Wave impedance gradient protection material design method and device based on neural network

Abstract

The invention provides a wave impedance gradient protection material design method and device based on a neural network, and the method comprises the steps: carrying out the simulation test of the performance of a wave impedance gradient material based on fluid dynamics, obtaining the surface density, layering proportion and protection performance values as samples, building a neural network prediction model through a multi-layer perception mechanism, and obtaining a prediction model; fitting a mapping relation among the surface density, the layering proportion and the protection performance numerical value to obtain a fitting curve function; inputting the target protection performance value to the neural network prediction model, and outputting a surface density prediction value and a paving layer proportion prediction value; a fitting curve function is used as a fitness function in a genetic algorithm to optimize a surface density predicted value and a paving layer proportion predicted value to obtain a target surface density and a target paving layer proportion, and a neural network prediction model and a genetic algorithm are established in combination with a multi-layer perception mechanism to perform optimization. The required target surface density and the target layering proportion can be generated, the application scene is wide, and the design efficiency of the wave impedance gradient material is improved.

Description

Technical Field

[0001] The present invention relates to the field of space debris protection technology, and in particular to a method and device for designing wave impedance gradient protection materials based on neural networks. Background Art

[0002] The basic principle of the passive protection structure of spacecraft space debris is to set up a protective screen on the outside of the spacecraft cabin wall, so that the initial incident debris will collide with the protective screen at ultra-high speed, causing it to break, melt or even vaporize to form a secondary debris cloud, thereby minimizing and dispersing the kinetic energy of the incident debris, significantly reducing the collision kinetic energy density acting on the spacecraft cabin wall, and alleviating the damage and destructive effect of the rear target plate. Therefore, the ability of the protective screen material to dissipate and convert the kinetic energy of the projectile is an important indicator for evaluating the performance of the protective structure. Wave impedance gradient material refers to a typical type of functional gradient material in which the wave impedance (the product of the zero-pressure volume sound velocity and the density) changes according to a certain rule along the thickness direction. When the wave impedance gradient material is used as a protective screen, the unique energy dissipation characteristics under ultra-high-speed impact can be used to improve the protection ability against projectiles.

[0003] However, most of the current wave impedance gradient protection material design methods have relatively single application scenarios, resulting in low design efficiency. Summary of the invention

[0004] The present invention provides a wave impedance gradient protective material design method and device based on neural network, which are used to solve the defects of single application scenario and low design efficiency of wave impedance gradient protective material design in the prior art.

[0005] In a first aspect, the present invention provides a method for designing a wave impedance gradient protective material based on a neural network, comprising:

[0006] Based on fluid dynamics, the performance of the wave impedance gradient material is simulated and tested to obtain material design parameters, wherein the material design parameters include surface density, ply ratio and protective performance value;

[0007] Taking the surface density, the ply ratio and the protective performance value as samples, a neural network prediction model is constructed using a multi-layer perceptron to fit the mapping relationship between the surface density, the ply ratio and the protective performance value to obtain a fitting curve function;

[0008] Inputting a target protection performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value;

[0009] The fitting curve function is used as a fitness function in a genetic algorithm to optimize the predicted value of the surface density and the predicted value of the ply ratio, so as to obtain a target surface density and a target ply ratio.

[0010] According to a wave impedance gradient protective material design method based on a neural network provided by the present invention, the use of the fitting curve function as a fitness function in a genetic algorithm to optimize the surface density prediction value and the ply ratio prediction value to obtain a target surface density and a target ply ratio includes:

[0011] Using the predicted value of the surface density and the predicted value of the ply ratio as genes, genetic iteration is performed, and parameter configurations of the surface density and the ply ratio are generated as candidate solutions in each generation;

[0012] Using the fitting curve function as a fitness function, determining the performance of the candidate solution;

[0013] When the performance reaches the stop condition, the optimized target surface density and target ply ratio are output.

[0014] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, before performing genetic iteration, the method further includes:

[0015] The genes are efficiently represented within the framework of the genetic algorithm using binary coding.

[0016] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, after determining the performance of the candidate solution, the method further includes:

[0017] The candidate solution that meets the target requirements is selected as the parent, and then new candidate solutions are generated through continuous iterations of crossover and mutation operations.

[0018] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, the stopping condition is: the number of iterations corresponding to the minimum mean square error.

