Interlayer superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm

Abstract

The invention discloses an interlayer superstructure electromagnetic-mechanical collaborative design method based on machine learning and a heuristic optimization algorithm, relates to the technical field of material design, and is used for solving the technical problems that the mechanical property of an interlayer superstructure obtained through existing design is poor, and the requirements for strength, stability and durability in practical application are difficult to meet. The collaborative design method comprises the following steps: setting an objective function of an interlayer superstructure, and determining the configuration of the interlayer superstructure to be designed according to the objective function; establishing a proxy model according to a machine learning algorithm, and integrating the proxy model into a heuristic optimization algorithm; generating geometric characteristic variables of the interlayer superstructure in batches; obtaining the mechanical property and the electromagnetic property of the interlayer superstructure; feeding back the mechanical property and the electromagnetic property to a heuristic optimization algorithm to obtain a numerical value of an objective function; and according to the numerical value of the target function, carrying out iterative updating on the geometric feature variables until an optimal solution is obtained, and completing the design.

Description

Technical Field

[0001] The present invention relates to the technical field of material design, and more particularly, to a collaborative electromagnetic-mechanical design method for sandwich superstructures based on machine learning and heuristic optimization algorithms. Background Art

[0002] With the development of electronic information technology and radar detection technology, electromagnetic technologies and devices are widely used in civilian and military fields such as medical, electronics, and radar. While achieving specific technical functions, many problems and challenges such as electromagnetic pollution, electromagnetic interference, and electromagnetic stealth have emerged around electromagnetic technologies.

[0003] The sandwich superstructure with both broadband wave absorption and high load-bearing integration is a key structure for reducing the radar cross-section of weaponry, and for reducing electromagnetic pollution and electromagnetic interference, and has important engineering application value. Most of the existing designs of sandwich superstructures focus on the electromagnetic performance design of the wave-absorbing layer, but lack systematic analysis and design of their mechanical properties, failing to fully exploit the comprehensive electromagnetic-mechanical advantages of the sandwich superstructure. The designed sandwich superstructures have poor mechanical properties and are difficult to meet the requirements for strength, stability, and durability in practical applications. Summary of the Invention

[0004] The purpose of the present invention is to provide a collaborative electromagnetic-mechanical design method for sandwich superstructures based on machine learning and heuristic optimization algorithms, which is used to solve the technical problems that most of the existing designs of sandwich superstructures focus on the electromagnetic performance design of the wave-absorbing layer, but lack systematic analysis and design of their mechanical properties, failing to fully exploit the comprehensive electromagnetic-mechanical advantages of the sandwich superstructure, resulting in poor mechanical properties of the designed sandwich superstructures and being difficult to meet the requirements for strength, stability, and durability in practical applications. In view of this, the present invention provides the following technical solutions.

[0005] The present invention provides a collaborative electromagnetic-mechanical design method for sandwich superstructures based on machine learning and heuristic optimization algorithms, including: Setting the objective function of the sandwich superstructure, and determining the configuration of the sandwich superstructure to be designed according to the objective function; Establishing a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model according to machine learning algorithms, and integrating the surrogate models into the heuristic optimization algorithm; Batch generating the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; Obtaining the mechanical properties and electromagnetic properties of the sandwich superstructure by sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model; Feeding back the mechanical properties and electromagnetic properties into the heuristic optimization algorithm to obtain the numerical value of the objective function; According to the value of the objective function, the geometric feature variables are iteratively updated by the heuristic optimization algorithm until the optimal solution corresponding to the objective function is obtained, and the design of the sandwich superstructure is completed.

[0006] Compared with the prior art, the electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm of the present invention is used to design a sandwich superstructure with excellent mechanical and electromagnetic properties. In this electromagnetic-mechanical collaborative design method, the configuration of the sandwich superstructure to be designed is first determined, and the configuration of the sandwich superstructure further determines the geometric feature variables of the sandwich superstructure. In order to make the designed sandwich superstructure have excellent mechanical and electromagnetic properties at the same time, this technical solution further establishes a geometric feature-mechanical property surrogate model (hereinafter referred to as the first model) and a geometric feature-electromagnetic property surrogate model (hereinafter referred to as the second model) according to the machine learning algorithm, and integrates the surrogate models into the heuristic optimization algorithm, and then batch generates the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; after inputting the geometric feature variables into the above two surrogate models respectively, the mechanical properties and electromagnetic properties corresponding to the geometric feature variables can be obtained. That is to say, the geometric feature variables correspond to the mechanical properties in the first model and the electromagnetic properties in the second model; the obtained mechanical properties and electromagnetic properties are fed back into the heuristic optimization algorithm. After obtaining the value of the objective function, according to the value of the objective function, the geometric feature variables are iteratively updated by the heuristic optimization algorithm until the optimal solution corresponding to the objective function is obtained. The optimal solution corresponds to the optimal geometric feature in the geometric feature variables, and then the geometric feature of the sandwich superstructure is determined. So far, the design of the sandwich superstructure is completed. It can be seen that the above technical solution of the present invention considers both the mechanical properties and the electromagnetic properties of the sandwich superstructure during the design process of the sandwich superstructure, so that the designed sandwich superstructure has excellent mechanical properties and electromagnetic properties at the same time. Through the above technical solution of the present invention, the problems in the prior art that the design of the existing sandwich superstructure mostly focuses on the electromagnetic performance design of the absorbing layer, lacks systematic analysis and design of its mechanical properties, fails to give full play to the electromagnetic-mechanical comprehensive advantages of the sandwich superstructure, resulting in poor mechanical properties of the designed sandwich superstructure and being difficult to meet the requirements of strength, stability and durability in practical applications are solved.

