Rapid intelligent design method and system for surface acoustic wave filter

Abstract

The invention provides a rapid intelligent design method and system for a surface acoustic wave filter, and the method comprises the steps: building a physical model of a surface acoustic wave resonator based on electromagnetic simulation software, carrying out the calculation to obtain a corresponding admittance response, and obtaining a series of surface acoustic wave resonator admittance responses and key structure parameters for constructing a target surface acoustic wave filter; preprocessing the collected sample data set to form an input vector of a deep learning model; constructing a reversely designed neural network model, learning a mapping relation between structure parameters and admittance response, and predicting the structure parameters of the serial and parallel surface acoustic wave resonators; and in combination with a COM model, the admittance of the series resonator and the admittance of the parallel resonator are converted into an S matrix, so that a filtering passband is obtained. According to the method, the problems of high calculation cost and accuracy caused by complex acoustic characteristics and material parameters are avoided, the defects that a traditional numerical simulation method is time-consuming, tedious and complex in circuit topology process are overcome, and the rapid and intelligent design requirement of the surface acoustic wave filter is met.

Description

Technical Field

[0001] The invention belongs to the technical field of surface acoustic wave devices and artificial intelligence, and in particular relates to a fast intelligent design method and system for a surface acoustic wave filter. Background Art

[0002] Surface acoustic wave filters have the advantages of miniaturization, strong anti-interference ability and easy manufacturing, and are widely used in wireless communications, radar systems, and biosensors. However, traditional surface acoustic wave device design usually requires a lot of electromagnetic simulation, complex circuit topology, expensive process verification, and professional engineer experience, resulting in a very cumbersome design process and a lot of time and hardware costs.

[0003] Traditional methods mainly include fitting method and simulation method. Among them, the fitting method is a widely used design method for surface acoustic wave filters, which requires a lot of experiments, repeated verification and adjustment, and finally makes the characteristics of the filter meet the requirements. The simulation method considers the acoustic characteristics of the surface acoustic wave device and the structure and material of the physical model, and then simulates the surface acoustic wave filter, but with the increase of complexity, the consumption of computing resources and time also increases accordingly.

[0004] As deep learning technology continues to improve, neural networks have demonstrated superior data learning capabilities and are widely used in multiple fields. Deep learning algorithms can be applied to different types of data and tasks, including images, text, and audio, and can process massive amounts of data to extract useful information. In the design of surface acoustic wave filters, deep learning technology can break through the limitations of traditional methods, achieve more efficient and accurate designs, and significantly reduce the design costs of surface acoustic wave filters.

[0005] In view of this, it is very meaningful to propose a fast and intelligent design method and system for surface acoustic wave filters. Summary of the invention

[0006] In order to solve the problems that the traditional surface acoustic wave filter design process is cumbersome, complicated and inefficient, the present invention provides a fast and intelligent design method and system for a surface acoustic wave filter to solve the above-mentioned technical defects.

[0007] In a first aspect, the present invention provides a fast and intelligent design method for a surface acoustic wave filter, the method comprising the following steps:

[0008] S1. Based on electromagnetic simulation software, a physical model of the surface acoustic wave resonator is established, the corresponding admittance response is calculated, the required sample data set is randomly generated, and a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing the target surface acoustic wave filter are obtained;

[0009] S2. Preprocessing the collected sample data set, including data cleaning and normalization, to form an input vector of a deep learning model;

[0010] S3, dividing the sample data set into a training set and a validation set, applying the deep learning model on the training set, building a reverse design neural network model, learning the mapping relationship between structural parameters and admittance response, and evaluating the algorithm performance on the validation set, and predicting the structural parameters of series and parallel surface acoustic wave resonators based on the trained deep learning model;

[0011] S4. Combined with the COM model, the COM parameters are extracted according to the admittance response of the series and parallel surface acoustic wave resonators. The SAW single-ended resonators are combined in a trapezoidal structure, and the admittance of the series and parallel resonators is converted into an S matrix to obtain the filter passband, thereby realizing the intelligent and rapid design of the surface acoustic wave filter.

[0012] This method uses electromagnetic simulation software to calculate the corresponding admittance response by random combination within a given periodic layered electrode-substrate structure, and uses these data for training and verification of deep learning algorithms, so that the neural network can learn the complex mapping relationship between electrode structure and material, substrate width and admittance response.

[0013] Preferably, the structural parameters of the surface acoustic wave resonator physical model established in S1 include the number of layers, material, geometry, metallization rate, width, thickness of the periodic layered electrode structure and the material and width of the substrate structure.

[0014] Further preferably, the number of layers of the periodic layered electrode structure is any number, the material of the periodic layered electrode structure is any one of titanium, aluminum, copper, gold, and silver metals or a composite thereof, and the geometric shape of the periodic layered electrode structure is a rectangle or a trapezoid.