[0019] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, the neural network prediction model includes: an input layer, an output layer and a hidden layer;

[0020] The input layer includes the surface density and ply ratio, and the output layer includes the protection performance value.

[0021] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, the number of hidden layers is 3, and the number of neurons in each hidden layer is 10, 6 and 3.

[0022] According to a neural network-based wave impedance gradient protective material design method provided by the present invention, the protective performance value includes a critical projectile diameter;

[0023] The wave impedance gradient material performance is simulated and tested based on fluid dynamics to obtain material design parameters, including:

[0024] Set the surface density to multiple levels, set the ply ratio to multiple levels;

[0025] The AUTODYN dynamic simulation program was used to carry out numerical simulation of the process of spherical projectiles hypervelocity impacting the PTFE / Al energetic wave impedance gradient material protection structure, and the critical projectile diameter corresponding to different surface density and ply ratio parameter settings was obtained.

[0026] In a second aspect, the present invention further provides a wave impedance gradient protective material design device based on a neural network, comprising:

[0027] A simulation module, used to simulate and test the performance of the wave impedance gradient material based on fluid dynamics to obtain material design parameters, wherein the material design parameters include surface density, ply ratio and protective performance value;

[0028] A construction module is used to use the surface density, the ply ratio and the protective performance value as samples, construct a neural network prediction model using a multi-layer perceptron, fit the mapping relationship between the surface density, the ply ratio and the protective performance value, and obtain a fitting curve function;

[0029] A prediction module, used for inputting a target protection performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value;

[0030] The optimization module is used to optimize the predicted value of the surface density and the predicted value of the ply ratio by using the fitting curve function as the fitness function in the genetic algorithm to obtain the target surface density and the target ply ratio.

[0031] In a third aspect, the present invention further provides a protective screen, wherein the wave impedance gradient protective material in the protective screen is obtained by the wave impedance gradient protective material design method based on a neural network as described in any one of the above items.

[0032] In a fourth aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for designing wave impedance gradient protective materials based on a neural network as described above is implemented.

[0033] In a fifth aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the neural network-based wave impedance gradient protective material design methods described above.

[0034] In a sixth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the neural network-based wave impedance gradient protective material design methods described above.

[0035] The present invention provides a method and device for designing wave impedance gradient protective materials based on a neural network, comprising: performing simulation tests on wave impedance gradient material performance based on fluid dynamics to obtain material design parameters, wherein the material design parameters include surface density, ply ratio and protective performance values; using the surface density, ply ratio and protective performance values ​​as samples, constructing a neural network prediction model using a multi-layer perceptron, fitting the mapping relationship between the surface density, ply ratio and protective performance values, and obtaining a fitting curve function; inputting a target protective performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value; optimizing the surface density prediction value and the ply ratio prediction value using the fitting curve function as a fitness function in a genetic algorithm to obtain a target surface density and a target ply ratio, constructing a neural network prediction model using a multi-layer perceptron, and then optimizing it using a legacy algorithm to generate the required target surface density and target ply ratio, which has a wide range of application scenarios and high accuracy, and effectively improves the design efficiency of wave impedance gradient materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 is a flow chart of a method for designing a wave impedance gradient protective material based on a neural network provided in this embodiment;

[0038] Figure 2 is a schematic diagram of the structure of the neural network prediction model provided in this embodiment;

[0039] Figure 3 It is a schematic diagram of the overall flow of the optimization algorithm provided in this embodiment;

[0040] Figure 4 is a schematic diagram of the encoding form adopted by the genetic algorithm provided in this embodiment;

[0041] Figure 5 is a schematic structural diagram of a wave impedance gradient protective material design device based on a neural network provided in this embodiment;

[0042] Figure 6 It is a schematic diagram of the structure of the electronic device provided in this embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Figure 1 It is a flow chart of the neural network-based wave impedance gradient protective material design method provided in this embodiment.

[0045] like Figure 1 As shown, the wave impedance gradient protective material design method based on neural network provided in the embodiment of the present invention mainly includes the following steps:

[0046] 101. Based on fluid dynamics, the performance of wave impedance gradient materials is simulated and tested to obtain material design parameters, which include surface density, ply ratio and protective performance value.

[0047] In a specific implementation process, the surface density and layer ratio of the wave impedance gradient material directly affect the anti-collision performance of the shield. The surface density refers to the mass per unit area of ​​the material, while the layer ratio indicates the relative proportion of different wave impedance gradient materials in the shield. By adjusting the surface density and layer ratio, the response of the shield to impact and shock can be controlled.

