider scope of application and a more flexible implementation method.

[0041] Step 104: Initialize the design variables of the modular virtual framework, and introduce the Bayesian optimization method to construct the Bayesian proxy model. Specifically, initialize the design variables, wherein the design variables include generalized density design variables, module rotation variables, and module type variables. In the modular virtual framework, the design variables are as follows.

[0042] ; In the formula, represents the generalized density design variable, represents the rotation variable, Represents a module type variable, represents the transposed matrix, represents the number of positions to be selected, Represents the total number of randomly placed candidate modules.

[0043] The expected recombinant configuration is generated by controlling the generalized density design variables at the candidate positions, the rotation variables of the modules at the candidate positions and the module type variables.

[0044] By introducing the Bayesian optimization method, the optimization target, design constraints and number of iterations of Bayesian optimization are set. In an embodiment of the present application, the design variables can be used as a set of hyperparameter combinations, and the final transposed matrix of the optimization result is a hyperparameter combination, i.e., a design variable combination. In high-dimensional space, data sparsity increases, sampling search is very time-consuming and difficult, and requires more computing resources. With the increase of the complexity of the problem, the modeling of the proxy model of the hyperparameter space, i.e., the design variable space, also becomes more difficult, greatly increasing the uncertainty of the entire system. To avoid the disaster of size, according to the characteristics of modular structure layout design, the design variable space is reduced in dimension. A comprehensive design variable is proposed, which includes a generalized density variable and a module type variable of a position to be selected. To avoid the computational cost brought by high-dimensional space, the embodiment of the present application reduces the dimension of the design variable space to obtain a design variable combination, as follows.

[0045] , , ; In the formula, represents the search for design variables after dimensionality reduction, represents the comprehensive generalized density design variable, represents the module rotation variable, represents the transposed matrix, Represents a module type variable, represents the objective function, represents the number of positions to be selected, represents the total number of randomly placed candidate modules, represents the number constraint of the modules to be selected, express Scale constraints in the axis direction, Indicates The number of modules to be selected on the layer is constrained. and denote the upper and lower constraints of the comprehensive generalized density design variable, respectively. Indicates comprehensive generalized density design variables, Indicates A rotating variable, and They respectively represent the upper and lower limit constraints of the module rotation angle.

[0046] Among them, the design variables describing the selection results of the candidate locations and the generalized density design variables describing the module selection locations are Updated to the comprehensive generalized density design variable Through dimensionality reduction, the dimension of the design variable space is reduced from the original Dimensionality dimensional, effectively reducing the latitude of the design variable space.

[0047] The optimization objective function of the Bayesian optimization method is the system The problem of maximizing the natural frequency is converted into a minimization problem by taking its inverse. In the embodiment of the present application, the constructed Bayesian proxy model is a Bayesian proxy model based on Gaussian process and a Bayesian proxy model based on random forest.

[0048] Step 105: Train the Bayesian proxy model until the preset stop condition is met to obtain the optimized parameters. For example, the Gaussian process is selected as the Gaussian proxy model for Bayesian optimization, and the optimization space design of the two-dimensional modular structure is performed. The maximum number of iterations is set to 1000 times, and the module size is set to 50mm×50mm. The design space is as follows: Figure 3 As shown, the initial layout is as Figure 4 As shown in the figure, the global optimal solution is obtained within the maximum number of iterations, and the optimized layout is as follows Figure 5 As shown, the maximum degree of freedom of the module is 4. Define a prediction that follows a normal distribution as follows.

[0049] ;in, ; ; , ; In the formula, represents the prediction of the normal distribution of the Bayesian surrogate model based on Gaussian processes, represents the target value, Represents a sample, represents a sampled data set, represents the total number of samples, represents a normal distribution, represents the mean function, represents the variance function, represents the sample point set, Indicates samples, Indicates scalar output.

[0050] For example, the random forest algorithm of probability regression is introduced as the random forest proxy model of Bayesian optimization to optimize the spatial design of the three-dimensional modular structure. The maximum number of iterations of the three-dimensional linear modular structure is 200 times, and the module size is set to 50mm×50mm×100mm. The design space is as follows Figure 6 As shown, the initial layout is as Figure 7 As shown, the optimized layout is as follows Figure 8 As shown in the figure, the maximum degree of freedom of the module is 8. The maximum number of iterations of the three-dimensional nonlinear modular structure is 100 times, and the diameter of the module outer circle is set to 50 mm. The design space is as follows: Fig. 9 As shown, the initial layout is as Fig.10 As shown, the optimized layout is as follows Fig.11 As shown in the figure, the maximum degree of freedom of the module is 16. In the initial layout, there are 4 constrained modules, 3 modules of each type, and the positions are randomly distributed in the design space. The initial natural frequency of the three-dimensional nonlinear modular structure is 64.47Hz, and the natural frequency after optimization is 73.05Hz, which is 13.31% higher than the initial frequency. Because the center of gravity, mass and fundamental frequency of the four modules are different. Although the difference is small. However, the specific distribution trend of the optimized layout meets the general rule that the section moment of inertia increases and the overall center of gravity of the system tends to the constraint end, which improves the dynamic characteristics of the structure.