[0015] Further preferably, the layered electrode structure is a three-layer metal layered nanostructure with a trapezoidal geometric shape. The three-layer electrode structure material is titanium aluminum titanium, the substrate material is a 42°YX tangential lithium tantalate piezoelectric crystal 42°YX-LT, the width of the substrate is λ, the metallization rates of the two electrodes are r1 and r2, wherein the metallization rate r=w / λ, w is the electrode width, and the thicknesses of the three-layer electrodes from top to bottom are t1, t2, and t3, respectively.

[0016] Preferably, the deep learning model in S3 includes an input layer, a hidden layer and an output layer in sequence to establish a complex mapping relationship between structural parameters and admittance response spectra;

[0017] The input layer receives the admittance response spectrum of the surface acoustic wave resonator as an input feature, and the output layer predicts the structural parameters and electrode materials of the corresponding resonator; the length of the input layer is 800, corresponding to the number of sampling points on the admittance response spectrum;

[0018] The hidden layer includes 8 fully connected layers, each of which contains multiple neurons, and each neuron performs nonlinear mapping by weighting and processing the input of the previous layer through an activation function;

[0019] The output layer is composed of several fully connected layers, and the output length is 9, that is, the output substrate width λ, the electrode widths w1 and w2, the electrode thicknesses t1, t2 and t3, and the material properties of the three-layer electrodes.

[0020] Preferably, in S3, the training set is input into a deep learning model to establish a mapping relationship between structural parameters and admittance response, and the training method of the deep learning model specifically includes:

[0021] Initialize the parameters, set the batch size to 128, the total number of epochs to 500, the initial learning rate to 0.0001, the learning rate decay strategy to reduce the learning rate by 60% every 30 epochs of training, use Adam as the gradient descent optimizer, use mean square error MSE as the loss function, use ReLU function as the activation function of the entire network model, and use the chain rule to reversely transfer the gradient to each neuron to optimize the parameters in the neuron.

[0022] Preferably, the reverse-engineered neural network includes a fully connected neural network, a convolutional neural network, a Transformer, a residual neural network, a recurrent neural network, a generative adversarial network, and a neural network of an autoencoder.

[0023] Preferably, the method further comprises: using electromagnetic simulation software to calculate the admittance response of the full-period model of the surface acoustic wave resonator combined with the layered electrode and the substrate structure, using the predicted structural parameters, using the COM model to fit the full-period surface acoustic wave resonator admittance, setting the period to 89, and extracting the wave velocity V of the surface grazing body wave SSBW based on the predicted admittance response spectrum and structural parameters. B , resonant frequency f r and the anti-resonance frequency f ar , the COM parameter calculation formula is:

[0024] Δ=ω / V B -2π / P I

[0025]

[0026] Δ V =-η2 / 2

[0027] κ=|ε|(η+|ε| / 2)

[0028] κ B =-|ε| 2 η / (η+2|ε|)

[0029] Where Δ is the normalized frequency, ω is the angular frequency, and P I represents the propagation period characteristics of surface acoustic waves in specific structures, such as periodic layered electrodes and substrate structures, ε represents the parameter of SAW reflection, η is the parameter representing the SAW penetration depth, κ and κ B are different parameters of the coupling strength between different modes of surface acoustic waves during propagation.

[0030] Preferably, in S4, based on the obtained admittance response, the SAW single-ended resonators are combined in a trapezoidal structure, and the admittances of the series and parallel resonators are converted into an S matrix. The specific calculation process is:

[0031] The filter has only one series resonator and one parallel resonator, and the [abcd] matrix of the filter is:

[0032]

[0033] Normalize the port impedance Z0, Z0 is 50Ω:

[0034]

[0035] Then transform the [ABCD] matrix into the [S] matrix:

[0036]

[0037] Among them, y s and p are parameters related to matrix transformation. The [abcd] matrix represents the matrix that uses the total voltage and total current of the output port to represent the total voltage and total current of the input port. The [ABCD] matrix represents the [abcd] matrix after impedance normalization. S11 and S22 represent reflection coefficients, which reflect the degree of signal reflection at the port. S21 and S12 represent transmission coefficients, which reflect the degree of signal transmission between ports.