[0048] First, the increase in surface density means that more materials are involved in protection, which can provide higher mass and inertia, thereby absorbing and dispersing impact energy and reducing deformation and damage after impact. Secondly, by adjusting the ply ratio of different materials, a gradient change in wave impedance can be achieved, thereby optimizing the propagation and absorption of shock waves. Different materials have different wave impedance characteristics. By reasonably selecting the ply ratio, a gradually changing wave impedance can be achieved in the protective screen structure, so that the shock wave can be gradually weakened and dispersed when propagating in the structure. Therefore, the correct selection of the ply ratio can significantly improve the collision resistance of the protective screen.

[0049] By building a neural network prediction model, the protection performance values ​​under different surface densities and ply ratios can be predicted. Similarly, when the protection performance value is known, the corresponding surface density and ply ratio parameter configuration can be reversed through the neural network prediction model.

[0050] Therefore, in order to build a neural network prediction model, samples are first obtained. The performance of the wave impedance gradient material is simulated and tested through fluid dynamics to obtain material design parameters, which include surface density, ply ratio and protection performance value. The protection performance value can be the critical projectile diameter. Different surface density sizes and ply ratios correspond to different critical projectile diameters. Therefore, through simulation tests, material design parameters under various combinations can be effectively obtained.

[0051] This embodiment uses the classic Whipple protection structure, and the surface density of the wave impedance gradient protection screen is between 0.279 and 0.837 g / cm 2 , the rear wall is 2.5mm thick aluminum alloy, and the total spacing of the protective structure is 100mm. In order to comprehensively analyze the influence of different combinations of surface density and ply ratio, the surface density is set to 4 levels and the ply ratio is set to 15 levels, forming the parameter conditions shown in Table 1. The AUTODYN dynamic simulation program is used to carry out numerical simulation research on the process of spherical projectile hypervelocity impact on PTFE / Al energetic wave impedance gradient material protective structure. When conducting hypervelocity impact numerical simulation, in order to reduce the number of simulations, the following experimental design is adopted, and its parameter levelization experimental combination is shown in Table 2.

[0052] Table 1

[0053]

[0054] Table 2

[0055]

[0056] The impact speed of the projectile in the simulation is 6.5 km / s. The critical projectile diameter corresponding to the protective structure of each experimental scheme is obtained by changing the projectile diameter. The projectile diameter change step is 0.1 mm. The thickness of each component material and the critical projectile diameter corresponding to the simulation are shown in Table 3.

[0057] Table 3

[0058]

[0059]

[0060] 102. Taking surface density, ply ratio and protective performance values ​​as samples, a neural network prediction model is constructed using a multi-layer perceptron to fit the mapping relationship between surface density, ply ratio and protective performance values, and obtain a fitting curve function.

[0061] After obtaining the surface density, ply ratio and protective performance, they are preprocessed, including normalization to eliminate the impact of different magnitudes, as well as possible denoising and data cleaning to ensure the quality and consistency of the input data, which is then used as a sample for training the neural network prediction model.

[0062] The training process is the process of fitting the mapping relationship between surface density, ply ratio and protection performance data. The neural network prediction model is constructed through a multi-layer perceptron (MLP). The multi-layer perceptron (MLP) consists of multiple layers: an input layer, one or more hidden layers, and an output layer. Figure 2 The input parameters x1, x2, x3, and x4 include the proportion and surface density of each layer, and the value y of the protection performance (such as the critical projectile diameter) is used as the output layer of the neural network.

[0063] A 3-layer multilayer perceptron (MLP) was used to fit the input variables, i.e., ply ratio and surface density, and the output variable, i.e., the diameter of the projectile to be protected. Figure 2 As shown in the figure, Input Layer represents the input layer, Hidden Layer represents the hidden layer, which is 3 layers, the number of neurons in the hidden layer is 10, 6 and 3 respectively, and Output Layer represents the output layer. For the wave impedance gradient material composed of three component materials, the input parameters of the multilayer perceptron include four key physical quantities: the proportion of each layer and the surface density. The numerical value of the protection performance (such as the critical projectile diameter) is used as the output layer of the neural network.

