Electromagnetic wave absorbing / transmitting integrated frequency selection metamaterial intelligent design system

Abstract

The invention relates to the technical field of electromagnetic metamaterial design, in particular to an electromagnetic wave absorbing / transmitting integrated frequency selection metamaterial intelligent design system which comprises three artificial neural networks and a user interface, and the three artificial neural networks are a parameter prediction network SNN, a reverse prediction network INN and a forward prediction network FNN. According to the method, the structural parameters of the electromagnetic wave absorption / transmission integrated frequency selection metamaterial are directly predicted according to the input S21 parameters, the structural design of the electromagnetic wave absorption / transmission integrated frequency selection metamaterial is completed in a short time, the design efficiency is greatly improved, and real-time structural design according to the target performance can be achieved. By means of the simple graphical user interface, a designer can directly drag the points on the image to input the target performance expected to be achieved, the actual design result can be confirmed through the S11 and S21 parameter curves output by the interface in real time, and the design process is simplified.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic metamaterial design, and in particular to an intelligent design system for an electromagnetic wave absorbing / permeating integrated frequency selective metamaterial. Background Art

[0002] With the rapid development of science and technology, information warfare has gradually become the core of military struggle in the new era. Radar, as the most mature and widely used detection method, plays a vital role in modern warfare. Radar, capable of monitoring, tracking, and accurately locating targets in the air, at sea, and on the ground, is one of the primary threats facing multi-dimensional weapon platforms. Therefore, avoiding enemy radar detection is crucial to victory on the battlefield. Furthermore, these combat platforms are equipped with advanced radar systems for target detection and engagement, but their high-power electromagnetic radiation makes them vulnerable to enemy radar countermeasures and strikes.

[0003] To address this contradiction, stealth design for radar systems has become a hot topic of research. Stealth radar radomes offer an ideal solution, offering advantages such as requiring no major modifications to the radar itself, low cost, and high efficiency. However, traditional absorbing materials and stealth structures struggle to balance electromagnetic transmittance within the radar operating frequency band with stealth in other frequency bands. Recent advances in metamaterial technology have provided a new approach to addressing this challenge. Frequency-selective metamaterials, which combine electromagnetic wave absorption and transmission, offer dual-mode operation. They not only transmit electromagnetic waves within the radar operating frequency band but also absorb them in other frequency bands, significantly reducing echo reflections in multiple directions and improving the platform's overall stealth performance. However, due to their dual-mode operation, frequency-selective metamaterials combine traditional and emerging artificial electromagnetic structures, including frequency-selective metamaterials, circuit-simulated absorbers, and Salisbury screen absorbers, resulting in considerable complexity. Consequently, electromagnetic simulation and optimization methods employed in traditional metamaterial design methods struggle to flexibly adjust their performance in a short period of time, failing to meet the diverse performance requirements of radomes for different radar systems.

[0004] In recent years, artificial intelligence (AI), driven by the rapid development of computer hardware and the internet, has become a highly anticipated cutting-edge interdisciplinary technology, finding widespread application in a wide range of fields, including industry, healthcare, transportation, and even the military. Artificial neural networks, mimicking the operating principles of the biological brain, can rapidly generate inferences about complex problems based on existing data through scientific and rational training and learning. This opens up a new avenue for the rapid, real-time, and customized design of frequency-selective metamaterials that absorb and transmit electromagnetic waves. Summary of the Invention

[0005] In order to solve the problem that traditional metamaterial design methods are difficult to flexibly adjust the structure and performance of electromagnetic wave absorbing / transmitting integrated frequency selective metamaterials in a short period of time, the present invention provides an electromagnetic wave absorbing / transmitting integrated frequency selective metamaterial intelligent design system.

[0006] The present invention is achieved through the following technical solutions: an intelligent design system for frequency-selective metamaterials that absorb / transmit electromagnetic waves, including three artificial neural networks and a user interface. The three artificial neural networks are a parameter prediction network SNN, an inverse prediction network INN, and a forward prediction network FNN.