[0038] In a second aspect, an embodiment of the present invention provides a fast intelligent design system for a surface acoustic wave filter, comprising:

[0039] Data simulation module: configured to establish a physical model of the surface acoustic wave resonator based on electromagnetic simulation software, calculate the corresponding admittance response, randomly generate the required sample data set, and obtain a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing the target surface acoustic wave filter;

[0040] Sample processing module: configured to pre-process the collected sample data set, including data cleaning and normalization, to form an input vector of the deep learning model;

[0041] Algorithm training module: configured to build a deep learning model, learn the mapping relationship between structural parameters and admittance curves, and evaluate the performance of the deep learning model;

[0042] Filter prediction module: configured to input the admittance responses of the series and parallel resonators to the trained deep learning model, so that the two admittance curves representing the admittance responses of the series and parallel resonators satisfy a certain relationship, that is, one of the admittance curves is the resonant frequency of the parallel resonator and the other admittance curve is the anti-resonant frequency of the series resonator, and the key structural parameters of the series-parallel resonator are obtained based on the deep learning model;

[0043] Circuit Topology Module: Configured to extract COM parameters and calculate ladder filter design performance based on COM models and predicted key structural parameters.

[0044] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0045] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0046] Compared with the prior art, the beneficial results of the present invention are:

[0047] This method can efficiently learn the mapping relationship between complex structural parameters and admittance response through a deep learning model. The trained algorithm model can quickly predict structural parameters within milliseconds. Combined with the COM model, it can realize the real-time on-demand design of surface acoustic wave filters with different filtering passband requirements, significantly improving the design efficiency and accuracy of surface acoustic wave filters. Compared with traditional design methods, the present invention avoids the high computational cost and accuracy problems caused by complex acoustic properties and material parameters, overcomes the time-consuming and tedious circuit topology process of traditional numerical simulation methods, and can quickly provide reliable reference solutions for qualified structural parameters and materials for surface acoustic wave filters with different filtering passband requirements, meeting the needs of rapid and intelligent design of surface acoustic wave filters. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated into and constitute a part of this specification. The accompanying drawings illustrate the embodiments and are used together with the description to explain the principles of the present invention. It will be easy to recognize other embodiments and many expected advantages of the embodiments because they become better understood by reference to the following detailed description. The elements of the drawings are not necessarily to scale with each other. The same reference numerals refer to corresponding similar parts.

[0049] Figure 1 is a diagram of an exemplary device architecture in which an embodiment of the present invention may be applied;

[0050] Figure 2 A schematic flow chart of a fast and intelligent design method for a surface acoustic wave filter according to an embodiment of the present invention;

[0051] Figure 3 It is a structural schematic diagram and a periodic unit schematic diagram of a physical model of a surface acoustic wave resonator according to an embodiment of the present invention;

[0052] Figure 4 A schematic diagram of the structure of a neural network model of a deep learning algorithm according to an embodiment of the present invention and a schematic diagram of a ladder filter constructed of series-parallel resonators;

[0053] Figure 5 A schematic diagram illustrating the design principle of a ladder filter according to an embodiment of the present invention;

[0054] Figure 6 A schematic diagram of a ladder filter designed by an intelligent design method according to an embodiment of the present invention;

[0055] Figure 7 A schematic diagram of the architecture of a fast and intelligent design system for a surface acoustic wave filter according to an embodiment of the present invention;

[0056] Figure 8A schematic diagram of a rapid intelligent design platform for a surface acoustic wave filter according to an embodiment of the present invention;

[0057] Fig. 9 It is a schematic diagram of the structure of a computer device suitable for implementing an electronic device of an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0059] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0060] Figure 1 An exemplary system architecture 100 is shown to which a rapid and intelligent design method for a surface acoustic wave filter or a rapid and intelligent design system for a surface acoustic wave filter according to an embodiment of the present invention can be applied.

[0061] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0062] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0063] Terminal devices 101, 102, 103 can be hardware or software. When terminal devices 101, 102, 103 are hardware, they can be various electronic devices, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, etc. When terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules (for example, software or software modules used to provide distributed services), or they can be implemented as a single software or software module. No specific limitation is made here.

[0064] The server 105 may be a server providing various services, such as a background information processing server that processes the verification request information sent by the terminal devices 101, 102, and 103. The background information processing server may analyze the received verification request information and obtain a processing result.

[0065] It should be noted that the fast intelligent design method of the surface acoustic wave filter provided in the embodiment of the present invention is generally executed by the server 105, and accordingly, the fast intelligent design system of the surface acoustic wave filter is generally set in the server 105. In addition, the fast intelligent design method of the surface acoustic wave filter provided in the embodiment of the present invention is generally executed by the terminal devices 101, 102, 103, and accordingly, the fast intelligent design system of the surface acoustic wave filter is generally set in the terminal devices 101, 102, 103.

[0066] It should be noted that the server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, used to provide distributed services), or it can be implemented as a single software or multiple software modules, which is not specifically limited here.

[0067] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided as required. In the case where the processed data does not need to be obtained remotely, the above device architecture may not include a network, but only a server or a terminal device.