[0064] The ratio of training set and test set is set to 80% and 20%. The fitting accuracy of the training set is 92.67%, and the fitting accuracy of the test set is 93.63%, respectively, indicating that the neural network prediction model can better fit the results of hypervelocity impact numerical simulation calculations. During the training process, the weights and biases of the model will be continuously adjusted to minimize the difference between the predicted values ​​and the actual values. This process usually involves multiple iterations until the performance of the model reaches the predetermined standard or no longer improves significantly. After the training is completed, the model is tested with an independent test data set to evaluate its generalization ability and prediction accuracy, to help understand the performance of the model on unseen data, thereby ensuring its reliability and effectiveness in practical applications.

[0065] 103. Input the target protection performance value into the neural network prediction model, and output the surface density prediction value and the ply ratio prediction value.

[0066] Through simulation testing, we obtain training samples of the neural network prediction model, and then perform data training based on the multi-layer perceptron to obtain the neural network prediction model. Then, we can input the target protection performance value into the neural network prediction model, and after internal calculation, we can obtain the surface density prediction value and ply ratio protection value corresponding to the target protection performance value.

[0067] In order to verify the accuracy of the prediction results of the neural network prediction model, the results of the hypervelocity impact numerical simulation are compared with those of the MLP prediction, as shown in Table 4.

[0068] Table 4

[0069]

[0070]

[0071] 104. The surface density prediction value and the layer ratio prediction value are optimized by using the fitting curve function as the fitness function in the genetic algorithm to obtain the target surface density and the target layer ratio.

[0072] It can be seen from the data in Table 4 that there are slight differences between the predicted values ​​and the simulation results. Therefore, the predicted values ​​of surface density and ply ratio are optimized by genetic algorithm to obtain more accurate target surface density and target ply ratio. The results of neural network fitting are used to guide the search direction of genetic algorithm, so as to more effectively find the optimal parameter configuration of surface density and ply ratio.

[0073] like Figure 3 The figure shows a schematic diagram of the optimization algorithm flow. After the population is initialized, it is divided into a training set and a test set, and then the model is trained to obtain the prediction accuracy evaluation of the MLP model, that is, the fitting curve function representing the mapping relationship between the surface density, the ply ratio and the critical projectile diameter is obtained. The fitting curve function is used as the fitness function to evaluate the individual fitness in the population, and then the selection operation, crossover operation and mutation operation are iterated until the conditions are met, and the optimal combination of surface density and ply ratio is returned for simulation and visualization. By combining genetic algorithms with MLP, it can be applied to the parameter configuration of surface density and ply ratio combinations in a variety of scenarios, which improves the efficiency of the design of wave impedance gradient protection materials.

[0074] like Figure 4The figure shows the coding format used by the genetic algorithm. The specific genetic optimization process is as follows: using the surface density prediction value and the ply ratio prediction value as genes, using binary coding to effectively represent the genes within the framework of the genetic algorithm, and performing genetic iterations. Each generation generates parameter configurations of surface density and ply ratio as candidate solutions; using the fitting curve function as the fitness function to determine the performance of the candidate solution; selecting the candidate solution that meets the target requirements as the parent generation, and then continuously iterating through crossover and mutation operations to generate new candidate solutions. When the performance reaches the stopping condition, that is, the number of iterations corresponding to the minimum mean square error, the optimized target surface density and target ply ratio are output.

[0075] Based on Table 4, the above algorithm and parameters were used to optimize the wave impedance gradient material ply. The optimization algorithm performed 50 generations of genetic optimization and reached the termination condition. The protection performance of the offspring was enhanced with the increase of genetic generations, indicating that the algorithm can optimize the wave impedance gradient material ply.

[0076] Table 5 gives the optimal layer ratio of titanium alloy-aluminum alloy-PTFE / Al at typical surface density and the critical projectile diameter of the corresponding protective structure.

[0077] Table 5

[0078]

[0079] A hypervelocity impact numerical simulation was constructed based on the AUTODYN dynamics simulation program to verify the reliability of the artificial intelligence model prediction. The parameters of the energetic wave impedance gradient material in the simulation model are shown in the table, and the critical projectile diameter obtained by calculation is shown in the table. The results show that the maximum error between the critical projectile diameter calculated by simulation and the model prediction result does not exceed 5%, as shown in Table 6.