[0007] Furthermore, in the electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm of the present invention, the configuration of the sandwich superstructure includes a protective layer, an absorbing layer and a bearing layer.

[0008] Furthermore, in the electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm of the present invention, the heuristic optimization algorithm includes an immune genetic algorithm and a multi-population genetic algorithm; And / or, the machine learning algorithm includes a BP neural network and a perceptron neural network.

[0009] Further, in the sandwich superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm of the present invention, the input of the geometric feature-mechanical property surrogate model is the geometric feature variable, and the output includes bending strength, compressive strength or tensile strength according to the design goal; And / or, the input of the geometric feature-electromagnetic property surrogate model is the geometric feature variable, and the output includes wave absorption bandwidth and minimum reflection loss value according to the design goal.

[0010] Further, in the sandwich superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm of the present invention, the geometric features of the sandwich superstructure include the thickness of the protective layer, the thickness of the wave absorption layer, the thickness of the bearing layer, and the geometric configuration of the wave absorption layer.

[0011] Further, in the sandwich superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm of the present invention, the objective function is determined according to the design goal, and the design goal includes thin thickness, broadband electromagnetic wave absorption performance and high load-bearing performance; wherein: The broadband electromagnetic wave absorption performance includes wave absorption bandwidth and microwave absorption rate, and the high load-bearing performance includes bending strength, tensile strength and compressive strength.

[0012] Further, in the sandwich superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm of the present invention, the configuration of the sandwich superstructure includes a wave-transparent protective layer, a broadband wave absorption layer and a shielding bearing layer; And / or, the material of the wave-transparent protective layer includes a flat wave-transparent structural material, the material of the broadband wave absorption layer includes a periodic array structure metamaterial, and the material of the shielding bearing layer includes a flat high-mechanical-strength material.

[0013] Further, in the sandwich superstructure electromagnetic-mechanical collaborative design method based on machine learning and heuristic optimization algorithm of the present invention, the flat wave-transparent structural material includes glass fiber laminate, quartz fiber laminate and aramid fiber laminate, the periodic unit of the periodic array structure metamaterial includes a honeycomb structure, a cylindrical protrusion metamaterial structure and a groove metamaterial structure, and the flat high-mechanical-strength material includes a carbon fiber laminate; And / or, the periodic array structure metamaterial is made of an absorbing composite material, the absorbing composite material is a composite material with electromagnetic loss particles, and the electromagnetic loss particles include carbon nanotubes, graphene, carbonyl iron and MXene.

[0014] Further, in the collaborative electromagnetic-mechanical design method of the sandwich superstructure based on machine learning and heuristic optimization algorithms of the present invention, during the process of establishing the geometric feature-mechanical property surrogate model according to the machine learning algorithm, the input of the machine learning algorithm is the geometric features of the sandwich superstructure, and the output is the mechanical properties corresponding to the geometric features. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings: Figure 1 is a schematic configuration diagram of the sandwich superstructure of the present invention; Figure 2 is a schematic flow diagram of the collaborative design method of Embodiment 3 of the present invention; Figure 3 is a schematic flow diagram of the collaborative design method of Embodiment 4 of the present invention; Figure 4 is a schematic flow diagram of the collaborative design method of Embodiment 5 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0017] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0018] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, "a plurality" means two or more unless otherwise specifically defined. "Several" means one or more unless otherwise specifically defined.

[0019] The sandwich superstructure with integrated broadband wave absorption and high load-bearing capacity is a key structure for reducing the radar cross-section of weapons and equipment, reducing electromagnetic pollution and electromagnetic interference, and has important engineering application value. Most of the existing designs of sandwich superstructures focus on the electromagnetic performance design of the wave-absorbing layer, but lack systematic analysis and design of their mechanical properties, failing to fully utilize the comprehensive electromagnetic-mechanical advantages of the sandwich superstructure. The designed sandwich superstructures have poor mechanical properties and are difficult to meet the requirements for strength, stability and durability in practical applications.