[0068] First, Figure 2 It is shown that the embodiment of the present invention discloses a fast and intelligent design method for a surface acoustic wave filter, such as Figure 2 As shown, the method comprises the following steps:

[0069] S1. Based on electromagnetic simulation software, a physical model of the surface acoustic wave resonator is established, the corresponding admittance response is calculated, the required sample data set is randomly generated, and a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing the target surface acoustic wave filter are obtained;

[0070] S2. Preprocessing the collected sample data set, including data cleaning and normalization, to form an input vector of a deep learning model;

[0071] S3, dividing the sample data set into a training set and a validation set, applying the deep learning model on the training set, building a reverse design neural network model, learning the mapping relationship between structural parameters and admittance response, and evaluating the algorithm performance on the validation set, and predicting the structural parameters of series and parallel surface acoustic wave resonators based on the trained deep learning model;

[0072] S4. Combined with the COM model, the COM parameters are extracted according to the admittance response of the series and parallel surface acoustic wave resonators. The SAW single-ended resonators are combined in a trapezoidal structure, and the admittance of the series and parallel resonators is converted into an S matrix to obtain the filter passband, thereby realizing the intelligent and rapid design of the surface acoustic wave filter.

[0073] This method uses electromagnetic simulation software to calculate the corresponding admittance response by random combination within a given periodic layered electrode-substrate structure, and uses these data for training and verification of deep learning algorithms, so that the neural network can learn the complex mapping relationship between electrode structure and material, substrate width and admittance response.

[0074] In a specific embodiment, the structural diagram and periodic unit of the physical model of the surface acoustic wave resonator are as follows: Figure 3 As shown. The structural parameters of the physical model of the surface acoustic wave resonator include the number of layers, materials, geometric shapes, metallization rate, width, thickness of the periodic layered electrodes, and the materials and width of the substrate structure. The layered electrode structure is a three-layer vertical periodic alternating metal layered nanostructure, the geometric shape is a trapezoid, the three-layer electrode structure material is a titanium-aluminum-titanium composite structure from top to bottom, the substrate material is a 42°YX tangential lithium tantalate piezoelectric crystal (42°YX-LT), the width of the substrate is λ, the widths of the two electrodes are w1, w2, and the thicknesses of the three layers of electrodes are t1, t2, and t3 from top to bottom.

[0075] Optionally, the number of electrode layers can be any number.

[0076] Optionally, the electrode material may be any one of metals such as titanium, aluminum, copper, gold, silver, or a composite thereof.

[0077] Optionally, the electrode geometry may be a rectangle or a trapezoid.

[0078] Optionally, the deep learning algorithm can be a neural network such as a fully connected neural network, a convolutional neural network, a Transformer, a residual neural network, a recurrent neural network, a generative adversarial network, an autoencoder, etc.

[0079] In one embodiment, the layered electrode structure is a three-layer vertically periodically alternating metal layered nanostructure, the substrate material is a 42°YX tangential lithium tantalate piezoelectric crystal (42°YX-LT), the width of the substrate is λ, the metallization rates of the two electrodes are r1 and r2, where the metallization rate r is w / λ (w is the electrode width), and the thicknesses of the three layers of electrodes are t1, t2, and t3 from top to bottom, respectively.

[0080] In a specific embodiment, the geometric shape of the layered electrode is a trapezoid.

[0081] In a specific embodiment, the three-layer electrode structure material is a titanium-aluminum-titanium composite structure from top to bottom.

[0082] In a specific embodiment, the deep learning network is a fully connected network.

[0083] Specifically, this embodiment uses the joint simulation function of MATLAB and COMSOL to improve the efficiency of sample data generation. According to the design principle of surface acoustic wave filters, filter samples can be constructed from training samples. Figure 4 As shown, a total of 240.75 million sets of data were obtained as data for neural network training and verification.

[0084] The 24,075 collected samples were randomly divided into 19,260 training sets and 4,815 validation sets in a 4:1 ratio. The training set was used to train the deep learning algorithm, and the validation set that had not appeared in the training set was used to verify the predictive ability of the model.

[0085] In a specific embodiment, the deep learning algorithm neural network model structure is as follows Figure 4 As shown, the deep learning model includes an input layer, a hidden layer, and an output layer in sequence; the input layer receives the admittance response spectrum of the surface acoustic wave resonator as an input feature and the main task of the output layer is to predict the structural parameters and electrode materials of the corresponding resonator.

[0086] In the input layer, the admittance response spectrum of the SAW resonator is received as the input feature in this embodiment, and the main task of the output layer is to predict the structural parameters and electrode materials of the corresponding resonator. The length of the input layer is 800, corresponding to the number of sampling points on the admittance response spectrum.

[0087] The hidden layer includes 8 fully connected layers, each hidden layer contains multiple neurons and each neuron weights and processes the input of the previous layer, and then performs nonlinear mapping through the activation function. The design of these layers is to establish a complex mapping relationship between structural parameters and admittance response spectra.

