Reverse modeling method of microwave coupling filter

Abstract

The invention belongs to the technical field of filter modeling, and relates to a reverse modeling method of a microwave coupling filter, which comprises the following steps: acquiring a design target of the microwave coupling filter, defining an ideal coupling matrix, and obtaining a plurality of coupling matrix samples to construct a sample set; calculating an S parameter of each coupling matrix sample at a set frequency point to construct a condition set; the sample set and the condition set serve as input of a diffusion model, the diffusion model is trained, and multiple sets of initial coupling parameters are obtained; obtaining an S parameter corresponding to each group of initial coupling parameters according to the forward model; calculating and screening mean square errors of the target S parameter and the corresponding S parameter to obtain a plurality of groups of optimal coupling parameters; and clustering the multiple groups of optimal coupling parameters to obtain coupling parameter distribution so as to realize reverse modeling of the microwave coupling filter. According to the invention, reverse modeling of the microwave coupling filter can be realized.

Description

Technical Field

[0001] This application relates to the technical field of filter modeling, and particularly to an inverse modeling method for microwave coupled filters. Background Art

[0002] The modeling of microwave coupled filters is usually forward modeling, that is, taking physical or geometric variables as inputs and outputting corresponding electrical responses. Some forward modeling techniques have been developed and matured, improving the efficiency of filter design by avoiding repeated electromagnetic simulations.

[0003] However, inverse modeling takes the S-parameters of the filter as inputs and outputs corresponding coupling matrix parameters, thereby realizing the design and manufacture of filters with specific electrical performances, and has received increasing attention and development. Among them, for actual filters, the tuning process is essential, and the coupling matrix plays a key role in optimization techniques such as spatial mapping in the heading model. Therefore, accurately solving the inverse problem (i.e., the extraction of the coupling matrix) is of great significance.

[0004] In the prior art, the main challenge of inverse modeling lies in the lack of a clear one-to-one mapping relationship and the inability to directly solve it through analytical methods. The common methods are mainly divided into the following two categories.

[0005] The first category of methods relies on numerical optimization techniques, such as the Cauchy method, vector fitting, and global optimization algorithms; these methods usually require multiple iterations, which are particularly complex in the case of high-dimensional variables. Direct global optimization is not only time-consuming but also difficult to quickly find the optimal solution; in addition, there is also the problem of non-uniqueness. Non-uniqueness means that the same or similar S-parameters as inputs may correspond to multiple contradictory output coupling matrices, which leads to large model training errors and the accuracy is difficult to meet the actual requirements.

[0006] The second category of methods is based on deep learning. Through data generation and network training, artificial neural networks (ANNs) are used to approximate complex non-linear multi-dimensional functions and applied to the inverse modeling of filters; but there is also the problem of non-uniqueness, and it is difficult to obtain a high-precision inverse model directly using neural networks.

[0007] To address the non-uniqueness problem, some studies have proposed methods based on data grouping and model combination. For example, by dividing data groups to eliminate contradictory samples and constructing multiple sub-models for combination; in addition, multi-valued neural networks handle contradictory samples through special training error functions, realizing multi-solution processing in inverse modeling. However, these methods usually require additional prior knowledge or manual intervention and are difficult to be fully automated. Summary of the Invention

[0008] Based on this, it is necessary to provide an inverse modeling method for microwave coupled filters to address the above-mentioned technical problems, which can achieve the inverse modeling of microwave coupled filters, solve the non-uniqueness problem existing in the prior art, and realize full automation without additional prior knowledge or manual intervention.

[0009] An inverse modeling method for microwave coupled filters includes: Obtain the design objectives of the microwave coupled filter and define the ideal coupling matrix; According to the ideal coupling matrix, set the coupling matrix space and sample in the coupling matrix space to obtain multiple coupling matrix samples to construct a sample set; calculate the S-parameters of each coupling matrix sample at the set frequency points to construct a condition set; Use the sample set and the condition set as the input of the diffusion model, train the diffusion model to learn the mapping relationship between the coupling parameters and the S-parameters, and obtain the trained diffusion model; Obtain the target S-parameters, and input the target S-parameters into the trained diffusion model multiple times to obtain multiple model outputs corresponding to the target S-parameters; according to the multiple model outputs, obtain multiple groups of initial coupling parameters; Input each group of initial coupling parameters into the forward model respectively to obtain the corresponding S-parameters of each group of initial coupling parameters; calculate the mean square error between the target S-parameters and the corresponding S-parameters, and screen according to the set threshold to obtain multiple groups of optimal coupling parameters; Cluster the multiple groups of optimal coupling parameters, and according to the optimal coupling parameter with the smallest mean square error in each cluster, obtain the coupling parameter distribution to achieve the inverse modeling of the microwave coupled filter.

