Electromagnetic topology simulation method and device based on convolutional neural network

Abstract

The application provides an electromagnetic topology simulation method and device based on a convolutional neural network, comprising: obtaining an electromagnetic topology structure to be simulated, determining electromagnetic topology data information based on the electromagnetic topology structure to be simulated, performing data preprocessing on the electromagnetic topology data information to obtain preprocessed electromagnetic topology data, the data preprocessing comprising eliminating electromagnetic topology data information corresponding to a suspended metal part in the electromagnetic topology structure to be simulated; inputting the preprocessed electromagnetic topology data into a trained electromagnetic topology simulation model, the electromagnetic topology simulation model extracting topology features of the electromagnetic topology structure, determining scattering simulation parameters corresponding to the electromagnetic topology data based on the extracted topology features; the electromagnetic topology simulation model comprising an input layer, a plurality of convolutional layers, a ReLU activation layer, a cross-channel parameter pooling layer, a fully connected layer and an output layer. The application improves the speed of electromagnetic simulation and realizes rapid design and optimization of the electromagnetic topology structure.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic (EM) topology design technology, and in particular to an electromagnetic topology simulation method and apparatus based on convolutional neural networks. Background Technology

[0002] In the field of microwave engineering, electromagnetic simulation is a crucial step in evaluating the performance of microwave components. While commonly used electromagnetic simulation methods such as the finite element method, finite-difference time-domain method, method of moments, and boundary element method can all perform electromagnetic simulations, traditional methods suffer from drawbacks such as high computational complexity and long processing times. This is particularly true when optimizing electromagnetic topologies, requiring the evaluation of numerous candidate designs. Existing electromagnetic simulation methods limit the speed and potential for design innovation.

[0003] Therefore, how to improve the speed of electromagnetic simulation and thus achieve rapid optimization of electromagnetic topology is a technical problem that urgently needs to be solved. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide an electromagnetic topology simulation method and apparatus based on convolutional neural networks to eliminate or improve one or more defects existing in the prior art.

[0005] One aspect of the present invention provides an electromagnetic topology simulation method based on a convolutional neural network, the method comprising:

[0006] The electromagnetic topology structure to be simulated is obtained, and electromagnetic topology data information is determined based on the electromagnetic topology structure to be simulated. The electromagnetic topology data information is preprocessed to obtain preprocessed electromagnetic topology data. The data preprocessing includes removing electromagnetic topology data information corresponding to suspended metal parts in the electromagnetic topology structure to be simulated.

[0007] The preprocessed electromagnetic topology data is input into a trained electromagnetic topology simulation model. The electromagnetic topology simulation model extracts the topological features of the electromagnetic topology structure based on the received preprocessed electromagnetic topology data, and determines the scattering simulation parameters corresponding to the electromagnetic topology data based on the extracted topological features. The electromagnetic topology simulation model includes an input layer, multiple convolutional layers, a ReLU activation layer, a cross-channel parameter pooling layer, a fully connected layer, and an output layer.

[0008] In some embodiments of the present invention, the method further includes:

[0009] An initial electromagnetic topology simulation model, a sample dataset, and a model loss function are constructed. The sample dataset includes preprocessed electromagnetic topology data samples and scattering simulation parameter sample data.

[0010] The initial electromagnetic topology simulation model is pre-trained based on the sample dataset and the model loss function to obtain a trained electromagnetic topology simulation model.

[0011] In some embodiments of the present invention, the model loss function is:

[0012] ;

[0013] Where n represents the total number of sample data. This represents the true value of the scattering parameters. This represents the predicted value of the scattering parameters.

[0014] In some embodiments of the present invention, the convolutional kernel of the convolutional layer is 3x3, and the pooling kernel of the cross-channel parameter pooling layer is 1x1; and / or,

[0015] The method further includes:

[0016] The breadth-first search algorithm was used to verify whether all metal units in the electromagnetic topology were connected to the outer wall of the waveguide structure.

[0017] In some embodiments of the present invention, a trained electromagnetic topology simulation model is obtained by pre-training the initial electromagnetic topology simulation model based on the sample dataset and the model loss function, including:

[0018] The network weights of the initial electromagnetic topology simulation model are initialized based on the Kaiming initialization method; and / or,

[0019] The model parameters of the initial electromagnetic topology simulation model are optimized based on the Adam optimization algorithm.