[0020] To solve the above technical problems, the present invention provides an electromagnetic-mechanical collaborative design method for sandwich superstructures based on machine learning and heuristic optimization algorithms, including the following steps: S100, according to the design usage requirements, set the objective function of the sandwich superstructure including electromagnetic and mechanical properties, and determine the configuration of the sandwich superstructure to be designed according to the objective function; S200, establish a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model according to the machine learning algorithm, and integrate the surrogate models into the heuristic optimization algorithm; S300, based on the heuristic optimization algorithm, batch generate the geometric feature variables of the sandwich superstructure; S400, by sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model, obtain the mechanical properties and electromagnetic properties of the sandwich superstructure; S500, feedback the mechanical properties and electromagnetic properties into the heuristic optimization algorithm to obtain the numerical value of the objective function; S600, according to the numerical value of the objective function, and through the heuristic optimization algorithm, iteratively update the geometric feature variables until the optimal solution corresponding to the objective function is obtained, and complete the design of the sandwich superstructure.

[0021] In the case of adopting the above technical solution, the electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm of the present invention is used to design a sandwich superstructure with excellent mechanical and electromagnetic properties. In this electromagnetic-mechanical collaborative design method, the configuration of the sandwich superstructure to be designed is first determined, and the configuration of the sandwich superstructure further determines the geometric feature variables of the sandwich superstructure. In order to enable the designed sandwich superstructure to have excellent mechanical and electromagnetic properties at the same time, this technical solution further establishes a geometric feature-mechanical property surrogate model (hereinafter referred to as the first model) and a geometric feature-electromagnetic property surrogate model (hereinafter referred to as the second model) according to the machine learning algorithm, and integrates the surrogate models into the heuristic optimization algorithm, and then batch generates the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; after inputting the geometric feature variables into the above two surrogate models respectively, the mechanical properties and electromagnetic properties corresponding to the geometric feature variables can be obtained. That is to say, the geometric feature variables correspond to the mechanical properties in the first model and the electromagnetic properties in the second model; the obtained mechanical properties and electromagnetic properties are fed back into the heuristic optimization algorithm. After obtaining the value of the objective function, according to the value of the objective function, the geometric feature variables are iteratively updated through the heuristic optimization algorithm until the optimal solution corresponding to the objective function is obtained. The optimal solution corresponds to the optimal geometric feature in the geometric feature variables, and then the geometric feature of the sandwich superstructure is determined. Thus, the design of the sandwich superstructure is completed. It can be seen that the above technical solution of the present invention takes into account both the mechanical properties and electromagnetic properties of the sandwich superstructure during the design process of the sandwich superstructure, so that the designed sandwich superstructure has excellent mechanical properties and electromagnetic properties at the same time. Through the above technical solution of the present invention, the problem that most of the designs of the existing sandwich superstructures focus on the electromagnetic performance design of the absorbing layer, lack systematic analysis and design of its mechanical properties, fail to fully utilize the electromagnetic-mechanical comprehensive advantages of the sandwich superstructure, resulting in poor mechanical properties of the designed sandwich superstructure and being difficult to meet the requirements of strength, stability and durability in practical applications is solved.

[0022] To better understand the present invention, the content of the present invention will be further clarified below in conjunction with specific embodiments, but the content of the present invention is not limited to the following embodiments.

[0023] The raw materials or materials used in the following embodiments are all commercially available unless otherwise specified.

[0024] Embodiment 1

[0025] This embodiment provides an electromagnetic-mechanical collaborative design method of a sandwich superstructure based on machine learning and heuristic optimization algorithm, including: Step 1, according to the design usage requirements, set the objective function of the sandwich superstructure including electromagnetic and mechanical properties, and determine the configuration of the sandwich superstructure to be designed according to the objective function; Step 2: Establish a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model according to the machine learning algorithm, and integrate the surrogate models into the heuristic optimization algorithm; Step 3: Based on the heuristic optimization algorithm, batch generate the geometric feature variables of the sandwich superstructure; Step 4: By sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model, obtain the mechanical properties and electromagnetic properties of the sandwich superstructure; Step 5: Feed back the mechanical properties and electromagnetic properties into the heuristic optimization algorithm to obtain the value of the objective function; Step 6: According to the value of the objective function, and through the heuristic optimization algorithm, iteratively update the geometric feature variables until the optimal solution corresponding to the objective function is obtained, and complete the design of the sandwich superstructure.

