Rapid design and preparation method of wave-absorbing material based on machine learning
Through the combination of machine learning and simulation optimization, wide-frequency absorbing materials are quickly designed and prepared, which solves the problem of long development cycle of absorbing materials and realizes the rapid preparation of high-performance absorbing materials.
Patent Information
- Application Number
- CN202510476115.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-16
- Publication Date
- 2025-08-19
AI Technical Summary
The design and preparation process of existing absorbent materials have long cycles, making it difficult to quickly obtain high-performance wide-frequency absorbent materials.
Using a combination of machine learning and simulation optimization, the model is trained through a small sample database to quickly obtain electromagnetic parameters, and multi-layer absorbing materials are prepared through impedance gradient design and process inversion.
The development cycle of wave absorbing materials is greatly shortened, and an impedance gradient graphene-based composite wave absorbing material with wide-band electromagnetic wave absorption performance is obtained.
Smart Images
Figure CN120509280A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a rapid design of an absorbing material based on machine learning and a preparation method thereof, belonging to the technical field of absorbing material preparation. Background Art
[0002] With the increasing development of smart and wireless technologies, higher demands are being placed on absorbing materials, among which electromagnetic parameters are the key to designing high-performance absorbing materials. However, the current design and preparation of absorbing materials often adopt the traditional trial-and-error method, which uses orthogonal experiments to screen and design, resulting in a long research cycle. This has seriously restricted the development of absorbing materials, especially in the face of emerging new materials. With the improvement of machine computing power, machine-assisted design of absorbing materials and guidance of actual preparation are effective ways to solve current problems. Based on machine learning, through the optimization of machine learning models, a large number of electromagnetic parameters can be accurately predicted within minutes, providing a large amount of data for the design of absorbing materials. Through simulation optimization, high-performance absorbing materials with broadband absorption can be quickly obtained.
[0003] Therefore, the present invention provides a method for rapid design and preparation of absorbing materials based on machine learning to address the long research cycle of existing absorbing materials. Compared with traditional experimental methods, this method, which combines machine learning and simulation using small sample data sets, can accurately design and prepare high-performance absorbing materials, greatly shortening the research cycle of absorbing materials. Summary of the Invention
[0004] The purpose of the present invention is to provide a rapid design of absorbing materials based on machine learning and a preparation method thereof, which greatly shortens the development cycle of absorbing materials, and the obtained impedance gradient graphene-based composite absorbing material has broadband electromagnetic wave absorption performance.
[0005] The present invention also provides an impedance gradient graphene-based composite absorbing material prepared by the above-mentioned preparation method.
[0006] To achieve the above objectives, the technical solutions adopted by the rapid design of absorbing materials based on machine learning and the preparation method thereof of the present invention are:
[0007] A method for rapidly designing and preparing absorbing materials based on machine learning includes the following steps: first, obtaining a small sample database through experimental preparation; second, establishing a machine learning model and training the model using the small sample database to improve its accuracy; then, by inputting different process parameters to obtain corresponding electromagnetic parameters, optimizing the absorbing material's absorbing properties using impedance gradient design; and based on the optimization results, inverting the process parameters to prepare a multilayered absorbing material.
[0008] The present invention's rapid design and preparation method for absorbing materials based on machine learning combines machine learning with simulation optimization to rapidly obtain a large number of electromagnetic parameters. Furthermore, through simulation-based design optimization, broadband absorbing performance is achieved. Based on the design results, process inversion of the required electromagnetic parameters is then performed to guide actual preparation.
[0009] Preferably, the concentration of the graphene oxide slurry is 1-18 mg / g, preferably 4-10 mg / g.
[0010] Preferably, the mass ratio of the graphene oxide to the dielectric fiber membrane is 1:1-1:5, preferably 1:1-1:4.
[0011] Preferably, the machine learning algorithm includes random forest, decision tree, neural network, etc., preferably a random forest algorithm.
[0012] Preferably, the establishment of a small sample database includes the following steps: first, graphene, silica fiber membrane and barium titanate fiber membrane are laid and impregnated according to a certain mass ratio, and then freeze-dried, and the freeze-dried samples are annealed at different temperatures, and the obtained samples are tested for electromagnetic parameters to obtain a small sample database.
[0013] Preferably, the process parameters include graphene concentration, type and content of the composite, and annealing temperature, preferably annealing temperature.
[0014] Preferably, the establishment and training of a machine learning model includes the following steps: First, parameterize the process characteristics that affect electromagnetic parameters and organize them into a unified table format to facilitate model training. Then, select an appropriate machine learning algorithm based on the parameter type and adjust the model's hyperparameters using Bayesian optimization to improve model performance. Finally, evaluate the model using validation data and determine the model and its corresponding hyperparameters.
[0015] Preferably, the liquid nitrogen freezing molding is to introduce liquid nitrogen into the mold, and the temperature is controlled by a temperature control valve. The freezing time gradually increases with the thickness of the composite material, and the freezing time is 1-3 hours, preferably 1 hour.