[0007] As a further improvement of the technical solution of the present invention, the parameter prediction network SNN is a fully connected neural network with the following structure: the number of input layer nodes Input (S) = 101-1601; the hidden layer depth Hide (SD) = 1-10 layers, the number of hidden layer nodes in each layer Hide (SW) = 100-600; the number of output layer nodes Output (S) = 101-1601; the function of the parameter prediction network SNN is to: input the electromagnetic response S of the metamaterial target 21 The parameter prediction output corresponds to the metamaterial target electromagnetic response S 11 parameter.

[0008] As a further improvement of the technical solution of the present invention, the inverse prediction network INN is a network comprising S 11 Input network, S 21 A fully connected neural network with input network and structural parameter prediction network; Among them, the S 11 The structure of the input network is: Input layer node number Input (IS 11 )=101-1601; hidden layer depth Hide(IS 11 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(IS 11 -W)=100-300; Output layer node number Output(IS 11 )=101-1601; said S 11 The role of the input network is to process the input metamaterial target electromagnetic response S 11 parameter; Among them, the S 21 The structure of the input network is: Input layer node number Input (IS 21 )=101-1601; hidden layer depth Hide(IS 21 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(IS 21 -W)=100-300; Output layer node number Output(IS 21)=101-1601; said S 21 The role of the input network is to process the input metamaterial target electromagnetic response S 21 parameter; The structure of the structural parameter prediction network is: Input layer node number Input (IP) = Output (IS 11 )+Output(IS 21 ); Hidden layer depth Hide (IPD) = 1-5 layers, number of hidden layer nodes Hide (IPW) = 100-300; number of output layer nodes Output (IP) = 5; The function of the structural parameter prediction network is: through S 11 and S 21 The input network takes the incoming data and predicts the structural parameters of the metamaterial that conforms to the input target electromagnetic response.

[0009] As a further improvement of the technical solution of the present invention, the forward prediction network FNN is a network including a structural parameter input network, S 11 Prediction Network and S 21 A fully connected neural network for prediction network; The structure of the structural parameter input network is as follows: the number of input layer nodes Input (FP) = 5; the hidden layer depth Hide (FPD) = 1-5 layers, the number of hidden layer nodes Hide (FPW) = 100-400; the number of output layer nodes Output (FP) = 100-400; the function of the structural parameter input network is to process the input metamaterial structural parameters; Among them, the S 11 The structure of the prediction network is: Input layer node number Input (FS 11 )=50-350; hidden layer depth Hide(FS 11 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(FS 11 -W)=100-300; Output layer node number Output(FS 11 )=101-1601; said S 11 The role of the prediction network is to predict the S that meets the input metamaterial structural parameters through the data input by the structural parameter input network. 11 parameter; Among them, the S 21 The structure of the prediction network is: Input layer node number Input (FS 21 )=Output(FP)-Input(FS 11 ); Hidden layer depth Hide(FS 21-D)=1-5 layers, the number of hidden layer nodes in each layer Hide(FS 21 -W)=100-300; Output layer node number Output(FS 21 )=101-1601; said S 21 The role of the prediction network is to predict the S that meets the input metamaterial structural parameters through the data input by the structural parameter input network. 21 parameter.

[0010] As a further improvement of the technical solution of the present invention, the user interface includes: User-Point=9 points that the user can freely drag, S generated by User-Point through interpolation 21 Input curve, structural parameter design results predicted by the inverse prediction network INN, S reconstructed by the forward prediction network FNN 11 and S 21 A real-time adjustable and movable graphical interface consisting of a parameter curve and reference lines fixed at -3dB and -10dB respectively.