[0010] In one embodiment, according to the ideal coupling matrix, set the coupling matrix space and sample in the coupling matrix space to obtain multiple coupling matrix samples to construct a sample set, including: Set the tolerance range for each variable in the ideal coupling matrix to obtain the value space of the variable as the coupling matrix space; Perform Latin hypercube sampling in the coupling matrix space, and use a set of sampled coupling parameters as a sample to obtain multiple coupling matrix samples; Construct a sample set according to the multiple coupling matrix samples.

[0011] In one embodiment, calculate the S-parameters of each coupling matrix sample at the set frequency points to construct a condition set, including: Obtain the design objectives of the microwave coupled filter, define the frequency range and the number of sampling points; Generate a frequency point set according to the frequency range and the number of sampling points to obtain the set frequency points; Calculate the normalized frequency variable of each coupling matrix sample at the set frequency point, and obtain the S parameter of each coupling matrix sample at the set frequency point based on the coupling matrix sample and the normalized frequency variable; Convert the S parameters into dB format to obtain the condition set.

[0012] In one embodiment, based on the coupling matrix samples and the normalized frequency variables, obtaining the S parameters of each coupling matrix sample at a set frequency point includes: Based on the coupling matrix samples and the normalized frequency variables, a transition matrix is ​​constructed; According to the transition matrix and different unit vectors, two linear equations are constructed and solved to obtain two transition parameters; According to the two transition parameters, the S parameters of the coupling matrix sample at the set frequency point are obtained.

[0013] In one embodiment, a frequency point set is generated according to the frequency range and the number of sampling points to obtain a set frequency point, including: Generate a frequency point set based on the frequency range and number of sampling points: ; in, ; In the formula, is the frequency point set, is the minimum value of the frequency range, is the maximum value of the frequency range, is the step length, For the frequency points, is the number of sampling points; According to the frequency point set, the set frequency point is obtained.

[0014] In one embodiment, the sample set and the condition set are used as inputs of the diffusion model, and the diffusion model is trained to learn the mapping relationship between the coupling parameters and the S parameters, thereby obtaining a trained diffusion model, including: Adding noise to each set of coupling parameters in the sample set to generate noise-added parameters; The noise adding parameters, S parameters in the condition set and time steps are input into the diffusion model, and the diffusion model is trained. The training goal is to predict the added noise and minimize the mean square error between the predicted noise and the real noise, so as to learn the mapping relationship between the coupling parameters and the S parameters, so that the diffusion model can accurately predict the noise under different conditions and time steps, and obtain a trained diffusion model.

[0015] In one embodiment, multiple sets of initial coupling parameters are obtained according to multiple model outputs, including: The model output is the predicted noise, with the standard Gaussian noise as the initial noise; According to the predicted noise and the initial noise, calculate the intermediate state at each time step, perform conditional reverse generation and execute it multiple times, so as to map the added noise back to the coupling parameters through step-by-step denoising, and obtain a set of initial coupling parameters.

[0016] In one embodiment, according to the predicted noise and the initial noise, calculate the noise addition parameters of the intermediate state at each time step, including: ; wherein, ; ; In the formula, is the noise addition parameter at time step , is the noise addition parameter at time step , is the time step, is the noise weight of each time step, is the accumulation of the noise weights, is the time step count, is the predicted noise output by the diffusion model, is the diffusion model, is the S parameter in the conditional set, is the time step in the noise term, is used to add random noise to the intermediate state.

[0017] In one embodiment, input each group of initial coupling parameters into the forward model respectively to obtain the corresponding S parameter of each group of initial coupling parameters; calculate the mean square error between the target S parameter and the corresponding S parameter, and perform screening according to the set threshold to obtain multiple groups of preferred coupling parameters, including: Input each group of initial coupling parameters into the forward model respectively, calculate the frequency response of the filter, and obtain the corresponding S parameter of each group of initial coupling parameters; Calculate the mean square error between the target S parameter and the corresponding S parameter; Screen multiple groups of initial coupling parameters with the mean square error less than the set threshold as the preferred coupling parameters.

[0018] In one embodiment, according to the preferred coupling parameters with the smallest mean square error in each cluster, obtain the coupling parameter distribution, including: Use the preferred coupling parameter with the smallest mean square error in each cluster as the coupling parameter of this cluster; Use the set of coupling parameters of all clusters as the coupling parameter distribution.