[0016] Preferably, the freeze-drying treatment time gradually increases with the increase of the thickness of the composite material, and the freeze-drying time is 3-6 days, preferably 6 days.
[0017] Preferably, the design of the multi-layer structure includes the following steps: first, based on the trained machine learning model, the electromagnetic parameter data is expanded several times within a few minutes, and then based on these electromagnetic parameters, simulation software is used to perform simulation design optimization to obtain a broadband absorption performance combination.
[0018] Preferably, the inversion of process parameters based on the optimization results includes the following steps: first, screening the required electromagnetic parameters based on the results of the simulation design optimization, and then inverting the process parameters based on the prediction model to guide the preparation of actual samples.
[0019] The technical solutions adopted by the rapid design of absorbing materials based on machine learning and the preparation method thereof of the present invention are:
[0020] An impedance gradient graphene-based composite absorbing material prepared using the above-mentioned machine learning-based rapid design of absorbing materials and the preparation method thereof.
[0021] The present invention's rapid design and preparation method for absorbing materials based on machine learning combines machine learning with simulation optimization to rapidly obtain a large number of electromagnetic parameters. Furthermore, through simulation-based design optimization, broadband absorbing performance is achieved. Based on the design results, process inversion of the required electromagnetic parameters is then performed to guide actual preparation. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 A distribution diagram of a machine learning model and a small sample database in an embodiment of the present invention;
[0023] Figure 2 An electromagnetic parameter map predicted by a machine learning model in an embodiment of the present invention;
[0024] Figure 3 The diagram shows the simulation guidance and actual performance of the absorbing material prepared in Example 3 of the present invention. DETAILED DESCRIPTION
[0025] The technical solution of the present invention is further described below in conjunction with specific implementation methods.
[0026] Example 1
[0027] The rapid design and preparation method of absorbing materials based on machine learning in this embodiment includes the following steps:
[0028] 1) Graphene was mixed with silicon dioxide film and barium titanate film in a certain mass ratio, and then the sample was freeze-dried. Then, different annealing temperatures were applied to it, and its electromagnetic parameters were tested to obtain a small sample database.
[0029] 2) The small sample database is divided into training and test sets. The process characteristics are parameterized and organized into a unified table format to facilitate the training of the machine learning model. Based on Bayesian optimization, the model's hyperparameters are adjusted to improve the model's accuracy. The trained machine learning model is then tested on the validation set to determine its accuracy.
[0030] 3) Based on the trained machine learning model, the electromagnetic parameter data is expanded several times in a short period of time by inputting process parameter characteristics. Then, based on the predicted results, the electromagnetic parameters of each layer are screened using simulation software using impedance gradient design to determine the impedance and thickness of each layer. In this embodiment, the structure is divided into three layers with a total thickness of 5mm. The design achieves broadband absorption performance.
[0031] 4) Based on the design results, the required electromagnetic parameters are screened and inverted in the machine learning prediction model to obtain the process parameters required to prepare the corresponding electromagnetic parameters, including graphene concentration, composite type and content, and annealing temperature.
[0032] 5) Based on the process parameters obtained in step 4), each required layer of sample is prepared separately, and according to the design in step 3), a multi-layer impedance gradient absorbing material is prepared, and polyurethane is used as a binder between the layers.
[0033] Example 2
[0034] The rapid design and preparation method of absorbing materials based on machine learning in this embodiment includes the following steps:
[0035] 1) Graphene was mixed with silicon dioxide film and barium titanate film in a certain mass ratio, and then the sample was freeze-dried. Then, different annealing temperatures were applied to it, and its electromagnetic parameters were tested to obtain a small sample database.
[0036] 2) The small sample database is divided into training and test sets. The process characteristics are parameterized and organized into a unified table format to facilitate the training of the machine learning model. Based on Bayesian optimization, the model's hyperparameters are adjusted to improve the model's accuracy. The trained machine learning model is then tested on the validation set to determine its accuracy.
[0037] 3) Based on the trained machine learning model, by inputting process parameter characteristics, the electromagnetic parameter data is expanded several times in a short period of time. Then, based on the predicted results, the electromagnetic parameters of each layer are screened using simulation software using impedance gradient design to determine the impedance and thickness of each layer. In this embodiment, the structure is divided into three layers with a total thickness of 10mm. The design achieves broadband absorption performance.
[0038] 4) Based on the design results, the required electromagnetic parameters are screened and inverted in the machine learning prediction model to obtain the process parameters required to prepare the corresponding electromagnetic parameters, including graphene concentration, composite type and content, and annealing temperature.
[0039] 5) Based on the process parameters obtained in step 4), each required layer of sample is prepared separately, and according to the design in step 3), a multi-layer impedance gradient absorbing material is prepared, and polyurethane is used as a binder between the layers.
[0040] Example 3
[0041] The rapid design and preparation method of absorbing materials based on machine learning in this embodiment includes the following steps:
[0042] 1) Graphene was mixed with silicon dioxide film and barium titanate film in a certain mass ratio, and then the sample was freeze-dried. Then, different annealing temperatures were applied to it, and its electromagnetic parameters were tested to obtain a small sample database.