[0011] As a further improvement to the technical solution of the present invention, a structural design is provided for a symmetrically structured electromagnetic wave absorbing / transmitting frequency selective metamaterial comprising periodically arranged n×n unit structures, where n is a positive integer; the unit structure comprises a frequency selective superstructure, a dielectric layer, and a lossy frequency selective superstructure stacked in sequence from bottom to top; The unit structure of the frequency selective superstructure is in the shape of a square, with a unit size a1=15.23-23.71 mm, an inner side length a2=9.10-14.33 mm, and a square resistance of 3-5 Ω / sq; the unit structure of the dielectric layer is in the shape of a square, with a side length a1 and a thickness h =3.20-9.16 mm; the unit structure of the lossy frequency selective superstructure is in the shape of a cross, with a unit size a1, an arm length b1=15.00-20.75 mm, an arm width b2=1.51-3.25 mm, and a square resistance of 4-25 Ω / sq.

[0012] Compared with the prior art, the present invention has the following advantages: (1) The present invention is based on multiple neural networks to form an intelligent design system, according to the input S 21 The structural parameters of the integrated electromagnetic wave absorbing / transmitting frequency selective metamaterial are directly predicted by parameters, and the structural design of the integrated electromagnetic wave absorbing / transmitting frequency selective metamaterial is completed in a short time. Compared with traditional methods, the design efficiency is greatly improved, and real-time structural design according to target performance can be achieved.

[0013] (2) The simple graphical user interface of the present invention enables designers to input the target performance expected by directly dragging points on the image, and can output the S 11 and S 21 The parameter curve confirms the actual design results, simplifies the design process, and greatly reduces the design threshold for the frequency-selective metamaterial structure that absorbs / transmits electromagnetic waves. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0016] Figure 1 This is a schematic diagram of the overall structure of the intelligent design system for the electromagnetic wave absorbing / transmitting integrated frequency-selective metamaterial of the present invention.

[0017] Figure 2 Schematic diagram of the SNN network structure in the intelligent design system of the electromagnetic wave absorbing / transmitting integrated frequency selection metamaterial of the present invention.

[0018] Figure 3 Schematic diagram of the INN network structure in the intelligent design system of the electromagnetic wave absorbing / transmitting integrated frequency selection metamaterial of the present invention.

[0019] Figure 4 Schematic diagram of the FNN network structure in the intelligent design system of the electromagnetic wave absorbing / transmitting integrated frequency selection metamaterial of the present invention.

[0020] Figure 5 This is the user interface of the intelligent design system for the electromagnetic wave absorption / transmission integrated frequency selection metamaterial of the present invention.

[0021] Figure 6 Schematic diagram of the unit structure of the integrated electromagnetic wave absorbing / transmitting frequency selective metamaterial designed by the present invention.

[0022] Figure 7 This is a front view of the frequency-selective metamaterial that can absorb and transmit electromagnetic waves and can be designed by the present invention.

[0023] Figure 8 This is a backside photo of the electromagnetic wave absorbing / transmitting integrated frequency selective metamaterial that can be designed by the present invention.

[0024] Figure 9The images of the design results on the user interface of Example 1, Example 2, Example 3 and Example 4 of the present invention are shown. DETAILED DESCRIPTION

[0025] In order to more clearly understand the above-mentioned objectives, features and advantages of the present invention, the scheme of the present invention will be further described below. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features therein can be combined with each other.

[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present invention, rather than all the embodiments.

[0027] The specific embodiments of the present invention are described in detail below. Example 1

[0028] An intelligent design system for frequency-selective metamaterials that absorb and transmit electromagnetic waves. This system is a metamaterial structure design system that includes three artificial neural networks (a parameter prediction network (SNN), an inverse prediction network (INN), and a forward prediction network (FNN)) and a user interface. Among them, the parameter prediction network SNN is a fully connected neural network with the following structure: the number of input layer nodes Input(S)=101; the hidden layer depth Hide(SD)=1 layer, the number of nodes Hide(SW)=600; the number of output layer nodes Output(S)=101.

[0029] The inverse prediction network INN is a network that includes S 11 Input network, S 21 Fully connected neural network of input network and structural parameter prediction network: (1) S 11 The structure of the input network is: Input layer node number Input (IS 11 )=101; hidden layer depth Hide(IS 11 -D)=5 layers, the number of nodes in each layer Hide(IS 11 -W)=100; Output layer node number Output(IS 11 )=101.