[0088] The output layer consists of several fully connected layers with an output length of 9, namely λ (substrate width), w1, w2, t1, t2, t3 (electrode thickness), and the material properties of the three-layer electrode. The mean square error (MSE) of the output layer is calculated and back-propagated back to the hidden layer to update the weights, thereby minimizing the gap between the predicted output and the actual output.

[0089] Furthermore, in S3, the expected admittance response vector in the reverse design network training set is input into the reverse design network to be trained, the structural parameters are predicted, and the corresponding predicted structural parameter vector is output. The structural parameter mean square error loss is calculated according to the predicted structural parameter vector and the structural parameter vector corresponding to the training set; the corresponding reverse design network is back-propagated and the network parameters are updated according to the structural parameter mean square error loss, and the reverse design network is converged after multiple rounds of iterations.

[0090] The input layer length of the inverse neural network is 800, which is the number of spectral line sampling points corresponding to the admittance response; the hidden layer is composed of a fully connected layer, which is used to establish a complex mapping relationship between the admittance response spectrum and the structural parameters; the output layer is composed of several fully connected layers, and the output length is 9, which is λ (substrate width), w1, w2, t1, t2, t3 (electrode thickness), and the material properties of the three-layer electrode.

[0091] The training method of the deep learning model in this embodiment is as follows:

[0092] The training set is input into the deep learning model to establish the mapping relationship between the structural parameters and the admittance response. The specific method is as follows: initialize the parameters, set the batch size to 128, the total number of learning epochs to 500, the initial learning rate to 0.0001, and the learning rate decay strategy is to reduce the learning rate by 60% every 30 epochs of training. Adam is used as the gradient descent optimizer, the mean square error MSE is used as the loss function, and the ReLU function is used as the activation function of the entire network model. The gradient is reversely transferred to each neuron through the chain derivative rule to optimize the parameters in the neuron.

[0093] Specifically, in the deep learning model process, we divided 128 training samples into batches and input them into the neural network, using Adam as the gradient descent optimizer and mean square error as the loss function. After each training batch, the loss function is used to back-propagate the gradient and update the trainable parameters. After about 500 epochs, when the validation loss no longer decreases, we determine that the training of the neural network model has been completed. We use the validation set to verify the predictive performance of the deep learning algorithm model, and also use MSE to measure the error.

[0094] The ladder filter is composed of series and parallel resonators. The condition that needs to be met is that the resonant frequency of the series resonator is equal to the anti-resonant frequency of the parallel resonator. Using the predicted structural parameters, the full-cycle surface acoustic wave resonator admittance is fitted based on the COM model. The period is set to 89, the admittance response is obtained, and the wave velocity V of the surface grazing body wave (SSBW) is extracted. B , resonant frequency fr and anti-resonant frequency far, and obtain the admittance of the fitted SAW resonator. The SAW single-ended resonator is combined in a ladder structure so that the admittance response of the series and parallel resonators meets the conditions, and the admittance of the series and parallel resonators is converted into an S matrix. The multi-step ladder SAW filter design is realized.

[0095] Specifically, based on the predicted admittance response spectrum and structural parameters, the wave velocity V of the surface grazing body wave SSBW is extracted. B , resonant frequency f r and the anti-resonance frequency f ar , the COM parameter calculation formula is:

[0096] Δ=ω / V B -2π / P I

[0097]

[0098] Δ V =-η 2 / 2

[0099] κ=|ε|(η+|ε| / 2)

[0100] κ B =-|ε| 2 η / (η+2|ε|)

[0101] Where Δ is the normalized frequency, ω is the angular frequency, and P I represents the propagation period characteristics of surface acoustic waves in specific structures, such as periodic layered electrodes and substrate structures, ε represents the parameter of SAW reflection, η is the parameter representing the SAW penetration depth, κ and κ B are different parameters of the coupling strength between different modes of surface acoustic waves during propagation.

[0102] Furthermore, based on the obtained admittance response, the SAW single-ended resonators are combined in a trapezoidal structure, and the admittance of the series and parallel resonators is converted into an S matrix. The specific calculation process is:

[0103] The filter has only one series resonator and one parallel resonator, and the [abcd] matrix of the filter is:

[0104]

[0105] Normalize the port impedance Z0, Z0 is 50Ω:

[0106]

[0107] Then transform the [ABCD] matrix into the [S] matrix:

[0108]

[0109] Among them, y s and p are parameters related to matrix transformation. The [abcd] matrix represents the matrix that uses the total voltage and total current of the output port to represent the total voltage and total current of the input port; the [ABCD] matrix represents the [abcd] matrix after impedance normalization; S11 and S22 represent reflection coefficients, which reflect the degree of signal reflection at the port; S21 and...12 represent transmission coefficients, which reflect the degree of signal transmission between ports.