[0020] In some embodiments of the present invention, the data preprocessing includes:

[0021] The electromagnetic topology data information is normalized or binarized.

[0022] In some embodiments of the present invention, the electromagnetic topology data information corresponding to the suspended metal component in the electromagnetic topology structure to be simulated is removed, including:

[0023] Searching for metal elements in the edge elements of an electromagnetic topology based on an edge element priority search algorithm;

[0024] The connectivity between the metal units is verified based on the breadth-first search algorithm, and the suspended metal components in the electromagnetic topology are determined based on the connectivity.

[0025] The electromagnetic topology data information corresponding to the suspended metal parts in the electromagnetic topology structure is removed.

[0026] In some embodiments of the present invention, the method includes:

[0027] Verify the accuracy of the simulation results of the pre-trained electromagnetic topology simulation model;

[0028] Based on the accuracy of the simulation results, optimize the model structure and parameters of the electromagnetic topology simulation model;

[0029] The electromagnetic topology simulation model with optimized model structure and model parameters was re-pre-trained;

[0030] The model structure includes the number of convolutional layers and the number of cross-channel parameter pooling layers. The model parameters include the convolutional kernel size, pooling kernel size, number of channels, and the dimension of the fully connected layer.

[0031] According to another aspect of the present invention, an electromagnetic topology simulation system based on a convolutional neural network is also disclosed. The system includes a processor, a memory, and a computer program stored in the memory. The processor is used to execute the computer program. When the computer program is executed, the system implements the steps of the method as described in any of the above embodiments.

[0032] According to another aspect of the present invention, a computer-readable storage medium is also disclosed, on which a computer program is stored, which, when executed by a processor, implements the steps of the method as described in any of the above embodiments.

[0033] The electromagnetic topology simulation method and apparatus based on convolutional neural networks disclosed in the above embodiments of the present invention first remove the electromagnetic topology data information corresponding to the suspended metal parts in the electromagnetic topology structure to be simulated, and then quickly and accurately obtain the scattering simulation parameters corresponding to the electromagnetic topology data through the pre-processed electromagnetic topology data using a trained electromagnetic topology simulation model. This method can perform fast and accurate simulation of electromagnetic topology structures, improve the speed of electromagnetic simulation, and thus realize the efficient design and rapid optimization of electromagnetic topology structures.

[0034] Furthermore, the electromagnetic topology simulation model of this application utilizes multiple small convolutional filters and introduces a cross-channel parameter pooling layer to enhance network depth, effectively improving the model's generalization ability to microwave topology data while reducing the number of model parameters and weights. Thus, this method can accurately capture the key features of microwave topology and achieve relatively accurate electromagnetic characteristic prediction. In addition, while maintaining high accuracy, this method reduces the demand for computing resources, enabling it to run efficiently on platforms with limited computing power, thereby expanding its application scope.

[0035] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the description, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures specifically pointed out in the description and drawings.

[0036] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0037] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. The components in the drawings are not drawn to scale but are merely illustrative of the principles of the invention. For ease of illustration and description of certain parts of the invention, corresponding portions in the drawings may be enlarged, i.e., may appear larger relative to other components in an exemplary device actually manufactured according to the invention. In the drawings:

[0038] Figure 1 This is a flowchart illustrating an embodiment of the electromagnetic topology simulation method based on a convolutional neural network according to this application.

[0039] Figure 2 This is a schematic diagram of the architecture of an electromagnetic topology simulation model according to an embodiment of this application.

[0040] Figure 3 This is a schematic diagram of the convolution kernel of the convolutional layer in an electromagnetic topology simulation model according to an embodiment of this application.

[0041] Figure 4 This is a schematic diagram of electromagnetic topology data information corresponding to the suspended metal component of an electromagnetic topology structure according to an embodiment of this application.

[0042] Figure 5 This is a flowchart illustrating another embodiment of the electromagnetic topology simulation method based on a convolutional neural network in this application.

[0043] Figure 6a This is a schematic diagram of the electromagnetic topology of a first-order microwave filter.

[0044] Figure 6b This is a schematic diagram of the electromagnetic topology of a third-order microwave filter.

[0045] Figure 7a This is a schematic diagram of the simulation results corresponding to the first electromagnetic topology of a first-order microwave filter.

