Deep learning-based millimeter wave and terahertz amplifier design method and related device

Abstract

The invention discloses a millimeter wave and terahertz amplifier design method based on deep learning and a related device, and belongs to the field of millimeter wave and terahertz integrated circuit design, the amplifier design method comprises the following steps: pixelating a plurality of arbitrary embedded compensation network structures and constructing a corresponding two-dimensional matrix; inputting the two-dimensional matrix into a network parameter prediction model to obtain a corresponding embedded network S parameter; and after the relationship between the S parameter of the embedded network and the Y parameter of the amplifier is constructed, global optimization is carried out on the S parameter of the embedded network to obtain an optimal embedded compensation network structure, and design is carried out according to the optimal embedded compensation network structure to obtain the amplifier. According to the invention, the problem that the solution space of the embedded compensation network is limited and is highly related to the working frequency can be solved.

Description

Technical Field

[0001] The present invention belongs to the field of millimeter wave and terahertz integrated circuit design, and specifically relates to a millimeter wave and terahertz amplifier design method and related devices based on deep learning. Background Art

[0002] In the sub-terahertz frequency band, the limited gain of CMOS devices has become a key factor restricting the design of CMOS sub-terahertz amplifiers. Due to the high frequency characteristics of the sub-terahertz frequency band, the parasitic parameters inside CMOS devices, such as resistance, capacitance and inductance, have a particularly significant impact on the deterioration of device gain. These parasitic parameters will not only reduce the gain of the device, but also introduce additional phase delay and noise, thus affecting the overall performance of the amplifier.

[0003] At present, the technical means of device physical property analysis and layout optimization are usually used to reduce the influence of parasitic parameters. Through in-depth analysis of the physical properties of CMOS devices, we can understand the source and influencing factors of parasitic parameters, and then reduce the value of parasitic parameters by optimizing the device structure and layout. However, although this method can improve the gain of the device to a certain extent, it is still limited by the physical properties of the device itself, and it is difficult to achieve a significant performance improvement.

[0004] In order to improve the gain performance of CMOS sub-terahertz amplifiers, researchers introduced embedded compensation theory and methods. The embedded compensation network technology introduces additional compensation elements into the network to offset the influence of parasitic parameters, thereby restoring and improving the gain of the device. The patent with the authorization publication number CN114024514B discloses a neutralized amplifier structure with a series lossy embedded network. Under the two-port network composed of a traditional amplifier, a series lossy embedded network is introduced to improve the unidirectional power gain under the preset conditions, and based on this, the Y / Z type LLREN is used to further improve the maximum available gain, so that the power gain of the amplifier is effectively improved. However, since the embedded compensation network technology is based on the network parameter theoretical analysis method, and the embedded network usually adopts a transmission line structure, the solution space is limited. At the same time, the transmission line network parameters are strongly correlated with the operating frequency, resulting in the performance of the embedded compensation network being limited in the sub-terahertz frequency band, making it difficult to achieve the requirements of broadband and high gain at the same time. Summary of the invention

[0005] The purpose of the present invention is to provide a millimeter wave and terahertz amplifier design method and related devices based on deep learning, so as to solve the problem that the solution space of the embedded compensation network is limited and strongly related to the operating frequency, so as to simultaneously achieve broadband and high gain of the amplifier.

[0006] In order to solve the above technical problems, the embodiment of the present application provides a millimeter wave and terahertz amplifier design method based on deep learning, comprising the following steps: Pixelate multiple arbitrary embedded compensation network structures and construct corresponding two-dimensional matrices; Inputting the two-dimensional matrix into a network parameter prediction model to obtain corresponding embedded network S parameters; After constructing the relationship between the embedded network S parameters and the Y parameters of the amplifier, the embedded network S parameters are globally optimized to obtain an optimal embedded compensation network structure, and the amplifier is designed according to the optimal embedded compensation network structure.

[0007] In some embodiments, the process of establishing the network parameter prediction model is: performing electromagnetic simulation on multiple arbitrary embedded compensation network structures after pixelation to obtain a two-dimensional matrix and corresponding embedded network S parameters, using the two-dimensional matrix and embedded network S parameters as the input and output of a convolutional neural network (CNN), respectively, and training and verifying the convolutional neural network to obtain the network parameter prediction model.