[0019] The reverse modeling method of the above microwave coupled filter is based on a two-channel conditional diffusion model. An ideal coupling matrix is generated conditionally, and parameter space sampling is performed within a set tolerance range. In the sampling stage, the model generates the corresponding coupling matrix parameter distributions multiple times by inputting the same S-parameters and noises with different distributions, and constructs a sample set of coupling parameters and the corresponding S-parameters. The sample set is input into the diffusion model for training to enable the model to learn the mapping relationship between S-parameters and coupling parameters. During the training process, the model takes the S-parameters as the conditional input, gradually adds noise to the coupling matrix parameters, and captures the non-uniqueness characteristics in the inverse problem by predicting the error of the noise. For the target S-parameters, multiple sets of candidate coupling parameters are generated through multiple predictions. The actual S-parameters corresponding to the predicted parameters are calculated using the forward model, and the data set with an error less than the set threshold is screened out through the mean square error. High-dimensional space clustering is performed on the screened parameter set, and the parameters with the minimum mean square error, that is, high-precision parameters, are extracted from each cluster, thereby realizing the accurate modeling of the coupling matrix parameters under the target S-parameters.

[0020] This method effectively utilizes the generation ability of the diffusion model, avoids the complex manual grouping and optimization processes in traditional methods, significantly improves the automation level of reverse modeling, effectively solves the multi-solution problem, that is, the non-uniqueness problem, in the reverse modeling of traditional microwave coupled filters, and improves the design efficiency and accuracy. It can quickly and accurately extract the coupling matrix parameters, providing an efficient and reliable solution for the design and optimization of microwave filters. Brief Description of the Drawings

[0021] Figure 1 It is a schematic flowchart of a reverse modeling method for a microwave coupled filter in an embodiment; Figure 2 It is a graph showing the change trend of the loss function curve during the training process in a specific embodiment; Figure 3 It is a graph of the target S-parameters in a specific embodiment; Figure 4 It is a comparison schematic diagram of the target curve and the real curve of the S11 parameter of prediction sample 1 in a specific embodiment; Figure 5 It is a comparison schematic diagram of the target curve and the real curve of the S11 parameter of prediction sample 2 in a specific embodiment; Figure 6 It is a comparison schematic diagram of the target curve and the real curve of the S11 parameter of prediction sample 3 in a specific embodiment; Figure 7 It is a comparison schematic diagram of the target curve and the real curve of the S21 parameter of prediction sample 1 in a specific embodiment; Figure 8Schematic diagram of the comparison between the target curve and the true curve of the S21 parameter of prediction sample 2 in a specific embodiment; Figure 9 Schematic diagram of the comparison between the target curve and the true curve of the S21 parameter of prediction sample 3 in a specific embodiment; Figure 10 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0022] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0023] In addition, in the present application, descriptions such as "first" and "second" are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "multiple groups" is at least two groups, such as two groups, three groups, etc., unless otherwise specifically defined.

[0024] In the present application, unless otherwise clearly defined and limited, terms such as "connection" and "fixation" should be understood in a broad sense. For example, "fixation" may be a fixed connection, a detachable connection, or integrated; it may be a mechanical connection, an electrical connection, a physical connection or a wireless communication connection; it may be directly connected, or indirectly connected through an intermediate medium, and may be the communication inside two elements or the interaction relationship between two elements, unless otherwise clearly limited. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific circumstances.

[0025] In addition, the technical solutions between various embodiments of the present application can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present application.

[0026] The present application provides a reverse modeling method for a microwave coupled filter, as Figure 1 shown in the flow schematic diagram. In one embodiment, it includes: Step 101, obtain the design objective of the microwave coupled filter and define an ideal coupling matrix.

[0027] In this step, a coupling matrix includes a set of coupling parameters, and each coupling parameter is a variable in the coupling matrix.

[0028] As for how to define an ideal coupling matrix according to the design objectives of the microwave coupling filter, it belongs to the prior art and will not be elaborated here.

[0029] Step 102: Set a coupling matrix space according to the ideal coupling matrix, and perform sampling in the coupling matrix space to obtain multiple coupling matrix samples so as to construct a sample set; calculate the S-parameters of each coupling matrix sample at the set frequency points to construct a condition set.

[0030] Specifically: Set a tolerance range for each variable in the ideal coupling matrix to obtain the value space of the variable, which serves as the coupling matrix space; perform Latin hypercube sampling in the coupling matrix space, and use a set of sampled coupling parameters as a sample to obtain multiple coupling matrix samples; construct a sample set according to the multiple coupling matrix samples; Obtain the design objectives of the microwave coupling filter, and define the frequency range and the number of sampling points; generate a frequency point set according to the frequency range and the number of sampling points to obtain the set frequency points; calculate the normalized frequency variables of each coupling matrix sample at the set frequency points, and obtain the S-parameters of each coupling matrix sample at the set frequency points based on the coupling matrix samples and the normalized frequency variables; convert the S-parameters into dB format to obtain the condition set.