[0110] Figure 5 The example of a ladder filter constructed of a series resonator and a parallel resonator is shown to illustrate the design principle of the ladder filter, that is, the resonant frequency of the admittance curve of the parallel resonator is equal to the anti-resonant frequency of the admittance curve of the other series resonator.

[0111] like Figure 6 As shown, in a specific example, two admittance curves are input, where the resonant frequency of the admittance curve of the series resonator is equal to the anti-resonant frequency of the admittance curve of the series resonator, and the deep learning model gives the corresponding predicted structural parameters of the surface acoustic wave resonator, where the electrode thickness t1 of the series resonator is 18nm, t2 is 131.3nm, t3 is 10.8nm, the substrate width λ is 1930nm, the two electrode widths w1 are 387nm, w2 is 367nm, and the corresponding electrode structure material is a titanium-aluminum-titanium composite structure. The electrode thickness t1 of the parallel resonator is 16.8nm, t2 is 128.5nm, t3 is 9.3nm, the substrate width λ is 2020nm, the two electrode widths w1 are 444nm, w2 is 444nm, and the corresponding electrode structure material is a titanium-aluminum-titanium composite structure.

[0112] Based on the key structural parameters of the series-parallel resonator obtained by the deep learning model, the COM model is used to extract the COM parameters, which can quickly design the following Figure 6 A surface acoustic wave filter that meets the target requirements is shown in FIG.

[0113] Further references Figure 7 As an implementation of the methods shown in the above figures, the present application provides an embodiment of a system, which is similar to Figure 2Corresponding to the method embodiment shown, the system can be specifically applied to various electronic devices.

[0114] In a second aspect, the present invention also discloses a fast intelligent design system for surface acoustic wave filters. Figure 7 and 8 As shown, including:

[0115] Data simulation module 71: configured to establish a surface acoustic wave resonator physical model based on electromagnetic simulation software, calculate the corresponding admittance response, randomly generate the required sample data set, and obtain a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing a target surface acoustic wave filter;

[0116] Sample processing module 72: configured to pre-process the collected sample data set, including data cleaning and normalization, to form an input vector of a deep learning model;

[0117] Algorithm training module 73: configured to build a deep learning model, learn the mapping relationship between structural parameters and admittance curves, and evaluate the performance of the deep learning model;

[0118] The filter prediction module 74 is configured to input the admittance responses of the series and parallel resonators to the trained deep learning model, so that the two admittance curves representing the admittance responses of the series and parallel resonators satisfy a certain relationship, that is, one of the admittance curves is the resonant frequency of the parallel resonator and the other admittance curve is the anti-resonant frequency of the series resonator, and the key structural parameters of the series-parallel resonator are obtained based on the deep learning model;

[0119] Circuit topology module 75: configured to extract COM parameters and calculate ladder filter design performance based on the COM model and predicted key structural parameters.

[0120] The functions of the above modules correspond to the methods and will not be repeated here.

[0121] Based on the proposed fast and intelligent design method of SAW filters, we developed a fast and intelligent design platform for SAW filters, such as Figure 7As shown, based on this platform, we can freely drag the admittance curves of the series resonator and the parallel resonator with the mouse to make it satisfy the resonant frequency of the admittance curve of one parallel resonator equal to the anti-resonant frequency of the admittance curve of the other series resonator. Based on the deep learning algorithm model, the key structural parameters of the series-parallel resonator are displayed on this platform in real time, and then the COM parameters are extracted through the back-end COM model to achieve the rapid design of different passband filters. We can also directly make the resonant frequency of the admittance curve of the parallel resonator equal to the anti-resonant frequency of the admittance curve of the other series resonator by inputting the target resonant frequency, thereby quickly realizing the rapid intelligent design of the surface acoustic wave filter.

[0122] The present invention can quickly and accurately predict the key structural parameters of the surface acoustic wave filter with different filter passband requirements based on the input series-parallel admittance curve, and realize the rapid intelligent design of the surface acoustic wave filter. This method overcomes the complex and time-consuming simulation solution process of the traditional surface acoustic wave filter, and the limitation that professionals are required to perform complex circuit topology, greatly shortens the design cycle of the surface acoustic wave filter, and significantly reduces the time and cost in the design process.