[0046] Figure 7b This is a schematic diagram of the simulation results corresponding to the second electromagnetic topology of a first-order microwave filter.

[0047] Figure 7c This is a schematic diagram of the simulation results corresponding to the third electromagnetic topology of a first-order microwave filter.

[0048] Figure 7d This is a schematic diagram of the simulation results corresponding to the fourth electromagnetic topology of a first-order microwave filter.

[0049] Figure 8a This is a schematic diagram of the simulation results corresponding to the first electromagnetic topology of a third-order microwave filter.

[0050] Figure 8b This is a schematic diagram of the simulation results corresponding to the second electromagnetic topology of a third-order microwave filter.

[0051] Figure 8c This is a schematic diagram of the simulation results corresponding to the third electromagnetic topology of a third-order microwave filter.

[0052] Figure 8d This is a schematic diagram of the simulation results corresponding to the fourth electromagnetic topology of a third-order microwave filter.

[0053] Figure 8e This is a schematic diagram of the simulation results corresponding to the fifth electromagnetic topology of a third-order microwave filter.

[0054] Figure 8f This is a schematic diagram of the simulation results corresponding to the sixth electromagnetic topology of a third-order microwave filter.

[0055] Figure 8g This is a schematic diagram of the simulation results corresponding to the seventh electromagnetic topology of a third-order microwave filter.

[0056] Figure 8h This is a schematic diagram of the simulation results corresponding to the eighth electromagnetic topology of a third-order microwave filter. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0058] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0059] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0060] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection with an intermediary, and can refer not only to a wired connection, but also to a wireless connection. The specific meaning can be changed based on the actual application scenario.

[0061] The electromagnetic topology simulation method and apparatus based on convolutional neural networks of this application can quickly and accurately simulate the electromagnetic characteristics of electromagnetic topologies, thereby further evaluating and optimizing the designed electromagnetic topologies based on the simulation results. The electromagnetic topology simulation method and apparatus of this application can be applied to the design and optimization of electromagnetic topologies, as well as to electromagnetic simulation scenarios.

[0062] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0063] Figure 1 This is a flowchart illustrating an embodiment of the electromagnetic topology simulation method based on a convolutional neural network according to this application. Figure 1 As shown, the electromagnetic topology simulation method includes at least steps S10 and S20.

[0064] Step S10: Obtain the electromagnetic topology structure to be simulated, determine the electromagnetic topology data information based on the electromagnetic topology structure to be simulated, and perform data preprocessing on the electromagnetic topology data information to obtain preprocessed electromagnetic topology data. The data preprocessing includes removing the electromagnetic topology data information corresponding to the suspended metal parts in the electromagnetic topology structure to be simulated.

[0065] In this step, the electromagnetic topology to be simulated is first converted into electromagnetic topology data information, and then the electromagnetic topology data information is preprocessed to obtain preprocessed electromagnetic topology data. During the data preprocessing process, the electromagnetic topology data information corresponding to the suspended metal components in the electromagnetic topology to be simulated is further removed. The electromagnetic topology data information corresponding to the suspended metal components in the electromagnetic topology structure is as follows: Figure 4 As shown, Figure 4 The content within the dashed box indicated by the middle arrow represents the electromagnetic topology data information corresponding to the suspended metal component; Figure 4 The diagram illustrates whether the topology is connected or not. Figure 4 The connection between the metal unit and the outer wall of the waveguide is shown visually. Additionally... Figure 6a and Figure 6b Schematic diagrams of the electromagnetic topologies of first-order and third-order microwave filters are shown respectively. The diagrams visualize the key structural features and topological dimensions of the different filters.

[0066] In the above steps, removing the electromagnetic topology data information corresponding to the suspended metal parts in the electromagnetic topology structure to be simulated is to improve the connectivity of the electromagnetic topology structure, that is, to eliminate non-physically meaningful structures in the electromagnetic topology structure, thereby improving the rationality and accuracy of the scattering simulation parameters of the electromagnetic topology data. This allows for better guidance on the design and optimization of the electromagnetic topology structure based on the simulation results, reducing the number of optimization iterations in the manufacturing process, lowering costs, and maintaining or improving product performance.