[0030] (2) S 21 The structure of the input network is: Input layer node number Input (IS 21 )=101; hidden layer depth Hide(IS 21 -D)=5 layers, the number of nodes in each layer Hide(IS 21-W)=100; Output layer node number Output(IS 21 )=101.

[0031] (3) The structure of the structural parameter prediction network is as follows: the number of input layer nodes Input (IP) = 202; the hidden layer depth Hide (IPD) = 5 layers, the number of nodes in each layer Hide (IPW) = 100; the number of output layer nodes Output (IP) = 5.

[0032] The forward prediction network FNN is a network that includes structural parameter input, S 11 Prediction Network and S 21 A fully connected neural network for prediction network.

[0033] (1) Structural parameters The structure of the input network is as follows: the number of input layer nodes Input (FP) = 5; the hidden layer depth Hide (FPD) = 1 layer, the number of nodes Hide (FPW) = 400; the number of output layer nodes Output (FP) = 400.

[0034] (2) S 11 The structure of the prediction network is: Input layer node number Input (FS 11 )=50; hidden layer depth Hide(FS 11 -D)=1 layer, number of nodes Hide(FS 11 -W)=300; Output layer node number Output(FS 11 )=101.

[0035] (3) S 21 The structure of the prediction network is: Input layer node number Input (FS 21 )=350; hidden layer depth Hide(FS 21 -D)=1 layer, number of nodes Hide(FS 21 -W) = 300; Output layer node number Output(FS 21 )=101.

[0036] Depend on Figure 9 The user interface shown in (a) shows that the designed electromagnetic wave absorbing / transmitting integrated lightweight frequency selective metamaterial has the following structural parameters: a1=20.45 mm, a2=11.74 mm, b1=19 mm, b2=3.00 mm, and h=8.30 mm. The absorption bands (S) of 2.98-4.55, 6.15-8.00, and 16.72-18 GHz are achieved. 11 ≤-10 dB and S 21≤-10 dB) and the 10.06-11.41 GHz transmission band (S 11 ≤-10 dB and S 21 ≥-3 dB). Example 2

[0037] The basic structure of the system is the same as that of Example 1, where the structure of the parameter prediction network SNN is: the number of input layer nodes Input (S) = 1601; the hidden layer depth Hide (SD) = 10 layers, the number of nodes Hide (SW) = 100; the number of output layer nodes Output (S) = 1601.

[0038] The structure of its inverse prediction network INN is: (1) S 11 The structure of the input network is: Input layer node number Input (IS 11 )=1601; hidden layer depth Hide(IS 11 -D)=1 layer, number of nodes Hide(IS 11 -W)=300; Output layer node number Output(IS 11 )=1601.

[0039] (2) S 21 The structure of the input network is: Input layer node number Input (IS 21 ) = 1601; hidden layer depth Hide(IS 21 -D)=1 layer, number of nodes Hide(IS 21 -W)=300; Output layer node number Output(IS 21 )=1601.

[0040] (3) The structure of the structural parameter prediction network is as follows: the number of input layer nodes Input (IP) = 3202; the hidden layer depth Hide (IPD) = 1 layer, the number of nodes Hide (IPW) = 300; the number of output layer nodes Output (IP) = 5.

[0041] The structure of the forward prediction network FNN is: (1) Structural parameters The structure of the input network is as follows: the number of input layer nodes Input (FP) = 5; the hidden layer depth Hide (FPD) = 5 layers, the number of nodes Hide (FPW) = 100; the number of output layer nodes Output (FP) = 100.

[0042] (2) S 11 The structure of the prediction network is: Input layer node number Input (FS 11)=350; hidden layer depth Hide(FS 11 -D)=5 layers, number of nodes Hide(FS 11 -W)=100; Output layer node number Output(FS 11 )=1601.