[0008] In some embodiments, the convolutional neural network includes n convolutional layers and m fully connected layers, where n is the number of convolutional layers and m is the number of fully connected layers. Each convolutional layer is sequentially connected to a normalization layer and an activation layer, and the first m-1 levels of fully connected layers are sequentially connected to a normalization layer, an activation layer, and a random inactivation layer, and the activation layer uses a Leaky ReLU function.

[0009] In some implementations, a genetic algorithm is used to globally optimize the embedded network S parameters to obtain an optimal embedded compensation network structure, specifically comprising the following steps: The two-dimensional matrices corresponding to the plurality of arbitrary embedded compensation network structures are used as a parent generation, and the mean square error between the embedded network S parameters and the expected S parameters is used as a cost function; Calculate the error between the embedded network S parameter and the expected S parameter according to the cost function, and select the two-dimensional matrices corresponding to several minimum errors as the first subset; Randomly select a number of the two-dimensional matrices in the parent generation and select two two-dimensional matrices corresponding to the minimum errors to cross-generate a child generation, repeat this step until the number of the child generations is equal to the number of the parent generations, and the child generations constitute a second subset; The first subset and the second subset are fused as a new population to iterate the parent generation, and the above steps are repeated until a preset number of iterations is reached to obtain an optimal embedding compensation network structure.

[0010] In some implementations, the step of constructing the relationship between the embedded network S parameters and the amplifier Y parameters specifically includes: The Y parameters of the amplifier are composed of the Y parameters of the embedded compensation network structure and the Y parameters of the transistor; Obtaining Y parameters of the embedded compensation network structure according to the embedded network S parameters and the unit matrix; According to the Y parameters and the unit matrix of the embedded compensation network structure, and in combination with the Y parameters of the transistor, the relationship between the embedded network S parameters and the Y parameters of the amplifier is obtained.

[0011] In some embodiments, after the step of performing electromagnetic simulation on the pixelated multiple arbitrary embedded compensation network structures to obtain a two-dimensional matrix and corresponding embedded network S parameters, it also includes: using a generative adversarial network to expand the two-dimensional matrix and its corresponding embedded network S parameters.

[0012] In order to solve the above technical problems, the embodiment of the present application provides a millimeter wave and terahertz amplifier design system based on deep learning, including: An embedded compensation network pixelation module, used to pixelate a plurality of arbitrary embedded compensation network structures and construct a corresponding two-dimensional matrix; An embedded network S parameter prediction module inputs the two-dimensional matrix into a network parameter prediction model to obtain corresponding embedded network S parameters; The amplifier design module constructs the relationship between the embedded network S parameters and the amplifier Y parameters, performs global optimization on the embedded network S parameters to obtain an optimal embedded compensation network structure, and designs an amplifier based on the optimal embedded compensation network structure.

[0013] In order to solve the above technical problems, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable in the processor, wherein the processor implements the steps of the deep learning-based millimeter wave and terahertz amplifier design method when executing the computer program.

[0014] In order to solve the above technical problems, an embodiment of the present application provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the millimeter wave and terahertz amplifier design method based on deep learning are implemented.