[0026] Embodiment 2

[0027] This embodiment provides an electromagnetic-mechanical collaborative design method for a sandwich superstructure based on machine learning and a heuristic optimization algorithm, including: S100: According to the design usage requirements, set the objective function of the sandwich superstructure including electromagnetic and mechanical properties, and determine the configuration of the sandwich superstructure to be designed according to the objective function; Among them, the configuration of the sandwich superstructure includes a protective layer, an absorbing layer, and a bearing layer; further, the configuration of the sandwich superstructure includes a wave-transparent protective layer, a broadband absorbing layer, and a shielding bearing layer; the material of the wave-transparent protective layer includes a flat wave-transparent structural material, the material of the broadband absorbing layer includes a periodic array structure metamaterial, and the material of the shielding bearing layer includes a flat high-mechanical-strength material; Further, the flat wave-transparent structural material includes a glass fiber laminate, a quartz fiber laminate, and an aramid fiber laminate, the periodic unit of the periodic array structure metamaterial includes a honeycomb structure, a cylindrical protrusion metamaterial structure, and a groove metamaterial structure, and the flat high-mechanical-strength material includes a carbon fiber laminate; the periodic array structure metamaterial is made of an absorbing composite material, the absorbing composite material is a composite material with electromagnetic loss particles, and the electromagnetic loss particles include carbon nanotubes, graphene, carbonyl iron, and MXene; S200: According to the machine learning algorithm, establish a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model, and integrate the surrogate models into the heuristic optimization algorithm; Among them, the machine learning algorithm includes a BP neural network and a perceptron neural network; the heuristic optimization algorithm includes an immune genetic algorithm and a multi-population genetic algorithm; Further, the input of the geometric feature-mechanical property surrogate model is the geometric feature variable, and the output includes flexural strength, compressive strength, or tensile strength according to the design objective; the input of the geometric feature-electromagnetic property surrogate model is the geometric feature variable, and the output includes absorption bandwidth and minimum reflection loss value according to the design objective. S300, batch-generate the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; Among them, the geometric features of the sandwich superstructure include the thickness of the protective layer, the thickness of the absorbing layer, the thickness of the bearing layer, and the geometric configuration of the absorbing layer; S400, obtain the mechanical properties and electromagnetic properties of the sandwich superstructure by sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model; S500, feedback the mechanical properties and electromagnetic properties into the heuristic optimization algorithm to obtain the value of the objective function; Among them, the objective function is determined according to the design objective, and the design objective includes thin-thickness broadband electromagnetic wave absorption performance and high load-bearing performance; the broadband electromagnetic wave absorption performance includes absorption bandwidth and microwave absorption rate, and the high load-bearing performance includes flexural strength, tensile strength, and compressive strength; S600, according to the value of the objective function, and through the heuristic optimization algorithm, iteratively update the geometric feature variables until the optimal solution corresponding to the objective function is obtained, and complete the design of the sandwich superstructure; Further, as described above, the above geometric feature-mechanical property surrogate model and geometric feature-electromagnetic property surrogate model are established based on a machine learning algorithm. Specifically, in the establishment process of the geometric feature-mechanical property surrogate model, the input of the machine learning algorithm is the geometric feature of the sandwich superstructure, and the output is the mechanical property corresponding to the geometric feature.