[0031] More specifically: The ideal coupling matrix is: .

[0032] The ideal coupling matrix has symmetry. Set a tolerance range for each variable in the ideal coupling matrix to obtain the high-dimensional value space of the variable , which serves as the coupling matrix space; Perform Latin hypercube sampling in the coupling matrix space, and use a set of sampled coupling parameters as a sample to obtain multiple (N) coupling matrix samples ; construct a sample set according to the multiple coupling matrix samples; Obtain the design objectives of the microwave coupling filter, and define the frequency range and the number of sampling points; Generate a frequency point set according to the frequency range and the number of sampling points: ; wherein, ; In the formula, is a set of frequency points, is the minimum value of the frequency range, is the maximum value of the frequency range, that is, the frequency range is expressed as ; is the step size, is the th frequency point, is the number of sampling points; Based on the set of frequency points, the set frequency points are obtained; Calculate the normalized frequency variable of each coupling matrix sample at the set frequency points: ; In the formula, is the normalized frequency variable, is the center frequency of the filter, is the normalized bandwidth; Based on the coupling matrix sample and the normalized frequency variable, construct a transition matrix: ; Among them, ; In the formula, is the transition matrix, is the coupling matrix sample, is the identity matrix, is the port matching matrix, is the imaginary unit, is the loss term, is the normalized quality factor, indicating the energy loss characteristic of the system; According to the transition matrix and different unit vectors, construct two linear equations and solve them to obtain two transition parameters: ; In the formula, and are two different transition parameters, and are two different unit vectors; According to the two transition parameters, obtain the S parameters of the coupling matrix sample at the set frequency points: ; ;

[0033] In the formula, is the parameter, is the amplitude of the outgoing wave at the exit of port 1, is the parameter, is the amplitude of the outgoing wave at the exit of Port 2; Convert the S-parameters to dB format to obtain a set of conditions:

[0034] ; wherein, is the parameter in dB format, is the parameter in dB format.

[0035] In this step, perturb the ideal coupling matrix to obtain the corresponding S-parameters.

[0036] How to set the tolerance range for variables and Latin Hypercube Sampling (LHS) both belong to the prior art.

[0037] The number of coupling matrix samples can be determined according to actual needs and is not limited here. For example, 60,000 coupling matrix samples are obtained by sampling.

[0038] Step 103: Use the sample set and the condition set as the input of the diffusion model, and train the diffusion model to learn the mapping relationship between the coupling parameters and the S-parameters, so as to obtain a trained diffusion model.

[0039] Specifically: Add noise to each group of coupling parameters in the sample set to generate noisy parameters; Input the noisy parameters, the S-parameters in the condition set, and the time step into the diffusion model, and train the diffusion model. The training objective is to predict the added noise and minimize the mean square error between the predicted noise and the true noise, so as to learn the mapping relationship between the coupling parameters and the S-parameters, and make the diffusion model accurately predict the noise under different conditions and time steps, and obtain a trained diffusion model.

[0040] More specifically: Add Gaussian noise to each group of coupling parameters in the sample set to generate noisy parameters. Through this forward diffusion process, gradually transform into data close to the standard normal distribution: ; wherein, is the noisy parameter at time step , is the coupling parameter (the coupling parameter without added noise), is the noise weight related to time step , controlling the intensity of the noise to increase with time; is the Gaussian noise; Input the noise addition parameter, the S parameter in the conditional set, and the time step into the diffusion model to achieve conditional generation: ; In the formula, is the predicted noise output by the diffusion model; is the diffusion model; is the time step 's noise addition parameter; is the S parameter in the conditional set, providing conditional constraints for the model; is the time step, used to guide the model to learn the features of different diffusion stages; Train the diffusion model, and the training objective is to predict the added noise and minimize the mean square error between the predicted noise and the true noise: ; In the formula, is the loss function of the mean square error, is the average loss under the current random variable distribution; By continuously optimizing the model parameters , to learn the mapping relationship between the coupling parameter and the S parameter, so that the diffusion model can accurately predict different conditions and time step under the noise, so as to master the distribution characteristics of the noise-added data and obtain a trained diffusion model.

[0041] In this step, the diffusion model can adopt a two-channel conditional diffusion model to make the trained diffusion model have stronger fitting ability and better performance.

[0042] Step 104, obtain the target S parameter, and input the target S parameter into the trained diffusion model multiple times to obtain multiple model outputs corresponding to the target S parameter; according to the multiple model outputs, obtain multiple groups of initial coupling parameters.