[0015] In order to solve the above technical problems, an embodiment of the present application provides a computer program product, which includes a computer program, and is characterized in that when the computer program is executed by a processor, the steps of the millimeter wave and terahertz amplifier design method based on deep learning are implemented in the claims.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a millimeter wave and terahertz amplifier design method based on deep learning, which constructs a two-dimensional matrix corresponding to any embedded compensation network after pixel decomposition by pixel decomposing any embedded compensation network, and then inputs the two-dimensional matrix into a network parameter prediction model to obtain the embedded network S parameters, and after constructing the relationship between the embedded network S parameters and the Y parameters of the amplifier, globally optimizes the embedded network S parameters to obtain the optimal embedded compensation network structure, and finally obtains the amplifier according to the optimal embedded compensation network design. The method has a large solution space, and after obtaining the optimal embedded compensation network structure, a broadband and high-gain amplifier can be obtained, which solves the problem that the solution space of the embedded compensation network is limited and is strongly related to the operating frequency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of an amplifier topology structure using an embedded compensation network in an embodiment of the present invention; Figure 2 An overall scheme diagram of a millimeter wave and terahertz amplifier design method based on deep learning provided in an embodiment of the present invention; Figure 3 A flowchart of a millimeter wave and terahertz amplifier design method based on deep learning provided in an embodiment of the present invention; Figure 4 A schematic diagram of the structure of a millimeter wave and terahertz amplifier design system based on deep learning provided in an embodiment of the present invention; Figure 5 A schematic diagram of a process for optimizing the S parameters of an embedded network using a genetic algorithm according to an embodiment of the present invention; Figure 6 A simulation diagram of a broadband amplifier design provided in an embodiment of the present invention, wherein (a) is a circuit schematic diagram of the broadband amplifier; and (b) is a simulation result of the broadband amplifier. DETAILED DESCRIPTION

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used in the specification of the application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0019] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0021] like Figure 2 and Figure 3 As shown, this embodiment provides a millimeter wave and terahertz amplifier design method based on deep learning, comprising the following steps: S1, pixelate multiple arbitrary embedded compensation network structures and construct corresponding two-dimensional matrices. Specifically, use the idea of ​​image pixelation to randomly generate 400,000 20×20 two-dimensional matrices for arbitrary embedded compensation network structures using MATLAB.

[0022] S2, inputting the two-dimensional matrix into the network parameter prediction model to obtain the corresponding embedded network S parameters. Specifically, on the PyTorch platform, the two-dimensional matrix and the embedded network S parameters generated by the electromagnetic simulation are used as the input and output of the convolutional neural network, respectively, and the convolutional neural network is trained and verified to obtain the network parameter prediction model, and the function fitting relationship between the embedded network S parameters and the embedded compensation network structure is constructed through the network parameter prediction model.

[0023] The convolutional neural network architecture includes 12 convolutional layers and 5 fully connected layers. Each level of convolutional layer is connected to a normalization layer and an activation layer in sequence. The first four levels of fully connected layers are connected to a normalization layer, an activation layer, and a random dropout layer in sequence. The activation layer uses a Leaky ReLU function to prevent the gradient from disappearing to ensure a good learning effect of the convolutional neural network. The random dropout layer can reduce the possibility of overfitting in the convolutional neural network learning. In this embodiment, the number of channels of each convolutional layer is 102, and the convolution kernel sizes of the first to twelfth convolutional layers are respectively: 16×16, 12×12, 10×10, 6×6, 5×5, 5×5, 4×4, 4×4, 4×4, 3×3, 3×3, 3×3; the neuron data of the first to fifth fully connected layers are respectively 1000, 1000, 800, 800, 120.

[0024] The 20×20 two-dimensional matrix in S1 is input into the above-mentioned CNN, where the first row, first column, eighteenth row and eighteenth column of the two-dimensional matrix represent the excitation port positions respectively, and the CNN output is the embedded network S parameters of the electromagnetic simulation of the embedded compensation network structure represented by the two-dimensional matrix.

[0025] Preferably, in the electromagnetic simulation process, after using a generative adversarial network to use the two-dimensional matrix and the embedded network S parameters as the input and output of the generative adversarial network, a new two-dimensional matrix and a corresponding embedded network S parameter data set are expanded to reduce the time required for data generation, and then on the PyTorch platform, the two-dimensional matrix and the embedded network S parameters are used as the input and output of the convolutional neural network, and the convolutional neural network is trained and verified to obtain a network parameter prediction model. The generative adversarial network can be in the form of a deep convolutional adversarial network (DCGAN), a progressive growing generative adversarial network (PGAN), a bulldozer distance generative adversarial network (WGAN), etc.

[0026] S3, after constructing the relationship between the embedded network S parameters and the amplifier Y parameters, globally optimize the embedded network S parameters to obtain an optimal embedded compensation network structure, and design the amplifier according to the optimal embedded compensation network.