[0067] Furthermore, the electromagnetic topology data information corresponding to the suspended metal components in the electromagnetic topology structure to be simulated is removed, specifically including the following steps: searching for metal units in the edge units of the electromagnetic topology structure based on the edge unit priority search algorithm; verifying the connectivity between the metal units based on the breadth-first search algorithm, and determining the suspended metal components in the electromagnetic topology structure based on the connectivity; and removing the electromagnetic topology data information corresponding to the suspended metal components in the electromagnetic topology structure. This embodiment preprocesses the electromagnetic topology data information based on the breadth-first search (BFS) algorithm to avoid the appearance of non-physically meaningful suspended metal components in the electromagnetic topology structure; and preprocesses the training dataset of the electromagnetic topology simulation model based on this method, which can also reduce the training data during the training process of the electromagnetic topology simulation model, thereby accelerating the design process of the electromagnetic topology structure.

[0068] In graph theory and computer science, the Breadth-First Search (BFS) algorithm is used to traverse or search tree and graph structures. Starting from a given source node, it systematically explores nodes in the graph in the width direction, first visiting all neighboring nodes and then delving deeper layer by layer, ensuring the shortest path from the source node to all reachable nodes. In this application, the BFS algorithm is reimagined and applied to EM topology design. The optimized BFS algorithm is used to verify whether all metal elements in the EM topology design are connected to the outer wall of the waveguide structure. Specifically, firstly, metal elements are searched in the edge elements of the design space, and all edge metal elements are enqueued; by marking visited elements, loops and infinite loops are avoided; furthermore, a queue data structure is used to manage the order of node exploration, ensuring that each node is visited exactly once. In the specific implementation, starting from the edge elements of each rectangular plate, all metal elements in the edge elements are searched and added to the queue; then, the metal elements in the queue are processed one by one. For each metal element, its directly connected metal elements are found, and unvisited metal elements are added to the end of the queue; this process expands progressively until all metal elements directly or indirectly connected to the first metal element are found. Finally, the unvisited cells in the entire rectangular plate are checked. If there are no metal cells among these unvisited cells, it indicates that all metal cells are directly or indirectly connected to the edge metal cells, meaning that the electromagnetic topology does not have any suspended metal components, and thus the electromagnetic topology meets the manufacturability condition. Otherwise, if there are metal cells not connected to the outer wall, it indicates that the electromagnetic topology has suspended metal components, and thus the electromagnetic topology does not meet the manufacturability condition. It is understood that the electromagnetic topology simulation method of this application performs data preprocessing to remove electromagnetic topology data information corresponding to suspended metal components during both the actual electromagnetic simulation process and the training process of the electromagnetic topology simulation model. This ensures that the data received by the electromagnetic topology simulation model corresponds to a manufacturable electromagnetic topology, ignoring meaningless suspended metal components during the electromagnetic simulation process, thereby improving the training efficiency and prediction accuracy of the model.

[0069] For example, when removing electromagnetic topology data information corresponding to suspended metal components in an electromagnetic topology structure, the specific process includes the following:

[0070] 1. Edge cell priority search: First, search for metal cells in the edge cells of the design space. This is based on the understanding that edge cells have an important impact on connectivity in microwave component design.

[0071] 2. Access Marking Optimization: By introducing an access marking mechanism, explored nodes can be managed efficiently, avoiding the problems of repeated visits and infinite loops in traditional BFS.

[0072] 3. Connectivity verification logic: The connectivity verification logic based on the BFS algorithm ensures that the connectivity of the metal unit meets the manufacturability requirements. This logic is specially designed according to the physical characteristics and manufacturing process of microwave components.

[0073] 4. Queue Management Mechanism: The queue management mechanism makes the search and verification process of metal units more efficient.

[0074] Step S20: Input the preprocessed electromagnetic topology data into the trained electromagnetic topology simulation model. The electromagnetic topology simulation model extracts the topological features of the electromagnetic topology structure based on the received preprocessed electromagnetic topology data, and determines the scattering simulation parameters corresponding to the electromagnetic topology data based on the extracted topological features. The electromagnetic topology simulation model includes an input layer, multiple convolutional layers, ReLU activation layers, cross-channel parameter pooling layers, fully connected layers, and an output layer.