[0043] Specifically: Obtain the target S parameter. How to obtain it is prior art. For example, the dataset includes a training set and a validation set. The training set is used for training, and the validation set is used for validation. Randomly select the target S parameter from the validation set; Input the target S parameter into the trained diffusion model multiple times to obtain multiple model outputs corresponding to the target S parameter; The model output is the predicted noise, with the standard Gaussian noise as the initial noise. That is to say, the reverse generation process starts from the standard Gaussian noise, where T represents the number of time steps in the diffusion process. At initialization is completely random Gaussian noise; Calculate the noise addition parameter of the intermediate state at each time step based on the predicted noise and the initial noise: ; where, ; ;

[0044] In the formula, is the noise addition parameter at time step , which is also the noise addition parameter of the intermediate state and can be restored to the un-noised coupling parameter through step-by-step iteration; is the noise addition parameter at time step ; is the time step, is the noise weight related to time step , that is, the noise weight of each time step, is the accumulation of the noise weights, is the time step count, is the predicted noise output by the diffusion model, is the diffusion model, is the S parameter in the conditional set, is the noise term in time step , which is used to add random noise to the intermediate state to prevent the model from deterministically converging to a single state; Calculate the intermediate state at each time step, perform the conditional reverse generation process and execute it multiple times (e.g., 20 times), generating a sample of the coupling matrix parameter each time, so as to map the added noise back to the coupling parameter through step-by-step denoising, thereby capturing the multi-solution characteristics and obtaining a set of initial coupling parameters: ; In the formula, is the first set of predicted coupling parameters, is the second set of predicted coupling parameters, is the Nth set of predicted coupling parameters.

[0045] In this step, input the target S parameter as conditional data into the trained diffusion model, and use the reverse generation ability of the diffusion model to gradually predict the corresponding coupling parameters.

[0046] Input the target S parameter into the trained diffusion model, obtain a model output corresponding to the target S parameter, and obtain a set of initial coupling parameters; input multiple times, then obtain multiple sets of initial coupling parameters.

[0047] Step 105: Input each set of initial coupling parameters into the forward model to obtain the corresponding S-parameters of each set of initial coupling parameters; calculate the mean square error between the target S-parameters and the corresponding S-parameters, and perform screening according to the set threshold to obtain multiple sets of optimal coupling parameters.

[0048] Specifically: Input each set of initial coupling parameters into the forward model, calculate the frequency response of the filter, and obtain the corresponding S-parameters of each set of initial coupling parameters; Calculate the mean square error between the target S-parameters and the corresponding S-parameters: ; In the formula, is the mean square error, is the total number of frequency points, is the corresponding S-parameter (which is also the actual S-parameter), is the target S-parameter, is the set of frequency points; Screen multiple sets of initial coupling parameters with a mean square error less than the set threshold (such as 0.1) as the optimal coupling parameters:

[0049] In the formula, is the optimal coupling parameter.

[0050] In this step, the screening result retains the coupling parameters that highly match the target S-parameters and removes the data that does not meet the design requirements.

[0051] As for the forward model, it belongs to the prior art and will not be elaborated here.

[0052] Step 106: Cluster the multiple sets of optimal coupling parameters, and obtain the coupling parameter distribution according to the optimal coupling parameter with the smallest mean square error in each cluster to realize the inverse modeling of the microwave coupling filter.

[0053] Specifically: Perform high-dimensional space clustering on the multiple sets of optimal coupling parameters to obtain multiple clusters. Each cluster represents a similar distribution in the parameter space to find all different possibilities through clustering; Let the clustering result be where is the th cluster; use the optimal coupling parameter with the smallest mean square error in each cluster as the coupling parameter of this cluster: ; In the formula, is the coupling parameter of the cluster; Take the set of coupling parameters of all clusters as the coupling parameter distribution under the target S-parameters: ; Thus, the inverse modeling of the microwave coupled filter is realized.

[0054] In this step, clustering belongs to the prior art.

[0055] The above inverse modeling method of the microwave coupled filter is based on a two-channel conditional diffusion model. By conditionally generating an ideal coupling matrix and sampling the parameter space within a set tolerance range, in the sampling stage, the model generates the corresponding coupling matrix parameter distributions multiple times by inputting the same S-parameters and noises with different distributions, and constructs a sample set of coupling parameters and the corresponding S-parameters; the sample set is input into the diffusion model for training to enable the model to learn the mapping relationship between S-parameters and coupling parameters. During the training process, the model takes the S-parameters as the conditional input, gradually adds noise to the coupling matrix parameters, and captures the non-uniqueness characteristics in the inverse problem by predicting the error of the noise; for the target S-parameters, multiple groups of candidate coupling parameters are generated through multiple predictions; the actual S-parameters corresponding to the predicted parameters are calculated using the forward model, and the data set with an error less than the set threshold is selected through the mean square error; the screened parameter set is clustered in a high-dimensional space, and the parameter with the minimum mean square error, that is, the high-precision parameter, is extracted from each cluster, thus realizing the accurate modeling of the coupling matrix parameters under the target S-parameters.