[0027] like Figure 1 As shown, the Y parameter of the amplifier is composed of the Y parameter of the embedded compensation network structure and the Y parameter of the transistor, that is, the following formula (1): (1) In formula (1), is the Y parameter of the amplifier, is the Y parameter of the transistor, is the Y parameter of the embedded compensation network structure.

[0028] The relationship between the embedded network S parameters and the embedded compensation network structure Y parameters is constructed as follows (2): (2) In formula (2), is the identity matrix, is the embedded network S parameter.

[0029] According to equations (1) and (2), the relationship between the embedded network S parameters and the amplifier Y parameters is obtained, namely equation (3): (3) Combined with the relationship between the embedded network S parameters and the amplifier's Y parameters, the embedded network S parameters are globally optimized through a genetic algorithm to obtain the optimal embedded compensation network structure. The genetic algorithm can be replaced by other global optimization algorithms, such as: imperial competition algorithm, particle swarm algorithm, simulated annealing algorithm, greedy algorithm and domain search algorithm.

[0030] like Figure 5 As shown in Figure 2, the specific genetic algorithm optimization process is as follows: S3.1, first initialize the population, randomly generate 2048 two-dimensional matrices as parents through MATLAB described in S1, and use the CNN trained in S2 to predict the embedding network S parameters of the embedding compensation network represented by the 2048 two-dimensional matrices; S3.2, taking the expected S parameters as the target, embedding the mean square error between the network S parameters and the expected S parameters as the cost function, as shown in the following formula:

[0031]

[0032]

[0033] In the above formula, (Cost Function) is the cost function, To embed the network S parameters, is the expected S parameter, It is the function fitting relationship between the embedded network S parameters and the embedded compensation network structure.

[0034] The error between the embedded network S parameters and the expected S parameters is calculated by the above cost function, and the minimum error between the expected S parameters and the embedded network S parameters calculated by the cost function is used as the objective function, among which the two-dimensional matrices corresponding to the 16 minimum errors are selected and directly inherited to the offspring.

[0035] S3.3, randomly select 256 two-dimensional matrices from the parent generation, read the corresponding errors, select the two-dimensional matrices corresponding to the two smallest errors as the parent generation individuals, generate a child generation through crossover, and repeat S3.3 until the number of children generated is equal to the number of parents described in S3.1.

[0036] S3.4, update the population through the offspring generated in S3.3 and the two-dimensional matrix corresponding to the 16 minimum errors in S3.2. The new round of population also includes the offspring generated by the crossover in S3.3 and the two-dimensional matrix corresponding to the 16 minimum errors in S3.2.

[0037] Repeat S3.2~S3.4 until the preset number of iterations reaches 100 times, generate the 100th generation population, and finally select the two-dimensional matrix corresponding to the minimum error as the optimal solution to obtain the optimal embedding compensation network structure.

[0038] Finally, the amplifier is designed according to the optimal embedded compensation network structure, such as Figure 6 As shown, the electromagnetic simulation results show that the method designs a sub-terahertz amplifier with an operating frequency range of 180GHz~220GHz, a gain greater than 1.8dB, good stability, and a total power consumption of about 7.8mA. The electromagnetic simulation results verify that the amplifier design method based on convolutional neural network and genetic algorithm provided in this embodiment can solve the problem that the existing embedded compensation network has limited solution space and is strongly related to the operating frequency, and realizes a broadband and high-gain amplifier.

[0039] The millimeter wave and terahertz amplifier design method based on deep learning provided in the above embodiment has the following advantages: (1) Using the idea of ​​image pixelation, any embedded compensation network structure is pixelated and decomposed. A two-dimensional matrix is ​​used to represent any embedded compensation network structure. Then, a convolutional neural network is used to construct a fitting relationship between the embedded compensation network structure and the embedded network S parameters. The solution space is large. Finally, a broadband and high-gain millimeter-wave and terahertz amplifier can be obtained through genetic algorithm iteration. This method can solve the problem that the solution space of the embedded compensation network is limited and strongly related to the operating frequency, and tap the ability of active devices to provide gain in the sub-terahertz band, so that broadband and high-gain amplifiers can be designed. (2) Using a generative adversarial network to expand the data set formed by the embedded compensation network structure and its corresponding embedded network S parameters can reduce the time required for electromagnetic simulation and improve design efficiency; (3) By modifying the genetic algorithm cost function according to the different requirements of the amplifier, the fitting relationship between the arbitrary pixelated embedded compensation network structure constructed by the convolutional neural network algorithm and the embedded network S parameters can be used to comprehensively generate the optimal embedded compensation network structure that meets the performance requirements. This has the advantages of strong applicability and efficient design.