[0080] Table 6

[0081]

[0082]

[0083] The present invention uses an experimental design method to perform experimental simulation design on the factors affecting the performance of wave impedance gradient materials, and the experiment is carried out based on fluid dynamics simulation software. The material design parameters corresponding to the experiment and the protective performance results obtained by simulation are used as artificial neural network learning samples, and the mapping relationship between material design parameters and protective performance is established using artificial neural networks to obtain an artificial neural network model that can predict material design parameters and protective performance. The relevant material design parameters are optimized based on and through genetic algorithms, thereby greatly improving the design efficiency. This optimization strategy that combines the MLP prediction algorithm and the genetic optimization algorithm not only improves the calculation efficiency and accuracy, but also greatly enhances the algorithm's ability to solve complex optimization problems.

[0084] Based on the same general inventive concept, the present invention also protects a wave impedance gradient protective material design device based on a neural network. The wave impedance gradient protective material design device based on a neural network provided by the present invention is described below. The wave impedance gradient protective material design device based on a neural network described below and the wave impedance gradient protective material design method based on a neural network described above can be referenced to each other.

[0085] Figure 5 It is a structural schematic diagram of the wave impedance gradient protective material design device based on neural network provided in this embodiment.

[0086] like Figure 5 As shown, this embodiment provides a wave impedance gradient protective material design device based on a neural network, comprising:

[0087] A simulation module 501 is used to simulate and test the performance of the wave impedance gradient material based on fluid dynamics to obtain material design parameters, where the material design parameters include surface density, ply ratio, and protective performance value;

[0088] A construction module 502 is used to construct a neural network prediction model using a multi-layer perceptron based on the surface density, the ply ratio and the protective performance value as samples, fit the mapping relationship between the surface density, the ply ratio and the protective performance value, and obtain a fitting curve function;

[0089] Prediction module 503, used to input the target protection performance value into the neural network prediction model, and output the surface density prediction value and the ply ratio prediction value;

[0090] The optimization module 504 is used to optimize the predicted values ​​of the surface density and the predicted values ​​of the ply ratio by using the fitting curve function as the fitness function in the genetic algorithm to obtain the target surface density and the target ply ratio.

[0091] Based on the same general inventive concept, the present invention also protects a protective screen, in which the wave impedance gradient protective material is obtained by any of the above-mentioned wave impedance gradient protective material design methods based on neural networks.

[0092] Figure 6 It is a schematic diagram of the structure of the electronic device provided in this embodiment.

[0093] like Figure 6As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630 and a communication bus 640, wherein the processor 610, the communication interface 620 and the memory 630 communicate with each other through the communication bus 640. The processor 610 may call the logic instructions in the memory 630 to execute the wave impedance gradient protection material design method based on the neural network, the method comprising: performing simulation test on the wave impedance gradient material performance based on fluid dynamics to obtain material design parameters, the material design parameters including surface density, ply ratio and protection performance value; using the surface density, the ply ratio and the protection performance value as samples, constructing a neural network prediction model using a multi-layer perceptron, fitting the mapping relationship between the surface density, the ply ratio and the protection performance value, and obtaining a fitting curve function; inputting the target protection performance value into the neural network prediction model, outputting the surface density prediction value and the ply ratio prediction value; using the fitting curve function as the fitness function in the genetic algorithm to optimize the surface density prediction value and the ply ratio prediction value, and obtaining the target surface density and the target ply ratio.

[0094] In addition, the logic instructions in the above-mentioned memory 630 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0095] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the wave impedance gradient protective material design method based on a neural network provided by the above methods, the method including: simulating and testing the performance of the wave impedance gradient material based on fluid dynamics to obtain material design parameters, the material design parameters including surface density, ply ratio and protective performance values; using the surface density, the ply ratio and the protective performance values ​​as samples, constructing a neural network prediction model using a multi-layer perceptron, fitting the mapping relationship between the surface density, the ply ratio and the protective performance values, and obtaining a fitting curve function; inputting the target protective performance value into the neural network prediction model, and outputting the surface density prediction value and the ply ratio prediction value; using the fitting curve function as the fitness function in the genetic algorithm to optimize the surface density prediction value and the ply ratio prediction value, and obtaining the target surface density and the target ply ratio.