[0028] Embodiment 3

[0029] This embodiment provides a method for collaborative electromagnetic-mechanical design of a sandwich superstructure based on machine learning and heuristic optimization algorithm, including: S100, according to the design and usage requirements, set the objective function of the sandwich superstructure including electromagnetic and mechanical properties, and determine the configuration of the sandwich superstructure to be designed according to the objective function; Please refer to Figure 1 , the configuration of the sandwich superstructure to be designed in this embodiment is as shown in the sandwich superstructure 1 in Figure 1 , and its front view and top view are shown in detail in Figure 1 , and this sandwich superstructure is composed of a wave-transparent protective layer, a broadband absorbing layer, and a shielding bearing layer; Further, in this embodiment, in order to ensure that the electromagnetic wave energy can smoothly enter the wave-absorbing layer, unidirectional glass fiber reinforced plastic (GFRP, belonging to glass fiber reinforced laminate) is selected as the wave-transmitting protective layer material. In addition, unidirectional carbon fiber reinforced plastic is selected as the shielding bearing layer material; the broadband wave-absorbing layer selects a periodic array structure metamaterial, and this periodic array structure metamaterial is made of a wave-absorbing composite material, and this wave-absorbing composite material is a composite material with carbon nanotubes and carbonyl iron powder; S200, establish a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model according to the machine learning algorithm, and integrate the surrogate models into the heuristic optimization algorithm; wherein, the machine learning algorithm is a BP neural network or a perceptron neural network; the heuristic optimization algorithm is an immune genetic algorithm or a multi-population genetic algorithm; Further, the input of the geometric feature-mechanical property surrogate model is the geometric feature variable, and the output includes the flexural strength, compressive strength or tensile strength according to the design objective; the input of the geometric feature-electromagnetic property surrogate model is the geometric feature variable, and the output includes the wave-absorbing bandwidth and the minimum reflection loss value according to the design objective; Please refer to Figure 1 , the sandwich superstructure of this embodiment uses the eight geometric feature variables of sandwich superstructure 1 as the input of the surrogate model. These geometric feature variables are respectively the thickness of the wave-transmitting protective layer ( t 1), the thickness of the shielding bearing layer ( t 5), the thickness of the upper cylindrical convex structure of the wave-absorbing metamaterial ( t 2), the thickness of the lower cylindrical convex structure of the wave-absorbing metamaterial ( t 3), the thickness of the bottom plate of the wave-absorbing metamaterial ( t 4), the period spacing of the periodic unit of the sandwich superstructure ( l ), and the diameter d1 of the upper cylindrical convex structure of the wave-absorbing metamaterial and the diameter d2 of the lower cylindrical convex structure of the wave-absorbing metamaterial; Further, the above geometric feature-mechanical property surrogate model and geometric feature-electromagnetic property surrogate model are established based on the machine learning algorithm. Specifically, in the establishment process of the geometric feature-mechanical property surrogate model, the input of the machine learning algorithm is the geometric features of the sandwich superstructure, and the output is the mechanical properties corresponding to the geometric features; S300, batch generate the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; Among them, the geometric features of the sandwich superstructure include the thickness of the protective layer, the thickness of the wave-absorbing layer, the thickness of the bearing layer, and the geometric configuration of the wave-absorbing layer; S400. Obtain the mechanical properties and electromagnetic properties of the sandwich superstructure by sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model; S500. Feed back the mechanical properties and electromagnetic properties into the heuristic optimization algorithm to obtain the value of the objective function. Among them, the objective function is determined according to the design objective, and the design objective includes thin-thickness broadband electromagnetic wave absorption performance and high load-bearing performance; the broadband electromagnetic wave absorption performance includes the wave absorption bandwidth and microwave absorption rate, and the high load-bearing performance includes bending strength, tensile strength, and compressive strength; In this embodiment, taking the wave absorption bandwidth and bending strength as the design objectives, for the convenience of calculation, the design objectives are converted into the following function form, that is, the objective function of step S100 is obtained: ; Among them, represents the functional relationship, represents the maximum value function symbol, and are both weight coefficients, denoted as the first weight coefficient, the second weight coefficient, and the third weight coefficient respectively, represents the normalized effective wave absorption bandwidth, represents the normalized bending strength, represents the normalized thickness of the sandwich superstructure; It should be noted that in this embodiment, the normalized effective wave absorption bandwidth is obtained by dividing the effective absorption band EAB by the entire bandwidth of 2-40 GHz, such as EAB / 38 GHz; the normalized bending strength is obtained by dividing the bending strength value B s by 100 MPa, such as B s / 100 MPa; the weight coefficient α i is determined according to the actual functional characteristic requirements of the sandwich superstructure. The value of the weight coefficient α i reflects the relative importance of various functions in the overall design objective. The primary performance requirement of the sandwich superstructure is ultra-wideband electromagnetic wave absorption performance. Therefore, in this embodiment, the corresponding weight coefficient α 1 is set to 0.7. Correspondingly, α 2 and α 3 are set to 0.3 and 0.2 respectively; S600. Iteratively update the geometric feature variables according to the value of the above objective function through the heuristic optimization algorithm until the optimal solution corresponding to the objective function is obtained, thus completing the design of the sandwich superstructure. Specifically, it includes: Regarding the geometric features of the sandwich superstructure as design variables and the above objective function as the design goal, a multi-population genetic algorithm for the sandwich superstructure is established to achieve electromagnetic-mechanical collaborative optimization design. Please refer to Figure 2 , in this multi-population genetic algorithm, first, a batch of geometric feature variables are generated in the geometric feature design space of the sandwich superstructure. The generated geometric feature variables are input into the geometric feature-mechanical property and geometric feature-electromagnetic property surrogate models established based on machine learning to quickly calculate the absorption bandwidth (electromagnetic response) and bending strength (mechanical response). Then, the calculated electromagnetic and mechanical property responses are fed back into the multi-population genetic algorithm to calculate the design objective function. According to the value of the calculated objective function, the geometric feature variables are iteratively updated through genetic operations such as selection, crossover, and mutation until the optimal solution is obtained. Specifically, from Figure 2 it can be seen that, that is to say, during the process of iteratively updating the geometric feature variables, after being calculated by the optimization objective function and meeting the termination condition, the optimal structural parameters of the sandwich superstructure are obtained, thus completing the design of the sandwich superstructure. Additionally, if the termination condition is not met, return to the above step S300, and repeat steps S300 to S600 until the termination condition is met. The termination condition can be the number of iterations. For example, the number of iterations can be 10 times, 15 times, or 20 times. For example, in the geometric feature-mechanical property surrogate model, select the BP (Back Propagation) artificial neural network as the surrogate model algorithm. Using the geometric feature design variables of the sandwich superstructure as input data and the bending strength of the sandwich superstructure as the output result, the relevant geometric feature-mechanical property training data can be obtained from finite element simulation analysis. In the geometric feature-electromagnetic property surrogate model, select the BP artificial neural network as the surrogate model algorithm. Using the geometric feature design variables of the sandwich superstructure as input data and the absorption bandwidth of the sandwich superstructure as the output result, the relevant geometric feature-mechanical property training data can be obtained from electromagnetic simulation analysis.