[0040] like Figure 4 As shown, this embodiment also provides a millimeter wave and terahertz amplifier design system based on deep learning, including: An embedded compensation network pixelation module, used to pixelate a plurality of arbitrary embedded compensation network structures and construct a corresponding two-dimensional matrix; An embedded network S parameter prediction module inputs the two-dimensional matrix into a network parameter prediction model to obtain corresponding embedded network S parameters; The embedded network S parameter prediction module is specifically used to input the two-dimensional matrix into the convolutional neural network and then output the embedded network S parameters, obtain a data set consisting of the embedded network S parameters corresponding to the two-dimensional matrix, train and verify the convolutional neural network through the data set, and obtain a network parameter prediction model. The convolutional neural network includes n convolutional layers and m fully connected layers, n is the number of convolutional layers, m is the number of fully connected layers, each of the convolutional layers is sequentially connected to a normalization layer and an activation layer, and the first m-1 levels of the fully connected layers are sequentially connected to a normalization layer, an activation layer, and a random inactivation layer, and the activation layer uses a Leaky ReLU function.

[0041] The embedded network S parameter prediction module also includes: A generative adversarial module is used to expand the two-dimensional matrix and its corresponding embedded network S parameters using a generative adversarial network.

[0042] The amplifier design module constructs the relationship between the embedded network S parameters and the amplifier Y parameters, performs global optimization on the embedded network S parameters to obtain an optimal embedded compensation network structure, and designs an amplifier based on the optimal embedded compensation network structure.

[0043] The amplifier design module is specifically used to obtain the Y parameters of the embedded compensation network structure according to the embedded network S parameters and the unit matrix; according to the Y parameters of the embedded compensation network structure and the unit matrix, and in combination with the Y parameters of the transistor, obtain the relationship between the embedded network S parameters and the Y parameters of the amplifier.

[0044] The amplifier design module is further specifically used to: combine the relationship between the above-mentioned embedded network S parameters and the Y parameters of the amplifier, form a parent generation with the two-dimensional matrices corresponding to the multiple arbitrary embedded compensation network structures, and use the mean square error between the embedded network S parameters and the expected S parameters as a cost function; calculate the error between the embedded network S parameters and the expected S parameters according to the cost function, and screen out the two-dimensional matrices corresponding to the minimum errors as a first subset; randomly select a number of the two-dimensional matrices in the parent generation and screen two two-dimensional matrices corresponding to the minimum errors to cross-produce a child generation, repeat this step until the number of the children generations is equal to the number of the parent generation, and the children generations constitute a second subset; merge the first subset and the second subset as a new population to iterate the parent generation, repeat the above steps until the preset number of iterations is reached, and then obtain the optimal embedded compensation network structure, and design the amplifier according to the optimal embedded compensation network structure.

[0045] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0046] In this embodiment, a computer device is also provided, which includes a processor and a memory, the memory is used to store a computer program (in this embodiment, the computer program includes a computing component and an iterative component, which can perform model calculation and model update), the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the millimeter wave and terahertz amplifier design method based on deep learning.

[0047] This embodiment also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space that stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the millimeter wave and terahertz amplifier design method based on deep learning in the above embodiment.

[0048] This embodiment also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the corresponding steps of the millimeter wave and terahertz amplifier design method based on deep learning in the above embodiment are implemented.

[0049] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0050] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0051] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0052] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0053] 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 above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A millimeter wave and terahertz amplifier design method based on deep learning, characterized in that: The following steps are involved: Pixelate multiple arbitrary embedded compensation network structures and construct corresponding two-dimensional matrices; Inputting the two-dimensional matrix into a network parameter prediction model to obtain corresponding embedded network S parameters; After constructing the relationship between the embedded network S parameters and the Y parameters of the amplifier, the embedded network S parameters are globally optimized to obtain an optimal embedded compensation network structure, and the amplifier is designed according to the optimal embedded compensation network structure.