[0096] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the wave impedance gradient protective material design method based on a neural network provided by the above-mentioned methods, the method comprising: simulating and testing the performance of the wave impedance gradient material based on fluid dynamics to obtain material design parameters, the material design parameters including surface density, ply ratio and protective performance values; using the surface density, the ply ratio and the protective performance values ​​as samples, constructing a neural network prediction model using a multi-layer perceptron, fitting the mapping relationship between the surface density, the ply ratio and the protective performance values, and obtaining a fitting curve function; inputting the target protective performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value; using the fitting curve function as a fitness function in a genetic algorithm to optimize the surface density prediction value and the ply ratio prediction value, and obtaining a target surface density and a target ply ratio.

[0097] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0098] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable 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 server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for designing wave impedance gradient protective materials based on neural network, characterized in that: include: Based on fluid dynamics, the performance of the wave impedance gradient material is simulated and tested to obtain material design parameters, wherein the material design parameters include surface density, ply ratio and protective performance value; Taking the surface density, the ply ratio and the protective performance value as samples, a neural network prediction model is constructed using a multi-layer perceptron to fit the mapping relationship between the surface density, the ply ratio and the protective performance value to obtain a fitting curve function; Inputting a target protection performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value; The fitting curve function is used as a fitness function in a genetic algorithm to optimize the predicted value of the surface density and the predicted value of the ply ratio, so as to obtain a target surface density and a target ply ratio.

2. The method for designing wave impedance gradient protective materials based on neural network according to claim 1, characterized in that: The method of using the fitting curve function as a fitness function in a genetic algorithm to optimize the predicted value of the surface density and the predicted value of the ply ratio to obtain a target surface density and a target ply ratio includes: Using the predicted value of the surface density and the predicted value of the ply ratio as genes, genetic iteration is performed, and parameter configurations of the surface density and the ply ratio are generated as candidate solutions in each generation; Using the fitting curve function as a fitness function, determining the performance of the candidate solution; When the performance reaches the stop condition, the optimized target surface density and target ply ratio are output.

3. The method for designing wave impedance gradient protective materials based on neural network according to claim 2, characterized in that: Before the genetic iteration, the method further includes: The genes are efficiently represented within the framework of the genetic algorithm using binary coding.

4. The method for designing wave impedance gradient protective materials based on neural network according to claim 2, characterized in that: After determining the performance of the candidate solution, the method further includes: The candidate solution that meets the target requirements is selected as the parent, and then new candidate solutions are generated through continuous iterations of crossover and mutation operations.

5. The method for designing wave impedance gradient protective materials based on neural network according to claim 2, characterized in that: The stopping condition is: the number of iterations corresponding to the minimum mean square error.

6. The method for designing wave impedance gradient protective materials based on neural network according to claim 1, characterized in that: The neural network prediction model includes: an input layer, an output layer and a hidden layer; The input layer includes the surface density and ply ratio, and the output layer includes the protection performance value.

7. The method for designing wave impedance gradient protective materials based on neural network according to claim 6, characterized in that: The number of hidden layers is 3, and the number of neurons in each hidden layer is 10, 6 and 3.

8. The method for designing wave impedance gradient protective materials based on neural network according to any one of claims 1 to 6, characterized in that: The protection performance values ​​include critical projectile diameter; The wave impedance gradient material performance is simulated and tested based on fluid dynamics to obtain material design parameters, including: Set the surface density to multiple levels, set the ply ratio to multiple levels; The AUTODYN dynamic simulation program was used to carry out numerical simulation of the process of spherical projectiles hypervelocity impacting the PTFE / Al energetic wave impedance gradient material protection structure, and the critical projectile diameter corresponding to different surface density and ply ratio parameter settings was obtained.

9. A wave impedance gradient protective material design device based on neural network, characterized in that: include: A simulation module, used to simulate and test the performance of the wave impedance gradient material based on fluid dynamics to obtain material design parameters, wherein the material design parameters include surface density, ply ratio and protective performance value; A construction module is used to use the surface density, the ply ratio and the protective performance value as samples, construct a neural network prediction model using a multi-layer perceptron, fit the mapping relationship between the surface density, the ply ratio and the protective performance value, and obtain a fitting curve function; A prediction module, used for inputting a target protection performance value into the neural network prediction model, and outputting a surface density prediction value and a ply ratio prediction value; The optimization module is used to optimize the predicted value of the surface density and the predicted value of the ply ratio by using the fitting curve function as the fitness function in the genetic algorithm to obtain the target surface density and the target ply ratio.

10. A protective screen, characterized in that: The wave impedance gradient protective material in the protective screen is obtained by the wave impedance gradient protective material design method based on neural network as described in any one of claims 1 to 8.

Images