[0030] Example 4 This embodiment provides a method for electromagnetic-mechanical collaborative design of a sandwich superstructure based on machine learning and heuristic optimization algorithm, including: S100. According to the design usage requirements, set the objective function of the sandwich superstructure including electromagnetic and mechanical properties, and determine the configuration of the sandwich superstructure to be designed according to the objective function; Please refer to Figure 1 , the configuration of the sandwich superstructure to be designed in this embodiment is as Figure 1As shown in the sandwich superstructure 3 therein, the front view and top view are shown in detail in Figure 1 , and the sandwich superstructure is composed of a wave-transparent protective layer, a broadband absorbing layer, and a shielding and bearing layer; Furthermore, in this embodiment, in order to ensure that the electromagnetic wave energy can smoothly enter the absorbing layer, unidirectional quartz fiber reinforced plastic is selected as the material of the wave-transparent layer. In addition, unidirectional carbon fiber reinforced plastic is selected as the material of the shielding layer; the broadband absorbing layer selects a periodic array structure metamaterial, and the periodic array structure metamaterial is made of an absorbing composite material, and the absorbing composite material is a composite material with carbon nanotubes. The periodic unit of the periodic array structure metamaterial is a honeycomb structure; S200. Establish a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model according to the machine learning algorithm, and integrate the surrogate models into the heuristic optimization algorithm; wherein, the machine learning algorithm includes a BP neural network and a perceptron neural network; the heuristic optimization algorithm includes an immune genetic algorithm and a multi-population genetic algorithm; Furthermore, the input of the geometric feature-mechanical property surrogate model is the geometric feature variable, and the output includes the bending strength, compressive strength or tensile strength according to the design goal; the input of the geometric feature-electromagnetic property surrogate model is the geometric feature variable, and the output includes the absorbing bandwidth and the minimum reflection loss value according to the design goal; Please refer to Figure 1 , the sandwich superstructure of this embodiment uses five geometric feature variables of the sandwich superstructure 3 as the input of the surrogate model. These geometric feature variables are respectively the thickness of the wave-transparent protective layer ( t 1), the thickness of the shielding and bearing layer ( t 3), the overall thickness of the honeycomb core ( t 2), the wall thickness of the honeycomb core ( w ), and the side length of the honeycomb core ( l ); Furthermore, the above geometric feature-mechanical property surrogate model and geometric feature-electromagnetic property surrogate model are established based on the machine learning algorithm. Specifically, in the establishment process of the geometric feature-mechanical property surrogate model, the input of the machine learning algorithm is the geometric feature of the sandwich superstructure, and the output is the mechanical property corresponding to the geometric feature; S300. Batch generate the geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; Among them, the geometric features of the sandwich superstructure include the thickness of the protective layer, the thickness of the absorbing layer, the thickness of the bearing layer, and the geometric configuration of the absorbing layer; S400. By sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model, the mechanical property and electromagnetic property of the sandwich superstructure are obtained; S500. Feed the mechanical properties and electromagnetic properties back into the heuristic optimization algorithm to obtain the value of the objective function. Among them, the objective function is determined according to the design objectives, and the design objectives include thin-thickness broadband electromagnetic wave absorption performance and high load-bearing performance. The broadband electromagnetic wave absorption performance includes the absorption bandwidth and microwave absorption rate, and the high load-bearing performance includes bending strength, tensile strength, and compressive strength. In this embodiment, taking the absorption bandwidth and bending strength as the design objectives, for the convenience of calculation, the design objectives are converted into the following function form, that is, the objective function of step S100 is obtained: ; Among them, represents the functional relationship, represents the maximum value function, and are both weight coefficients, which are respectively denoted as the first weight coefficient, the second weight coefficient, and the third weight coefficient. represents the normalized effective absorption bandwidth, represents the normalized compressive strength, represents the normalized thickness of the sandwich superstructure. It should be noted that in this embodiment, the normalized effective absorption bandwidth is obtained by dividing the effective absorption band EAB by the entire bandwidth of 2 - 40 GHz, such as EAB / 38 GHz; the normalized compressive strength is obtained by dividing the bending strength value M s by 100 MPa, such as M s / 100 MPa; the weight coefficient α i is determined according to the actual functional characteristic requirements of the sandwich superstructure. The numerical value of the weight coefficient α i reflects the relative importance of various functions in the overall design objectives. The primary performance requirement of the sandwich superstructure is ultra-wideband electromagnetic wave absorption performance. Therefore, in this embodiment, the corresponding weight coefficient α 1 is set to 0.7. Correspondingly, α 2 and α 3 are set as weight coefficients of 0.3 and 0.2 respectively. S600. According to the value of the above objective function, and through the heuristic optimization algorithm, iterate and update the geometric feature variables until the optimal solution corresponding to the objective function is obtained, and complete the design of the sandwich superstructure. Specifically, it includes: Taking the geometric characteristics of the sandwich superstructure as design variables and the above objective function as the design goal, a multi-population genetic algorithm for the sandwich superstructure is established to achieve the electromagnetic-mechanical collaborative optimization design. Please refer to Figure 3 , in this multi-population genetic algorithm, first, a batch of geometric feature variables are generated in the geometric feature design space of the sandwich superstructure. The generated geometric feature variables are input into the geometric feature-mechanical performance and geometric feature-electromagnetic performance surrogate models established based on machine learning to quickly calculate the absorption bandwidth (electromagnetic response) and bending strength (mechanical response). Then, the calculated electromagnetic performance and mechanical performance responses are fed back into the multi-population genetic algorithm to calculate the design objective function. According to the calculated objective function value, through genetic operations such as selection, crossover, and mutation, the geometric feature variables are iteratively updated until the optimal solution is obtained. Specifically, from Figure 3 viewpoint, that is to say, during the iterative update of the geometric feature variables, after the optimization objective function is calculated and the termination condition is met, the optimal structural parameters of the sandwich superstructure are obtained, and the design of the sandwich superstructure is completed; additionally, if the termination condition is not met, return to the above step S300 and repeat steps S300 to S600 until the termination condition is met. The termination condition can be the number of iterations. For example, the number of iterations can be 10 times, 15 times, or 20 times; for example, in the geometric feature-mechanical performance surrogate model, the BP (Back Propagation) artificial neural network is selected as the surrogate model algorithm. Taking the geometric feature design variables of the sandwich superstructure as input data and the bending strength of the sandwich superstructure as the output result, the relevant geometric feature-mechanical performance training data can be obtained from finite element simulation analysis; in the geometric feature-electromagnetic performance surrogate model, the BP artificial neural network is selected as the surrogate model algorithm. Taking the geometric feature design variables of the sandwich superstructure as input data and the absorption bandwidth of the sandwich superstructure as the output result, the relevant geometric feature-mechanical performance training data can be obtained from electromagnetic simulation analysis.