[0051] A Gaussian distribution is preset to construct a random forest prediction distribution. For example, ,in, and You can select the regressor value in the random forest regression tree collection.

[0052] Specifically, the Bayesian proxy model based on Gaussian process and the Bayesian proxy model based on random forest start from the dimension of the problem, perform optimization design, jump out of the local optimal solution within the least number of iterations, and construct a modular optimization design method.

[0053] Furthermore, the Bayesian proxy model is optimized using the empty position material-free calculation model and the hyperparameter dimensionality reduction method. A scaling factor is introduced into the hyperparameter dimensionality reduction method, and the superelement matrix is ​​scaled to obtain the optimized superelement matrix.

[0054] Step 106: Update the design variables according to the optimization parameters to obtain an optimized modular structure system.

[0055] In the embodiment of the present application, the empty position no material calculation model is an empty matrix that does not consider the empty position in the numerical calculation. At the same time, in order to avoid the numerical singularity problem caused by the empty matrix, a scaling factor is introduced to scale the super unit matrix and update the optimization design variables of the empty position without material. Specifically, the super unit matrix is ​​scaled as follows.

[0056] , ; In the formula, and represents the scaled optimized superelement matrix, and Indicates new system The initial superelement matrix, Indicates the scaling factor, which is 0.1. Represents a constant, set to 3.

[0057] Among them, by scaling the super unit matrix, the design variables of the material-free position are made close to 0. As the dimension of the design space increases, the embodiment of the present application adopts a calculation model of empty position without material, does not consider the empty matrix of the empty position in the numerical calculation of the structural analysis, updates the optimized design variables, and can reduce the new system The error is reduced, which significantly improves the calculation efficiency and completes the new system reconstruction.

[0058] For example, the optimization result of the Bayesian surrogate model based on Gaussian process is as follows: Fig.12 As shown, the new system after optimization The layout is shortened along the length direction, which conforms to the general rule of improving the dynamic characteristics of the structure. The linear optimization results of the Bayesian surrogate model based on random forest are shown in Fig.13 As shown in the figure, the high-dimensional design variable space is effectively reduced in dimension, making the optimization process more stable and the modeling process faster. The increase in the complexity of module connections also increases the flexibility of structural reorganization, providing a larger design space for exploring the optimal structural performance. The nonlinear optimization results of the Bayesian surrogate model based on random forest are shown in the figure. Fig.14 As shown, the highly complex and nonlinear design space is optimized more efficiently, presenting a structural layout form of high-freedom modules. In the embodiment of the present application, the modular structure optimization design under the complex stacking from two-dimensional to three-dimensional space is completed, and the optimized modular aerospace structure system is obtained, which gives the optimal hyperparameter combination of the modular structure layout more quickly and accurately, that is, the optimal design variables.

[0059] Although the present application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in this embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the actual device or client product is executed, it can be executed in sequence or in parallel according to the method shown in this embodiment or the accompanying drawings (for example, in a parallel processor or multi-threaded processing environment).

[0060] like Fig.15 As shown, the embodiment of the present application also provides a modular structure optimization design device 1500. The device includes: a decomposition module 1501, a reorganization module 1502, a structure optimization module 1503 and a reconstruction module 1504, as follows.

[0061] The decomposition module 1501 is used to construct a modular structure system based on the target structure and decompose it into multiple isomorphic subsystems and residual systems.

[0062] The reorganization module 1502 is used to perform finite element analysis on the homogeneous subsystem and construct a super-element matrix; transform the homogeneous subsystem from physical coordinates to modal coordinates, and construct a modular virtual framework based on the super-element matrix.

[0063] The structural optimization module 1503 is used to initialize the design variables of the modular virtual framework, introduce the Bayesian optimization method to construct the Bayesian proxy model; train the Bayesian proxy model until the preset stop condition is met to obtain the optimized parameters.

[0064] The reconstruction module 1504 is used to update the design variables according to the optimization parameters to obtain an optimized modular structure system.

[0065] The devices or modules described in the above application embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above devices are described in various modules according to their functions. When implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0066] The methods, devices or modules described in this application can be implemented in the form of computer-readable program codes. The controller can be implemented in any appropriate manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program codes (such as software or firmware) that can be executed by the (micro)processor, logic gates, switches, application-specific integrated circuits (English: Application Specific Integrated Circuit; Abbreviation: ASIC), programmable logic controllers and embedded microcontrollers. Examples of controllers include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program codes, the controller can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included in it for implementing various functions can also be regarded as structures within the hardware component. Or even, the means for realizing various functions may be regarded as both a software module for realizing the method and a structure within a hardware component.