[0056] This method effectively utilizes the generation ability of the diffusion model, avoids the complex manual grouping and optimization processes in traditional methods, significantly improves the automation level of inverse modeling, effectively solves the multi-solution problem, that is, the non-uniqueness problem, in traditional inverse modeling of microwave coupled filters, and improves the design efficiency and accuracy. It can quickly and accurately extract the coupling matrix parameters, providing an efficient and reliable solution for the design and optimization of microwave filters.

[0057] It should be noted that in this application, the diffusion model is applied to the inverse modeling of microwave filters. The core lies in that during the training stage, the transformation is made from capturing the non - linear relationship between electrical parameters (S - parameters) and geometric parameters to capturing the non - linear relationship between electrical parameters as conditional information and the geometric parameters with added noise and the denoised geometric parameters. Through this transformation, the diffusion model not only learns how to model the relationship between electrical parameters and geometric parameters, but also effectively captures this complex non - linear mapping through the noise perturbation process. In the diffusion model, it is mainly divided into three stages: forward diffusion, model training, model prediction and reverse generation. In the forward diffusion stage, noise is gradually added to the geometric parameters, and finally they are transformed into noise samples close to the Gaussian distribution. In the model training stage, the diffusion model learns how to guide the noise removal process according to electrical parameters by minimizing the error between the predicted noise and the real noise. In the model prediction and reverse generation stage, through the step - by - step denoising process, the geometric parameters matching the electrical parameters are recovered from the noise, and at the same time, multi - solution generation is achieved through multiple reverse generations, thus effectively solving the non - uniqueness problem in inverse modeling.

[0058] It should be understood that although Figure 1 the steps in the flowchart of Figure 1 are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise clearly stated in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover,

[0059] In a specific embodiment, according to the filter design objective, an ideal coupling matrix is defined as a benchmark. The specific indicators of the filter are a center frequency of 1.75 GHz, a bandwidth of 100 MHz, and a return loss of 20 dB. Its 4 - order ideal coupling matrix is: .

[0060] Set the coupling matrix space, and the tolerance range of each coupling parameter is ±0.8, where: ; ; .

[0061] Based on this space, 60,000 coupled matrix samples are uniformly selected using Latin Hypercube Sampling (LHS) to construct a sample set. The frequency range is set to ±10% of the central frequency (i.e., 1.575 GHz to 1.925 GHz), and the number of sampling points for frequency is 20. The S 11 parameters and S 21 parameters are calculated, and they are converted to dB format to construct a condition set.

[0062] The sample set and the corresponding condition set are input into the diffusion model for training. By adding Gaussian noise to the original coupling parameters to generate noisy parameters , and the noisy parameters , the S parameters in the condition set and the time step are jointly input into the diffusion model to minimize the mean square error between the predicted noise output by the diffusion model and the true noise . Taking this as the goal, the model parameters are optimized and an inverse mapping relationship between the S parameters and the coupling parameters is established. The change trend graph of the loss function curve during the training process is as shown in . It can be seen that the model is converging and can more accurately learn the "noise distribution characteristics of the coupling coefficient from X0 to X Figure 2 at a given time step t and conditional information (such as S11, S21)". That is to say, the model has captured the noise evolution law in this diffusion process and can establish an effective mapping relationship between the input data and its corresponding noise. t

[0063] As shown in Figure 3 the target S parameter curve graph, taking the target S parameters (including the and response curves of 20 frequency points) as conditional information , combined with the noise attenuation weight coefficient at the time step , input into the trained diffusion model. Starting from the initial Gaussian noise , based on the conditional information and the time step , through step-by-step iteration, the random noise is mapped to the coupled matrix parameter space that meets the target S parameter constraints. After 20 independent predictions are executed, 20 sets of coupled matrix parameters are generated.

[0064] According to the generated 20 sets of coupled matrix parameters, the actual S parameters and . By comparing with the target S-parameters and , calculate the mean square error: .

[0065] Filter out the coupling matrix parameters that satisfy , eliminate the samples with excessive errors, and finally form an effective parameter set for subsequent clustering optimization. There are 12 effective coupling matrix samples after screening.