2. The millimeter wave and terahertz amplifier design method based on deep learning according to claim 1, characterized in that: The process of establishing the network parameter prediction model is as follows: electromagnetic simulation is performed on multiple arbitrary embedded compensation network structures after pixelation to obtain a two-dimensional matrix and corresponding embedded network S parameters, the two-dimensional matrix and embedded network S parameters are used as the input and output of the convolutional neural network respectively, and the convolutional neural network is trained and verified to obtain the network parameter prediction model.

3. The millimeter wave and terahertz amplifier design method based on deep learning according to claim 2, characterized in that: The convolutional neural network includes n convolutional layers and m fully connected layers, where n is the number of convolutional layers and m is the number of fully connected layers. Each convolutional layer is sequentially connected to a normalization layer and an activation layer, and the first m-1 levels of fully connected layers are sequentially connected to a normalization layer, an activation layer and a random inactivation layer, and the activation layer adopts a Leaky ReLU function.

4. The millimeter wave and terahertz amplifier design method based on deep learning according to claim 1, characterized in that: The genetic algorithm is used to globally optimize the S parameters of the embedded network to obtain the optimal embedded compensation network structure, which specifically includes the following steps: The two-dimensional matrices corresponding to the plurality of arbitrary embedded compensation network structures are used as a parent generation, and the mean square error between the embedded network S parameters and the expected S parameters is used as a cost function; Calculate the error between the embedded network S parameter and the expected S parameter according to the cost function, and select the two-dimensional matrices corresponding to several minimum errors as the first subset; Randomly select a number of the two-dimensional matrices in the parent generation and select two two-dimensional matrices corresponding to the minimum errors to cross-generate a child generation, repeat this step until the number of the child generations is equal to the number of the parent generations, and the child generations constitute a second subset; The first subset and the second subset are fused as a new population to iterate the parent generation, and the above steps are repeated until a preset number of iterations is reached to obtain an optimal embedding compensation network structure.

5. The millimeter wave and terahertz amplifier design method based on deep learning according to claim 1, characterized in that: The step of constructing the relationship between the embedded network S parameters and the amplifier Y parameters specifically includes: The Y parameters of the amplifier are composed of the Y parameters of the embedded compensation network structure and the Y parameters of the transistor; Obtaining Y parameters of the embedded compensation network structure according to the embedded network S parameters and the unit matrix; According to the Y parameters and the unit matrix of the embedded compensation network structure, and in combination with the Y parameters of the transistor, the relationship between the embedded network S parameters and the Y parameters of the amplifier is obtained.

6. The method for designing a millimeter wave and terahertz amplifier based on deep learning according to claim 2, characterized in that: After the step of performing electromagnetic simulation on the pixelated multiple arbitrary embedded compensation network structures to obtain a two-dimensional matrix and corresponding embedded network S parameters, the method further includes: using a generative adversarial network to expand the two-dimensional matrix and its corresponding embedded network S parameters.

7. A millimeter wave and terahertz amplifier design system based on deep learning, characterized in that: include: An embedded compensation network pixelation module, used to pixelate a plurality of arbitrary embedded compensation network structures and construct a corresponding two-dimensional matrix; An embedded network S parameter prediction module inputs the two-dimensional matrix into a network parameter prediction model to obtain corresponding embedded network S parameters; The amplifier design module constructs the relationship between the embedded network S parameters and the amplifier Y parameters, performs global optimization on the embedded network S parameters to obtain an optimal embedded compensation network structure, and designs an amplifier based on the optimal embedded compensation network structure.

8. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable in the processor, wherein the processor implements the steps of the deep learning-based millimeter wave and terahertz amplifier design method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the millimeter wave and terahertz amplifier design method based on deep learning according to any one of claims 1 to 6 are implemented.

10. A computer program product, the computer product comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of the millimeter wave and terahertz amplifier design method based on deep learning according to any one of claims 1 to 6 are implemented.

Images