[0043] (3) S 21 The structure of the prediction network is: Input layer node number Input (FS 21 )=50; hidden layer depth Hide(FS 21 -D)=5 layers, number of nodes Hide(FS 21 -W)=100; Output layer node number Output(FS 21 )=1601.

[0044] Depend on Figure 9 The user interface shown in (b) shows that the designed electromagnetic wave absorbing / transmitting integrated lightweight frequency selective metamaterial has the following structural parameters: a1=15.32 mm, a2=9.04 mm, b=a1=15.32 mm, b1=14.99 mm, b2=1.40 mm, h=3.12 mm. The absorption band (S) of 4.09-10.04 GHz was achieved. 11 ≤-10 dB and S 21 ≤-10 dB) and a transmission band of 12.80-15.00 GHz (S 11 ≤-10 dB and S 21 ≥-3 dB). Example 3

[0045] The network structure in the system is the same as that in Example 1. Another type of electromagnetic wave absorbing / permeable integrated lightweight frequency selective metamaterial is designed by operating the user interface.

[0046] Depend on Figure 9 The user interface shown in (c) shows that the designed electromagnetic wave absorbing / transmitting integrated lightweight frequency selective metamaterial has the following structural parameters: a1=23.71 mm, a2=14.33 mm, b1=20.75 mm, b2=3.25 mm, h=9.16 mm. The absorption bands (S) of 3.17-6.88 and 15.84-18 GHz are achieved. 11 ≤-10 dB and S 21 ≤-10 dB) and a transmission band of 8.64-10.15 (S 11 ≤-10 dB and S 21 ≥-3 dB). Example 4

[0047] The network structure in the system is the same as that in Example 2. Another type of electromagnetic wave absorbing / permeable integrated lightweight frequency selective metamaterial is designed by operating the user interface.

[0048] Depend on Figure 9 The user interface shown in (d) shows that the designed electromagnetic wave absorbing / transmitting integrated lightweight frequency selective metamaterial has the following structural parameters: a1=15.15 mm, a2=9.19 mm, b1=14.72 mm, b2=1.23 mm, and h=3.65 mm. The absorption band (S) of 3.95-9.75 GHz is achieved. 11 ≤-10 dB and S 21 ≤-10 dB) and the 13.80-14.32 GHz transmission band (S 11 ≤-10 dB and S 21 ≥-3 dB).

[0049] The above description is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Although detailed descriptions have been made with reference to the aforementioned embodiments, those skilled in the art should understand that they may still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents; and such modifications or replacements do not deviate from the essence of the corresponding technical solutions within the scope of the technical solutions of the embodiments, and they should all be included in the scope of protection of the claims.

Claims

1. An intelligent design system for frequency-selective metamaterials that absorbs and transmits electromagnetic waves, characterized in that: The system comprises three artificial neural networks and a user interface, wherein the three artificial neural networks are a parameter prediction network SNN, an inverse prediction network INN and a forward prediction network FNN.

2. The intelligent design system for electromagnetic wave absorption / transmission frequency selective metamaterial according to claim 1, characterized in that: The parameter prediction network SNN is a fully connected neural network with the following structure: the number of input layer nodes Input (S) = 101-1601; the hidden layer depth Hide (SD) = 1-10 layers, the number of hidden layer nodes Hide (SW) = 100-600; the number of output layer nodes Output (S) = 101-1601; the function of the parameter prediction network SNN is to input the electromagnetic response S of the metamaterial target. 21 The parameter prediction output corresponds to the metamaterial target electromagnetic response S 11 parameter.