[0043] 2) The small sample database is divided into training and test sets. The process characteristics are parameterized and organized into a unified table format to facilitate the training of the machine learning model. Based on Bayesian optimization, the model's hyperparameters are adjusted to improve the model's accuracy. The trained machine learning model is then tested on the validation set to determine its accuracy.
[0044] 3) Based on the trained machine learning model, the electromagnetic parameter data is expanded several times in a short period of time by inputting process parameter characteristics. Then, based on the predicted results, the electromagnetic parameters of each layer are screened using simulation software using impedance gradient design to determine the impedance and thickness of each layer. In this embodiment, the structure is divided into three layers with a total thickness of 20mm. The design achieves broadband absorption performance.
[0045] 4) Based on the design results, the required electromagnetic parameters are screened and inverted in the machine learning prediction model to obtain the process parameters required to prepare the corresponding electromagnetic parameters, including graphene concentration, composite type and content, and annealing temperature.
[0046] 5) Based on the process parameters obtained in step 4), each required layer of sample is prepared separately, and according to the design in step 3), a multi-layer impedance gradient absorbing material is prepared, and polyurethane is used as a binder between the layers.
[0047] The machine learning model and small sample database used in step 2) of the embodiment are shown in FIG. Figure 1 ,Depend on Figure 1 It can be seen that the process parameters in the small sample database are distributed relatively evenly; the prediction results of the electromagnetic parameters by the machine learning model in step 3) of the embodiment are shown in FIG. Figure 2 The absorbing performance of the multilayer absorbing material prepared in Example 3 was tested. The simulation results and the actual test results are shown in Figure 2. Figure 3 shown.
[0048] Experimental example
[0049] The electromagnetic wave absorption performance of the impedance gradient graphene-based composite absorbing material prepared in Example 3 was tested.
[0050] The test method is to use an Agilent N5244A vector network analyzer to test the reflection loss of the impedance gradient graphene composite absorbing material obtained by the bow frame test method. The electromagnetic wave band is 2-18GHz.
[0051] As can be seen from the above examples, the present invention provides an impedance-gradient graphene-based composite absorbing material produced using the aforementioned machine learning-based rapid absorbing material design and preparation method. This machine learning-based rapid absorbing material design and preparation method combines machine learning with simulation optimization to rapidly obtain a large number of electromagnetic parameters and, through simulation-based design optimization, achieve broadband absorbing performance. Based on the design results, process inversion of the desired electromagnetic parameters guides actual production. This significantly shortens the research cycle for absorbing materials.
[0052] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.
Claims
1. A rapid design and preparation method of absorbing materials based on machine learning, characterized by: The following steps are involved: First, a small sample database was obtained through experimental preparation. A machine learning model was then established and trained using the small sample database to improve its accuracy. Subsequently, different process parameters were input to obtain corresponding electromagnetic parameters. An impedance gradient design was used to optimize the absorbing material's absorbing properties. Based on the optimization results, the process parameters were inverted to produce a multilayer absorbing material.
2. The rapid design and preparation method of the absorbing material according to claim 1, characterized in that: The small sample database is a graphene-based composite absorbing material, and the process factors affecting the electromagnetic parameters are graphene concentration, type and content of the composite, and annealing temperature.
3. The rapid design and preparation method of absorbing materials according to claim 1, characterized in that: The machine learning models are random forest and neural network.
4. The rapid design and preparation method of the absorbing material according to claim 1, characterized in that: The establishment of a small sample database includes the following steps: first, graphene, silica fiber membrane and barium titanate fiber membrane are laid and impregnated according to a certain mass ratio, and then freeze-dried. The freeze-dried samples are annealed at different temperatures, and the obtained samples are tested for electromagnetic parameters to obtain a small sample database.
5. The method for rapid design and preparation of an absorbing material according to any one of claims 1 to 4, characterized in that: The creation and training of a machine learning model involves the following steps: First, the process characteristics that influence electromagnetic parameters are parameterized and organized into a unified table format to facilitate model training. Then, an appropriate machine learning algorithm is selected based on the parameter type, and the model's hyperparameters are adjusted using Bayesian optimization to improve model performance. Finally, the model's performance is evaluated using validation data, and the corresponding hyperparameters are determined.
6. The rapid design and preparation method of absorbing materials according to claim 1, characterized in that: The design of the multi-layer structure includes the following steps: first, based on the trained machine learning model, the electromagnetic parameter data is expanded several times within minutes; then, based on these electromagnetic parameters, simulation software is used to perform simulation design optimization to obtain a broadband absorption performance combination.
7. The rapid design and preparation method of absorbing materials according to claim 1, characterized in that: The inversion of process parameters based on the optimization results includes the following steps: first, screening the required electromagnetic parameters according to the results of the simulation design optimization, and then inverting the process parameters according to the prediction model to guide the preparation of actual samples.
8. An impedance gradient graphene composite absorbing material produced by the rapid design and preparation method of absorbing materials according to any one of claims 1 to 7.