[0031] Example 5 The electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm provided in this embodiment is basically the same as that in Embodiment 4, except that in step S100, the configuration of the sandwich superstructure to be designed in this embodiment is the same as that of Figure 1 the sandwich superstructure shown in 2, and the front view and top view are shown in detail in Figure 1 ; in step S200, please refer to Figure 1 , in this embodiment, the sandwich superstructure uses five geometric feature variables of the sandwich superstructure 2 as the input of the surrogate model. These geometric feature variables are respectively the thickness of the wave-transparent layer ( t 1), the thickness of the wave-absorbing layer ( t 2), and the thickness of the shielding layer ( t3), wall thickness ( w ), and side length ( l ); In step S600, according to the value of the objective function, the geometric feature variables are iteratively updated through the heuristic optimization algorithm until the optimal solution corresponding to the objective function is obtained, and the design of the sandwich superstructure is completed; specifically including: Regarding the geometric features of the sandwich superstructure as design variables and the above objective function as the design goal, a multi-population genetic algorithm for the sandwich superstructure is established to achieve electromagnetic-mechanical collaborative optimization design. Please refer to Figure 4 , in this multi-population genetic algorithm, first, geometric feature variables are batch-generated in the geometric feature design space of the sandwich superstructure. The generated geometric feature variables are input into the geometric feature-mechanical performance and geometric feature-electromagnetic performance surrogate models established based on machine learning to quickly calculate the absorption bandwidth (electromagnetic response) and bending strength (mechanical response). Then, the calculated electromagnetic performance and mechanical performance responses are fed back into the multi-population genetic algorithm to calculate the design objective function. According to the calculated objective function value, through genetic operations such as selection, crossover, and mutation, the geometric feature variables are iteratively updated until the optimal solution is obtained. Specifically, from Figure 4 viewpoint, that is to say, during the process of iteratively updating the geometric feature variables, after calculating through the optimization objective function and meeting the termination condition, the optimal structural parameters of the sandwich superstructure are obtained, and the design of the sandwich superstructure is completed; additionally, if the termination condition is not met, return to the above step S300 and repeat steps S300 to S600 until the termination condition is met. The termination condition can be the number of iterations. For example, the number of iterations can be 10 times, 15 times, or 20 times. For example, in the geometric feature-mechanical performance surrogate model, the BP (Back Propagation) artificial neural network is selected as the surrogate model algorithm. Using the geometric feature design variables of the sandwich superstructure as input data and the bending strength of the sandwich superstructure as the output result, the relevant geometric feature-mechanical performance training data can be obtained from finite element simulation analysis; in the geometric feature-electromagnetic performance surrogate model, the BP artificial neural network is selected as the surrogate model algorithm. Using the geometric feature design variables of the sandwich superstructure as input data and the absorption bandwidth of the sandwich superstructure as the output result, the relevant geometric feature-mechanical performance training data can be obtained from electromagnetic simulation analysis.