3. The intelligent design system for electromagnetic wave absorption / transmission integrated frequency selective metamaterial according to claim 1, characterized in that: The inverse prediction network INN is a network comprising S 11 Input network, S 21 A fully connected neural network with input network and structural parameter prediction network; Among them, the S 11 The structure of the input network is: Input layer node number Input (IS 11 )=101-1601; hidden layer depth Hide(IS 11 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(IS 11 -W)=100-300; Output layer node number Output(IS 11 )=101-1601; said S 11 The role of the input network is to process the input metamaterial target electromagnetic response S 11 parameter; Among them, the S 21 The structure of the input network is: Input layer node number Input (IS 21 )=101-1601; hidden layer depth Hide(IS 21 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(IS 21 -W)=100-300; Output layer node number Output(IS 21 )=101-1601; said S 21 The role of the input network is to process the input metamaterial target electromagnetic response S 21 parameter; The structure of the structural parameter prediction network is: Input layer node number Input (IP) = Output (IS 11 )+Output(IS 21 ); Hidden layer depth Hide (IPD) = 1-5 layers, number of hidden layer nodes Hide (IPW) = 100-300; number of output layer nodes Output (IP) = 5; The function of the structural parameter prediction network is: through S 11 and S 21 The input network takes the incoming data and predicts the structural parameters of the metamaterial that conforms to the input target electromagnetic response.

4. The intelligent design system for electromagnetic wave absorption / transmission integrated frequency selective metamaterial according to claim 1, characterized in that: The forward prediction network FNN is a network including structural parameter input, S 11 Prediction Network and S 21 A fully connected neural network for prediction network; The structure of the structural parameter input network is as follows: the number of input layer nodes Input (FP) = 5; the hidden layer depth Hide (FPD) = 1-5 layers, the number of hidden layer nodes Hide (FPW) = 100-400; the number of output layer nodes Output (FP) = 100-400; the function of the structural parameter input network is to process the input metamaterial structural parameters; Among them, the S 11 The structure of the prediction network is: Input layer node number Input (FS 11 )=50-350; hidden layer depth Hide(FS 11 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(FS 11 -W)=100-300; Output layer node number Output(FS 11 )=101-1601; said S 11 The role of the prediction network is to predict the S that meets the input metamaterial structural parameters through the data input by the structural parameter input network. 11 parameter; Among them, the S 21 The structure of the prediction network is: Input layer node number Input (FS 21 )=Output(FP)- Input(FS 11 ); Hidden layer depth Hide(FS 21 -D)=1-5 layers, the number of hidden layer nodes in each layer Hide(FS 21 -W)=100-300; Output layer node number Output(FS 21 )=101-1601; said S 21 The role of the prediction network is to predict the S that meets the input metamaterial structural parameters through the data input by the structural parameter input network. 21 parameter.

5. The intelligent design system for electromagnetic wave absorption / transmission integrated frequency selective metamaterial according to claim 1, characterized in that: The user interface includes: User-Point = 9 points that the user can freely drag, S generated by interpolation of User-Point 21 Input curve, structural parameter design results predicted by the inverse prediction network INN, S reconstructed by the forward prediction network FNN 11 and S 21 A real-time adjustable and movable graphical interface consisting of a parameter curve and reference lines fixed at -3 dB and -10 dB respectively.

6. The intelligent design system for frequency selective metamaterials capable of absorbing and transmitting electromagnetic waves according to any one of claims 1 to 5, characterized in that: The system is capable of structurally designing a symmetrical structure of an electromagnetic wave absorbing / transmitting integrated frequency selective metamaterial comprising a periodically arranged n×n unit structure, where n is a positive integer; the unit structure comprises a frequency selective metastructure, a dielectric layer, and a lossy frequency selective metastructure stacked in sequence from bottom to top; The unit structure of the frequency selective superstructure is in the shape of a square, with a unit size a1=15.23-23.71 mm, an inner side length a2=9.10-14.33 mm, and a square resistance of 3-5 Ω / sq; the unit structure of the dielectric layer is in the shape of a square, with a side length a1 and a thickness h =3.20-9.16 mm; the unit structure of the lossy frequency selective superstructure is in the shape of a cross, with a unit size a1, an arm length b1=15.00-20.75 mm, an arm width b2 = 1.51-3.25 mm, and a square resistance of 4-25 Ω / sq.

Images