[0066] Perform high-dimensional space clustering on the filtered coupling matrix parameters, and select a set of parameters with the minimum mean square error in each cluster to obtain the final distribution of the coupling matrix parameters under the target S-parameters. The number of finally determined clusters is 3, and the three selected optimal samples are respectively:

[0067] The comparison between the corresponding target S-parameter curve and the true S-parameter curve is as Figures 4 to 9 shown. It can be seen that within the range of and at 20 frequency points, although there are certain differences in the three sets of calculated coupling matrix parameters, they all better meet the requirements of the target S-parameter points. That is to say, each sample meets the requirements, but the distribution of each sample is different, effectively solving the problem of reverse modeling of microstrip filters; in addition, the predicted S-parameter curve (the curve with squares) and the target / true data (the curve with dots) have quite high consistency throughout the frequency band, which indicates that the samples predicted by the model can accurately fit the main resonance characteristics and bandwidth changes presented by the target S-parameter curve and maintain a small error at each frequency point compared with the true measurement results.

[0068] This application also provides a reverse modeling device for a microwave coupled filter. In one embodiment, it includes: a first module, a second module, a third module, a fourth module, a fifth module, and a sixth module, where: The first module is used to obtain the design objective of the microwave coupled filter, obtain the target S-parameters, and define an ideal coupling matrix; The second module is used to set a coupling matrix space according to the ideal coupling matrix, sample in the coupling matrix space to obtain a plurality of coupling matrix samples to construct a sample set; calculate the S-parameters of each coupling matrix sample at the set frequency points to construct a condition set; The third module is used to use the sample set and the condition set as the input of the diffusion model, train the diffusion model to learn the mapping relationship between the coupling parameters and the S-parameters, and obtain a trained diffusion model; A fourth module is configured to input the target S-parameters into the trained diffusion model multiple times to obtain multiple model outputs corresponding to the target S-parameters; and obtain multiple sets of initial coupling parameters according to the multiple model outputs. A fifth module is configured to input each set of initial coupling parameters into the forward model respectively to obtain the corresponding S-parameters of each set of initial coupling parameters; calculate the mean square error between the target S-parameters and the corresponding S-parameters, and perform screening according to a set threshold to obtain multiple sets of optimal coupling parameters. A sixth module is configured to cluster the multiple sets of optimal coupling parameters, and obtain a coupling parameter distribution according to the optimal coupling parameter with the minimum mean square error in each cluster, so as to realize the inverse modeling of the microwave coupled filter.

[0069] For the specific limitations of an inverse modeling device of a microwave coupled filter, reference can be made to the limitations of an inverse modeling method of a microwave coupled filter in the foregoing text, which will not be elaborated herein. Each module in the above device can be implemented in whole or in part by software, hardware, and their combination. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0070] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structural diagram can be as Figure 10 shown. The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements an inverse modeling method of a microwave coupled filter. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0071] Those skilled in the art can understand that Figure 10 the structure shown in

[0072] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method in the above embodiment are implemented.

[0073] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the above embodiment are implemented.

[0074] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it may include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application may include non-volatile and / or volatile memories. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0075] The content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0076] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0077] The above-described embodiments only represent several implementation manners of the present application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for inverse modeling of a microwave coupled filter, characterized in that: include: Obtain the design objectives of the microwave coupled filter and define the ideal coupling matrix; According to the ideal coupling matrix, a coupling matrix space is set, and sampling is performed in the coupling matrix space to obtain a plurality of coupling matrix samples to construct a sample set; Calculate the S parameters of each coupling matrix sample at a set frequency point to construct a condition set; The sample set and the condition set are used as the input of the diffusion model, and the diffusion model is trained to learn the mapping relationship between the coupling parameters and the S parameters, thereby obtaining a trained diffusion model; Obtain target S parameters, and input the target S parameters into the trained diffusion model multiple times to obtain multiple model outputs corresponding to the target S parameters; obtain multiple sets of initial coupling parameters according to the multiple model outputs; Input each set of initial coupling parameters into the forward model respectively to obtain the corresponding S parameters of each set of initial coupling parameters; calculate the mean square error between the target S parameters and the corresponding S parameters, and screen them according to the set threshold to obtain multiple sets of optimal coupling parameters; Multiple groups of optimal coupling parameters are clustered, and coupling parameter distribution is obtained according to the optimal coupling parameter with the smallest mean square error in each cluster to achieve inverse modeling of the microwave coupling filter.