[0075] In this step, the preprocessed electromagnetic topology data is used as input data for the electromagnetic topology simulation model, so that the electromagnetic topology simulation model outputs the predicted results of scattering simulation parameters. Figure 2 The diagram illustrates the connection methods of the input layer, multiple convolutional layers, ReLU activation layers, cross-channel parameter pooling layers, fully connected layers, and output layer of the electromagnetic topology simulation model, and clearly expresses the depth and complexity of the model.

[0076] The electromagnetic topology simulation model in this application is suitable for processing two-dimensional or three-dimensional data of microwave topology. The electromagnetic topology simulation model includes a cross-channel parameter pooling layer, which allows the network to increase its depth and improve its performance without losing critical spatial information. Since the accurate representation of the spatial relationships of the electromagnetic topology is crucial for the accuracy of prediction, the electromagnetic topology simulation model in this application introduces a cross-channel parameter pooling layer design, thus making it suitable for handling complex electromagnetic problems.

[0077] Furthermore, the electromagnetic topology simulation model of this application can employ multiple small convolutional filters to capture subtle features in the electromagnetic topology; for example, the convolutional kernel of the convolutional layer is 3x3 (refer to...). Figure 3 The pooling kernel of the cross-channel parameter pooling layer is 1x1. In this embodiment, the use of a small convolutional filter with a 3x3 kernel improves the electromagnetic topology simulation model's ability to capture the details of microwave topology, thereby improving the model's expressive power and computational efficiency, and further improving the accuracy of the scattering simulation parameters predicted by the electromagnetic topology simulation method.

[0078] In the electromagnetic topology simulation model of the above embodiments, multiple small convolutional filters are used to enhance the model's feature extraction capability; a cross-channel parameter pooling layer is introduced to increase the network depth, improving model performance and generalization ability, reducing the number of model parameters and weights, and optimizing the model's computational efficiency. Furthermore, the above electromagnetic topology simulation model connects a ReLU activation function layer after the convolutional layer to enhance the network's nonlinear expressive ability; a cross-channel parameter pooling layer with a 1x1 pooling kernel is used to maintain spatial resolution and capture interactions between features.

[0079] In some embodiments of this application, the electromagnetic topology simulation method based on convolutional neural networks further includes the following steps: constructing an initial electromagnetic topology simulation model, a sample dataset, and a model loss function, wherein the sample data in the sample dataset includes preprocessed electromagnetic topology data samples and scattering simulation parameter sample data; and pre-training the initial electromagnetic topology simulation model based on the sample dataset and the model loss function to obtain a trained electromagnetic topology simulation model. This embodiment is used to construct an electromagnetic topology simulation model. In this embodiment, an initial electromagnetic topology simulation model is first constructed, which includes an input layer, multiple convolutional layers, a ReLU activation layer, a cross-channel parameter pooling layer, a fully connected layer, and an output layer; further, electromagnetic topology structures and S (scattering) parameters of different microwave components are collected, and the collected electromagnetic topology structures and S (scattering) parameters of different microwave components are preprocessed to form a sample dataset; the preprocessed sample data meets the input requirements of the initial electromagnetic topology simulation model. For example, the data preprocessing may specifically include removing electromagnetic topology data information corresponding to suspended metal parts, and normalizing or binarizing the electromagnetic topology data information.

[0080] In the process of constructing the sample dataset, the electromagnetic topology data information corresponding to the suspended metal parts is removed, so that the electromagnetic topology structure corresponding to the sample data is a manufacturable structure. The electromagnetic topology simulation model is trained based on the preprocessed sample data, which can reduce the number of model iterations and avoid the impact of electromagnetic topology simulation model on simulation results due to topology design defects.

[0081] Furthermore, during the pre-training phase of the initial electromagnetic topology simulation model, the network weights of the initial electromagnetic topology simulation model can be initialized based on the Kaiming initialization method; and / or, the model parameters of the initial electromagnetic topology simulation model can be optimized based on the Adam optimization algorithm. This embodiment initializes the model weights based on the Kaiming initialization technique, avoiding a series of problems such as gradient vanishing or exploding during model training, and also ensuring that the model can converge quickly and that the simulation results have high accuracy.

[0082] In one embodiment, the electromagnetic topology simulation model uses root mean square error as the loss function, and the specific model loss function can be:

[0083] ;

[0084] Where n represents the total number of sample data. This represents the true value of the scattering parameters. This represents the predicted value of the scattering parameters.