[0032] In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in a suitable manner in any one or more embodiments or examples.

[0033] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims described above.

Claims

1. A co - design method for electromagnetic - mechanical of sandwich superstructure based on machine learning and heuristic optimization algorithm, characterized in that, Including: Setting the objective function of the sandwich superstructure and determining the configuration of the sandwich superstructure to be designed according to the objective function; Establishing a geometric feature-mechanical property surrogate model and a geometric feature-electromagnetic property surrogate model based on a machine learning algorithm, and integrating the surrogate models into a heuristic optimization algorithm; Batch generating geometric feature variables of the sandwich superstructure based on the heuristic optimization algorithm; By sequentially inputting the geometric feature variables into the geometric feature-mechanical property surrogate model and the geometric feature-electromagnetic property surrogate model, obtaining the mechanical property and electromagnetic property of the sandwich superstructure; Feeding back the mechanical property and electromagnetic property into the heuristic optimization algorithm to obtain the value of the objective function; According to the value of the objective function and through the heuristic optimization algorithm, iteratively updating the geometric feature variables until the optimal solution corresponding to the objective function is obtained, and completing the design of the sandwich superstructure.

2. The electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 1, characterized in that The configuration of the sandwich superstructure includes a protective layer, an absorbing layer, and a load-bearing layer.

3. The electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 2, characterized in that, The heuristic optimization algorithm includes an immune genetic algorithm and a multi-population genetic algorithm; And / or, the machine learning algorithm includes a BP neural network and a perceptron neural network.

4. The electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 3, characterized in that, The input of the geometric feature-mechanical property surrogate model is the geometric feature variable, and the output includes bending strength, compressive strength, or tensile strength according to the design objective; And / or, the input of the geometric feature-electromagnetic property surrogate model is the geometric feature variable, and the output includes absorbing bandwidth and minimum reflection loss value according to the design objective.

5. The electromagnetic-mechanical collaborative design method for the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 4, characterized in that The geometric features of the sandwich superstructure include the thickness of the protective layer, the thickness of the absorbing layer, the thickness of the load-bearing layer, and the geometric configuration of the absorbing layer.

6. The electromagnetic-mechanical co-design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 5, characterized in that The objective function is determined according to the design objective, and the design objective includes thin thickness, broadband electromagnetic wave absorption performance, and high load-bearing performance; wherein: The broadband electromagnetic wave absorption performance includes absorbing bandwidth and microwave absorption rate, and the high load-bearing performance includes bending strength, tensile strength, and compressive strength.

7. The electromagnetic-mechanical collaborative design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 6, characterized in that The configuration of the sandwich superstructure includes a wave-transparent protective layer, a broadband absorbing layer, and a shielding load-bearing layer; And / or, the material of the wave-transparent protective layer includes a flat wave-transparent structural material, the material of the broadband absorbing layer includes a periodic array structure metamaterial, and the material of the shielding load-bearing layer includes a flat high-mechanical-strength material.

8. The electromagnetic-mechanical co-design method for sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 7, characterized in that The flat wave-transparent structural material includes a glass fiber laminate, a quartz fiber laminate, and an aramid fiber laminate. The periodic unit of the periodic array structure metamaterial includes a honeycomb structure, a cylindrical protrusion metamaterial structure, and a groove metamaterial structure. The flat high-mechanical-strength material includes a carbon fiber laminate; And / or, the periodic array structure metamaterial is made of an absorbing composite material, the absorbing composite material is a composite material with electromagnetic loss particles, and the electromagnetic loss particles include carbon nanotubes, graphene, carbonyl iron, and MXene.

9. The electromagnetic-mechanical co-design method of the sandwich superstructure based on machine learning and heuristic optimization algorithm according to claim 8, characterized in that, In the process of establishing a geometric feature-mechanical property surrogate model based on a machine learning algorithm, the input of the machine learning algorithm is the geometric feature of the sandwich superstructure, and the output is the mechanical property corresponding to the geometric feature.

Images