2. The inverse modeling method of a microwave coupled filter according to claim 1, characterized in that: According to the ideal coupling matrix, a coupling matrix space is set, and sampling is performed in the coupling matrix space to obtain multiple coupling matrix samples to construct a sample set, including: Setting a tolerance range for each variable in the ideal coupling matrix to obtain the value space of the variable as the coupling matrix space; Latin hypercube sampling is performed in the coupling matrix space, and a set of sampled coupling parameters is taken as a sample to obtain a plurality of coupling matrix samples; A sample set is constructed according to a plurality of coupling matrix samples.

3. The inverse modeling method of a microwave coupled filter according to claim 2, characterized in that: Calculate the S parameters of each coupling matrix sample at a set frequency point to construct a set of conditions, including: Obtain the design objectives of the microwave coupled filter and define the frequency range and number of sampling points; Generate a frequency point set according to the frequency range and the number of sampling points to obtain a set frequency point; Calculate the normalized frequency variable of each coupling matrix sample at the set frequency point, and obtain the S parameter of each coupling matrix sample at the set frequency point based on the coupling matrix sample and the normalized frequency variable; Convert the S parameters into dB format to obtain the condition set.

4. The inverse modeling method of a microwave coupled filter according to claim 3, characterized in that: Based on the coupling matrix samples and the normalized frequency variables, the S parameters of each coupling matrix sample at the set frequency point are obtained, including: Based on the coupling matrix samples and the normalized frequency variables, a transition matrix is ​​constructed; According to the transition matrix and different unit vectors, two linear equations are constructed and solved to obtain two transition parameters; According to the two transition parameters, the S parameters of the coupling matrix sample at the set frequency point are obtained.

5. The inverse modeling method of a microwave coupled filter according to claim 4, characterized in that: Generate a frequency point set based on the frequency range and the number of sampling points to obtain the set frequency point, including: Generate a frequency point set based on the frequency range and number of sampling points: in, In the formula, is the frequency point set, is the minimum value of the frequency range, is the maximum value of the frequency range, is the step length, For the frequency points, is the number of sampling points; According to the frequency point set, the set frequency point is obtained.

6. The inverse modeling method of a microwave coupled filter according to any one of claims 1 to 5, characterized in that: The sample set and the condition set are used as inputs of the diffusion model, and the diffusion model is trained to learn the mapping relationship between the coupling parameters and the S parameters, and a trained diffusion model is obtained, including: Adding noise to each set of coupling parameters in the sample set to generate noise-added parameters; The noise adding parameters, S parameters in the condition set and time steps are input into the diffusion model, and the diffusion model is trained. The training goal is to predict the added noise and minimize the mean square error between the predicted noise and the real noise, so as to learn the mapping relationship between the coupling parameters and the S parameters, so that the diffusion model can accurately predict the noise under different conditions and time steps, and obtain a trained diffusion model.

7. The inverse modeling method of a microwave coupled filter according to any one of claims 1 to 5, characterized in that: Based on multiple model outputs, multiple sets of initial coupling parameters are obtained, including: The model output is the predicted noise, with standard Gaussian noise as the initial noise; According to the predicted noise and the initial noise, the intermediate state of each time step is calculated, and the conditional reverse generation is performed and executed multiple times to map the added noise back to the coupling parameters through step-by-step denoising to obtain a set of initial coupling parameters.

8. The inverse modeling method of a microwave coupled filter according to claim 7, characterized in that: According to the predicted noise and initial noise, the noise parameters of the intermediate state of each time step are calculated, including: in, In the formula, is the time step The noise parameter, is the time step The noise parameter, is the time step, For each time step the noise weight is is the accumulation of noise weight, is the time step count, is the predicted noise output by the diffusion model, is the diffusion model, is the S parameter in the condition set, is the time step The noise term in Used to add random noise to intermediate states.

9. The inverse modeling method of a microwave coupled filter according to any one of claims 1 to 5, characterized in that: Input each set of initial coupling parameters into the forward model respectively to obtain the corresponding S parameters of each set of initial coupling parameters; Calculate the mean square error between the target S parameter and the corresponding S parameter, and screen them according to the set threshold to obtain multiple sets of optimal coupling parameters, including: Input each set of initial coupling parameters into the forward model respectively, calculate the frequency response of the filter, and obtain the corresponding S parameters of each set of initial coupling parameters; Calculate the mean square error between the target S parameter and the corresponding S parameter; Multiple groups of initial coupling parameters whose mean square errors are less than a set threshold are screened as preferred coupling parameters.

10. The inverse modeling method of a microwave coupled filter according to any one of claims 1 to 5, characterized in that: According to the optimal coupling parameters with the smallest mean square error in each cluster, the coupling parameter distribution is obtained, including: The optimal coupling parameter with the smallest mean square error in each cluster is used as the coupling parameter of the cluster; The set of coupling parameters of all clusters is taken as the coupling parameter distribution.

Images