[0085] Furthermore, the electromagnetic topology simulation model can use the relative root mean square error as the evaluation function. Specifically, the model evaluation function can be:

[0086] ;

[0087] Where n represents the total number of sample data. This represents the true value of the scattering parameters. This represents the predicted value of the scattering parameters. This represents the maximum predicted value of the scattering parameter. This represents the minimum predicted value of the scattering parameter.

[0088] As can be seen from the above embodiments, this application pre-trains the initial electromagnetic topology simulation model based on the preprocessed sample dataset. During the training process, the Kaiming initialization technique is used to avoid gradient vanishing or exploding problems, and appropriate loss functions and optimization algorithms are used to improve the performance of the model. As a result, the prediction accuracy, computation speed and generalization ability of the trained electromagnetic topology simulation model are all better.

[0089] In addition, to ensure model performance during pre-training, in some embodiments, the electromagnetic topology simulation method further includes the following steps: optimizing the model structure and parameters of the electromagnetic topology simulation model based on the accuracy of the simulation results; re-pre-training the electromagnetic topology simulation model with the optimized model structure and parameters; wherein the model structure includes the number of convolutional layers and the number of cross-channel parameter pooling layers, and the model parameters include the convolutional kernel size, pooling kernel size, number of channels, and fully connected layer dimension. In this embodiment, the trained electromagnetic topology simulation model is validated through simulation results to evaluate the training effect. If the simulation results of the trained electromagnetic topology simulation model do not meet the usage requirements, the model is further adjusted and trained until an electromagnetic topology simulation model that meets the usage requirements is obtained.

[0090] Furthermore, the accuracy and reliability of the electromagnetic topology simulation results presented in this application can be further verified through experimental results. Specifically, the performance of the electromagnetic topology simulation method in different microwave topology designs can be verified, including accuracy, computational speed, and generalization ability.

[0091] When verifying accuracy, the simulation results obtained by the electromagnetic topology simulation method based on this application can be compared with the simulation results obtained by existing electromagnetic simulation methods. This test includes predicting the S-parameters of microwave components with different configurations and evaluating the accuracy of the electromagnetic topology simulation method based on this application by comparing it with high-fidelity electromagnetic simulation results. The test results show that the electromagnetic topology simulation method based on this application can predict electromagnetic characteristics with high accuracy and has a significant performance improvement compared with existing electromagnetic simulation methods.

[0092] When verifying the computational speed, runtime tests can be conducted under different hardware configurations to evaluate the response speed of the electromagnetic topology simulation method in practical applications. The results show that even in environments with limited computing resources, the electromagnetic topology simulation method of this application can quickly predict electromagnetic characteristics, greatly shortening the design evaluation time.

[0093] When verifying the generalization ability of the electromagnetic topology simulation model in the electromagnetic topology simulation method, a batch of data that was not used in the model training was used for testing. The experimental results show that the model not only performs well on the training data, but also provides accurate predictions on unseen data, demonstrating good generalization ability. This generalization ability enables the electromagnetic topology simulation method of this application to adapt to changing design requirements and application scenarios.

[0094] In addition, when verifying the electromagnetic topology simulation method of this application, two different filters can be selected to verify the practicality of the method in microwave topology design; that is, the prediction results obtained by the electromagnetic topology simulation method of this application are applied to the actual microwave component manufacturing process, and the impact of the test model output on manufacturing accuracy and cost is demonstrated by proving the practicality of the electromagnetic topology simulation method of this application. Figure 7a , 7b 7c and 7d are schematic diagrams showing the simulation results for different electromagnetic topologies of a first-order microwave filter. Figure 8a , 8b Figures 8c, 8d, 8e, 8f, 8g, and 8h represent simulation results for different electromagnetic topologies of a third-order microwave filter. As shown in the above simulation results, the scattering simulation parameters predicted by the electromagnetic topology simulation method of this application are very close to the actual scattering parameters, thus proving the accuracy, reliability, generalization ability, and prediction accuracy of the electromagnetic topology simulation method based on convolutional neural networks in this application. Therefore, the electromagnetic topology simulation method of this application, by simulating the electromagnetic characteristics of the electromagnetic topology, can be directly applied to the design of electromagnetic topologies, providing an efficient and fast electromagnetic topology simulation method for electromagnetic topology design, thereby improving the efficiency and accuracy of microwave component design.