[0067] The embodiment of the present application also provides a device for a modular structure optimization design method, the device comprising: a processor; a memory for storing processor executable instructions; when the processor executes the executable instructions, the method described in the embodiment of the present application is implemented.

[0068] The embodiments of the present application also provide a non-volatile computer-readable storage medium on which a computer program or instruction is stored. When the computer program or instruction is executed, the method described in the embodiments of the present application is implemented.

[0069] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist independently, or two or more modules may be integrated into one module.

[0070] It can be known from the description of the above implementation mode that those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary hardware. Based on such an understanding, the technical solution of the present application can be essentially or partly reflected in the prior art in the form of a software product, or it can be reflected in the implementation process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a mobile terminal, a server, or a network device, etc.) to execute the methods of each embodiment of the present application or some parts of the embodiments.

[0071] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. All or part of this application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0072] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present application.

Claims

1. A modular structure optimization design method, characterized in that: include: Construct a modular structure system based on the target structure and decompose it into multiple isomorphic subsystems and residual systems; Performing finite element analysis on the isomorphic subsystem and constructing a super-element matrix; Converting the isomorphic subsystem from physical coordinates to modal coordinates, and constructing a modular virtual framework based on the super-element matrix; Initializing the design variables of the modular virtual framework, and introducing the Bayesian optimization method to construct a Bayesian proxy model; Training the Bayesian proxy model until a preset stop condition is met to obtain optimized parameters; The design variables are updated according to the optimization parameters to obtain an optimized modular structural system.

2. The method according to claim 1, characterized in that The step of performing finite element analysis on the isomorphic subsystem to construct a super-element matrix includes: constructing a finite element model of the isomorphic subsystem; Performing finite element analysis on the finite element model of the isomorphic subsystem to extract a local characteristic matrix of the finite element model of the isomorphic subsystem; wherein the local characteristic matrix includes a stiffness matrix and a mass matrix; Setting the degrees of freedom of nodes in the finite element model of the isomorphic subsystem, and reducing the dimension of the local characteristic matrix based on the degrees of freedom to obtain a reduced local matrix; The reduced local matrices of the finite element models of the homogeneous subsystems are assembled to obtain the super-element matrix.

3. The method according to claim 1, characterized in that The step of constructing a modular virtual framework based on the super unit matrix comprises: Introducing interface displacement coordination and force balance conditions in the isomorphic subsystems under physical coordinates; Determine the finite element model functions and boundaries of each of the isomorphic subsystems and define them as independent modules; Connections between the independent modules are established to construct the modular virtual framework.

4. The method according to claim 1, characterized in that The design variables include generalized density variables, module rotation variables and module type variables.

5. The method according to claim 4, characterized in that After the design variables, it also includes: Constructing a comprehensive generalized density design variable; wherein the comprehensive generalized density design variable includes the generalized density variable and the module type variable of the to-be-selected position; The comprehensive generalized density design variables are used to reduce the dimension of the design variables.

6. The method according to claim 5, characterized in that The step of reducing the dimension of the design variables by using the comprehensive generalized density design variables comprises: ， ， ； In the formula, represents the search for design variables after dimensionality reduction, represents the comprehensive generalized density design variable, represents the module rotation variable, represents the transposed matrix, Represents a module type variable, represents the objective function, represents the number of positions to be selected, represents the total number of randomly placed candidate modules, represents the number constraint of the modules to be selected, express Scale constraints in the axis direction, Indicates The number of modules to be selected on the layer is constrained. and denote the upper and lower constraints of the comprehensive generalized density design variable, respectively. Indicates comprehensive generalized density design variables, Indicates A rotating variable, and They respectively represent the upper and lower limit constraints of the module rotation angle.

7. The method according to claim 1, characterized in that The Bayesian proxy model includes: Bayesian surrogate model based on Gaussian process and Bayesian surrogate model based on random forest.

8. A modular structure optimization design device, characterized in that: include: Decomposition module, which is used to build a modular structure system based on the target structure and decompose it into multiple isomorphic subsystems and the remaining system; A reorganization module is used to perform finite element analysis on the isomorphic subsystem and construct a super-element matrix; Converting the isomorphic subsystem from physical coordinates to modal coordinates, and constructing a modular virtual framework based on the super-element matrix; A structural optimization module, used to initialize the design variables of the modular virtual framework and introduce a Bayesian optimization method to construct a Bayesian proxy model; Training the Bayesian proxy model until a preset stop condition is met to obtain optimized parameters; The reconstruction module is used to update the design variables according to the optimization parameters to obtain an optimized modular structure system.

9. A device for performing a modular structure optimization design method, characterized in that: include: processor; a memory for storing processor-executable instructions; When the processor executes the executable instructions, the method according to any one of claims 1 to 7 is implemented.

10. A non-volatile computer-readable storage medium, characterized in that: The device comprises a computer program or an instruction for storing the computer program or the instruction, which, when executed, enables the method according to any one of claims 1 to 7 to be implemented.

Images