[0095] In some embodiments of the present invention, after obtaining the scattering simulation parameters corresponding to the electromagnetic topology, a breadth-first search algorithm is further used to verify whether all metal units in the electromagnetic topology are connected to the outer wall of the waveguide structure. This method avoids the inclusion of non-physically meaningful suspended metal parts in the electromagnetic topology, prevents the presence of non-physically meaningful structures in the electromagnetic topology, ensures the manufacturability of microwave components, and reduces costs.

[0096] Furthermore, when the convolutional neural network-based electromagnetic topology simulation method of this application is applied to the design process of electromagnetic topology structures, designers can explore more design options within the same timeframe, significantly improving design efficiency. This provides more possibilities for innovative design and helps develop microwave components with superior performance and more compact structures. Moreover, this simulation method reduces reliance on expensive hardware simulation resources and time consumption, helping to lower overall R&D costs, especially in the early stages of product development, where cost savings are particularly significant. Additionally, completing the design of electromagnetic topology structures based on the electromagnetic topology simulation method of this application reduces iterations and rework, helping to reduce the manufacture of physical prototypes, thereby reducing material consumption and environmental impact, aligning with the concepts of green design and sustainable development. Beyond the above, this application demonstrates the application potential of deep learning technology in the field of electromagnetics, providing new ideas and tools for the integration of technologies across different disciplines, and contributing to the promotion of interdisciplinary research and innovation. With the development of 5G and future communication technologies, the demand for high-performance microwave components is increasing. The electromagnetic topology simulation method of this application provides an effective design means to meet these needs, laying a solid foundation for future technological development. In summary, the advantages of the electromagnetic topology simulation method proposed in this application are not only reflected in the technological breakthrough, but also in promoting industry development, reducing costs, improving design quality, and driving technological innovation.

[0097] Figure 5 This is a flowchart illustrating another embodiment of the electromagnetic topology simulation method based on convolutional neural networks, showing the construction and optimization process of the electromagnetic topology simulation model. Figure 5As shown, firstly, data variables in the topology are defined, and a dataset is randomly generated. These data variables can also be understood as electromagnetic topology data information. Next, the constrained EM topology design method is used to filter the data (preprocessing the data), and the data is randomly divided into training and test sets. The neural network structure is then designed and adjusted. The model is trained using the training set. If the training error is less than θ, the model is tested using the test set. If the test error is less than ε, the process stops. If the training error is not less than θ, the neural network structure is redesigned and adjusted. If the test error is not less than ε, it is determined whether the test error is decreasing. If it is decreasing, the neural network structure is redesigned and adjusted. If the test error is not decreasing, data is added again, and the constrained EM topology design method is used again to filter the data, randomly dividing it into training and test sets.

[0098] Accordingly, the present invention also provides an electromagnetic topology simulation system based on a convolutional neural network. The system includes a processor, a memory, and a computer program stored in the memory. The processor is used to execute the computer program. When the computer program is executed, the system implements the steps of the method as described in any of the above embodiments.

[0099] For example, an electromagnetic topology simulation system based on a convolutional neural network may specifically include a data generation module, a model training module, and a performance verification module. The data generation module is used to generate EM topology data that conforms to design constraints. The model training module is used to train the electromagnetic topology simulation model and optimize the model parameters. The performance verification module is used to verify the accuracy and reliability of the model through experimental results. Furthermore, the data generation module can also use the finite element method (FEM) to perform EM simulation and generate topology data of metals and non-metals within the design space.

[0100] This invention also provides a computer-readable storage medium and a computer program product having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of the above embodiments. The computer-readable storage medium may be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium known in the art.

[0101] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0102] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0103] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0104] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An electromagnetic topology simulation method based on convolutional neural networks, characterized in that, The method includes: The electromagnetic topology structure to be simulated is obtained, and electromagnetic topology data information is determined based on the electromagnetic topology structure to be simulated. The electromagnetic topology data information is preprocessed to obtain preprocessed electromagnetic topology data. The data preprocessing includes removing electromagnetic topology data information corresponding to suspended metal parts in the electromagnetic topology structure to be simulated. The preprocessed electromagnetic topology data is input into a trained electromagnetic topology simulation model. The electromagnetic topology simulation model extracts the topological features of the electromagnetic topology structure based on the received preprocessed electromagnetic topology data, and determines the scattering simulation parameters corresponding to the electromagnetic topology data based on the extracted topological features. The electromagnetic topology simulation model includes an input layer, multiple convolutional layers, ReLU activation layers, cross-channel parameter pooling layers, fully connected layers, and an output layer. During the training of the electromagnetic topology simulation model, data preprocessing is performed to remove electromagnetic topology data information corresponding to the suspended metal parts. The electromagnetic topology data information corresponding to the suspended metal components in the electromagnetic topology structure to be simulated is removed, including: Searching for metal elements in the edge elements of an electromagnetic topology based on an edge element priority search algorithm; The connectivity between the metal units is verified based on the breadth-first search algorithm, and the suspended metal components in the electromagnetic topology are determined based on the connectivity. The electromagnetic topology data information corresponding to the suspended metal parts in the electromagnetic topology structure is removed. The process begins with the edge cells of each rectangular plate, searching for all metal cells within the edge cells and adding them to a queue. Then, each metal cell in the queue is processed one by one. For each metal cell, the metal cells directly or indirectly connected to it are searched, and any unvisited metal cells are added to the end of the queue, until all metal cells directly or indirectly connected to the first metal cell are found. The unvisited cells in the entire rectangular plate are then checked. If no metal cells are found in the unvisited cells, it indicates that all metal cells are directly or indirectly connected to the edge metal cells, meaning the electromagnetic topology does not have a suspended metal component. Otherwise, if there are metal cells not connected to the outer wall, it indicates that the electromagnetic topology has a suspended metal component. The evaluation function for the electromagnetic topology simulation model is: Where n represents the total number of sample data, This represents the true value of the scattering parameters. This represents the predicted value of the scattering parameters. This represents the maximum predicted value of the scattering parameter. This represents the minimum predicted value of the scattering parameter.

2. The electromagnetic topology simulation method based on convolutional neural networks according to claim 1, characterized in that, The method further includes: An initial electromagnetic topology simulation model, a sample dataset, and a model loss function are constructed. The sample dataset includes preprocessed electromagnetic topology data samples and scattering simulation parameter sample data. The initial electromagnetic topology simulation model is pre-trained based on the sample dataset and the model loss function to obtain a trained electromagnetic topology simulation model.

3. The electromagnetic topology simulation method based on convolutional neural networks according to claim 2, characterized in that, The model loss function is: ； Where n represents the total number of sample data. This represents the true value of the scattering parameters. This represents the predicted value of the scattering parameters.

4. The electromagnetic topology simulation method based on convolutional neural networks according to claim 1, characterized in that, The convolutional kernel of the convolutional layer is 3x3, and the pooling kernel of the cross-channel parameter pooling layer is 1x1; and / or, The method further includes: The breadth-first search algorithm was used to verify whether all metal units in the electromagnetic topology were connected to the outer wall of the waveguide structure.

5. The electromagnetic topology simulation method based on convolutional neural networks according to claim 2, characterized in that, Based on the sample dataset and the model loss function, the initial electromagnetic topology simulation model is pre-trained to obtain a trained electromagnetic topology simulation model, including: The network weights of the initial electromagnetic topology simulation model are initialized based on the Kaiming initialization method; and / or, The model parameters of the initial electromagnetic topology simulation model are optimized based on the Adam optimization algorithm.

6. The electromagnetic topology simulation method based on convolutional neural networks according to claim 1, characterized in that, The data preprocessing includes: The electromagnetic topology data information is normalized or binarized.

7. The electromagnetic topology simulation method based on convolutional neural networks according to claim 2, characterized in that, The method includes: Verify the accuracy of the simulation results of the pre-trained electromagnetic topology simulation model; Based on the accuracy of the simulation results, optimize the model structure and parameters of the electromagnetic topology simulation model; The electromagnetic topology simulation model with optimized model structure and model parameters was re-pre-trained; The model structure includes the number of convolutional layers and the number of cross-channel parameter pooling layers. The model parameters include the convolutional kernel size, pooling kernel size, number of channels, and the dimension of the fully connected layer.

8. An electromagnetic topology simulation system based on a convolutional neural network, the system comprising a processor, a memory, and a computer program stored in the memory, characterized in that, The processor is configured to execute the computer program, and when the computer program is executed, the system implements the steps of the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method as described in any one of claims 1 to 7.

Images