[0123] The innovation of the embodiment of the present invention is to develop a fast and intelligent design method and system for surface acoustic wave filters, which optimizes and accelerates the design process of surface acoustic wave filters based on deep learning methods. By generating sample data through electromagnetic simulation software, building a deep learning neural network architecture, and conducting large-scale data training, the deep learning algorithm can effectively learn the complex mapping relationship between the structural parameters of the surface acoustic wave resonator and the admittance response spectrum. Using the predicted key structural parameters of the surface acoustic wave resonator, based on the coupled mode (COM) model, the key COM parameters are extracted, and the fitted surface acoustic wave resonator admittance is converted into an S matrix to obtain the filter passband, thereby realizing the bandpass filtering function of the surface acoustic wave filter. This innovative method can quickly determine the key structural parameters of the surface acoustic wave filter, significantly shorten the design time, simplify the design process, and save costs. In addition, this technology can be effectively extended to other fields such as radio frequency devices and optoelectronic devices to provide interdisciplinary design solutions.

[0124] Reference below Fig. 9 , which shows an electronic device (eg Figure 1 A schematic diagram of the structure of a computer device 600 (a server or terminal device as shown). Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0125] like Fig. 9As shown, the computer device 600 includes a central processing unit (CPU) 601 and a graphics processing unit (GPU) 602, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 603 or a program loaded from a storage part 609 to a random access memory (RAM) 604. In the RAM 604, various programs and data required for the operation of the device 600 are also stored. The CPU 601, GPU 602, ROM 603 and RAM 604 are connected to each other through a bus 605. An input / output (I / O) interface 606 is also connected to the bus 605.

[0126] The following components are connected to the I / O interface 606: an input section 607 including a keyboard, a mouse, etc.; an output section 608 including a display such as a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 609 including a hard disk, etc.; and a communication section 610 including a network interface card such as a LAN card, a modem, etc. The communication section 610 performs communication processing via a network such as the Internet. A drive 611 may also be connected to the I / O interface 606 as needed. A removable medium 612, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 611 as needed so that a computer program read therefrom is installed into the storage section 609 as needed.

[0127] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 610, and / or installed from the removable medium 612. When the computer program is executed by the central processing unit (CPU) 601 and the graphics processing unit (GPU) 602, the above-mentioned functions defined in the method of the present invention are executed.

[0128] It should be noted that the computer-readable medium described in the present invention may be a computer-readable signal medium or a computer-readable medium or any combination of the above two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor device, apparatus or device, or any combination of the above. More specific examples of computer-readable media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution device, apparatus or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium may also be any computer-readable medium other than a computer-readable medium that can send, propagate or transmit a program for use by or in conjunction with an instruction execution device, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0129] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0130] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the device, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based device that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0131] The modules involved in the embodiments of the present invention may be implemented in software or hardware, and the modules described may also be arranged in a processor.

[0132] As another aspect, the present invention further provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by the electronic device, the electronic device: executes the method steps described in the first aspect.

[0133] The above description is only a preferred embodiment of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present invention is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present invention (but not limited to) to form a technical solution.

Claims

1. A fast and intelligent design method for a surface acoustic wave filter, characterized in that: The method comprises the following steps: S1. Based on electromagnetic simulation software, a physical model of the surface acoustic wave resonator is established, the corresponding admittance response is calculated, the required sample data set is randomly generated, and a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing the target surface acoustic wave filter are obtained; S2. Preprocessing the collected sample data set, including data cleaning and normalization, to form an input vector of a deep learning model; S3, dividing the sample data set into a training set and a validation set, applying the deep learning model on the training set, building a reverse design neural network model, learning the mapping relationship between structural parameters and admittance response, and evaluating the algorithm performance on the validation set, and predicting the structural parameters of series and parallel surface acoustic wave resonators based on the trained deep learning model; S4. Combined with the COM model, the COM parameters are extracted according to the admittance response of the series and parallel surface acoustic wave resonators. The SAW single-ended resonators are combined in a trapezoidal structure, and the admittance of the series and parallel resonators is converted into an S matrix to obtain the filter passband, thereby realizing the intelligent and rapid design of the surface acoustic wave filter.

2. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: The structural parameters of the surface acoustic wave resonator physical model established in S1 include the number of layers, material, geometry, metallization rate, width, thickness of the periodic layered electrode structure and the material and width of the substrate structure.

3. The fast and intelligent design method for surface acoustic wave filters according to claim 2, characterized in that: The number of layers of the periodic layered electrode structure is arbitrary, the material of the periodic layered electrode structure is any one of titanium, aluminum, copper, gold, and silver metals or a composite thereof, and the geometric shape of the periodic layered electrode structure is rectangular or trapezoidal.

4. The fast and intelligent design method for surface acoustic wave filters according to claim 3, characterized in that: The layered electrode structure is a three-layer metal layered nanostructure with a trapezoidal geometric shape. The three-layer electrode structure material is titanium aluminum titanium, and the substrate material is a 42°YX tangential lithium tantalate piezoelectric crystal 42°YX-LT. The width of the substrate is λ, and the metallization rates of the two electrodes are r1 and r2, wherein the metallization rate r=w / λ, w is the electrode width, and the thicknesses of the three-layer electrodes are t1, t2, and t3 from top to bottom, respectively.

5. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: The deep learning model described in S3 includes an input layer, a hidden layer, and an output layer in sequence to establish a complex mapping relationship between structural parameters and admittance response spectra; The input layer receives the admittance response spectrum of the surface acoustic wave resonator as an input feature, and the output layer predicts the structural parameters and electrode materials of the corresponding resonator; the length of the input layer is 800, corresponding to the number of sampling points on the admittance response spectrum; The hidden layer includes 8 fully connected layers, each of which contains multiple neurons, and each neuron performs nonlinear mapping by weighting and processing the input of the previous layer through an activation function; The output layer is composed of several fully connected layers, and the output length is 9, that is, the output substrate width λ, the electrode widths w1 and w2, the electrode thicknesses t1, t2 and t3, and the material properties of the three-layer electrodes.

6. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: In S3, the training set is input into the deep learning model to establish a mapping relationship between the structural parameters and the admittance response. The training method of the deep learning model specifically includes: Initialize the parameters, set the batch size to 128, the total number of epochs to 500, the initial learning rate to 0.0001, the learning rate decay strategy to reduce the learning rate by 60% every 30 epochs of training, use Adam as the gradient descent optimizer, use mean square error MSE as the loss function, use ReLU function as the activation function of the entire network model, and use the chain rule to reversely transfer the gradient to each neuron to optimize the parameters in the neuron.

7. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: The reverse-designed neural networks include fully connected neural networks, convolutional neural networks, Transformer, residual neural networks, recurrent neural networks, generative adversarial networks, and autoencoder neural networks.

8. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: Also includes: Electromagnetic simulation software is used to calculate the admittance response of the full-period model of the surface acoustic wave resonator combined with the layered electrode and substrate structure. The predicted structural parameters are used to fit the full-period surface acoustic wave resonator admittance using the COM model. The period is set to 89. Based on the predicted admittance response spectrum and structural parameters, the wave velocity V of the surface grazing body wave SSBW is extracted. B , resonant frequency f r and the anti-resonance frequency f ar , the COM parameter calculation formula is: Δ=ω / V B -2π / P I D V =-η 2 / 2 κ=|ε|(η+|ε| / 2) k B =-|e| 2 n / (n+2|e|) Where Δ is the normalized frequency, ω is the angular frequency, and P I represents the propagation period characteristics of surface acoustic waves in specific structures, such as periodic layered electrodes and substrate structures, ε represents the parameter of SAW reflection, η is the parameter representing the SAW penetration depth, κ and κ B are different parameters of the coupling strength between different modes of surface acoustic waves during propagation.

9. The fast and intelligent design method for surface acoustic wave filters according to claim 1, characterized in that: In S4, based on the obtained admittance response, the SAW single-ended resonators are combined in a trapezoidal structure, and the admittance of the series and parallel resonators is converted into an S matrix. The specific calculation process is: The filter has only one series resonator and one parallel resonator, and the [abcd] matrix of the filter is: Normalize the port impedance Z0, Z0 is 50Ω: Then transform the [ABCD] matrix into the [S] matrix: Among them, y s and p are parameters related to matrix transformation. The [abcd] matrix represents the matrix that uses the total voltage and total current of the output port to represent the total voltage and total current of the input port. The [ABCD] matrix represents the [abcd] matrix after impedance normalization. S11 and S22 represent reflection coefficients, which reflect the degree of signal reflection at the port. S21 and S12 represent transmission coefficients, which reflect the degree of signal transmission between ports.

10. A fast intelligent design system for surface acoustic wave filters, characterized in that: include: Data simulation module: configured to establish a physical model of the surface acoustic wave resonator based on electromagnetic simulation software, calculate the corresponding admittance response, randomly generate the required sample data set, and obtain a series of surface acoustic wave resonator admittance responses and key structural parameters for constructing the target surface acoustic wave filter; Sample processing module: configured to pre-process the collected sample data set, including data cleaning and normalization, to form an input vector of the deep learning model; Algorithm training module: configured to build a deep learning model, learn the mapping relationship between structural parameters and admittance curves, and evaluate the performance of the deep learning model; Filter prediction module: configured to input the admittance responses of the series and parallel resonators to the trained deep learning model, so that the two admittance curves representing the admittance responses of the series and parallel resonators satisfy a certain relationship, that is, one of the admittance curves is the resonant frequency of the parallel resonator and the other admittance curve is the anti-resonant frequency of the series resonator, and the key structural parameters of the series-parallel resonator are obtained based on the deep learning model; Circuit Topology Module: Configured to extract COM parameters and calculate ladder filter design performance based on COM models and predicted key structural parameters.

11. An electronic device, comprising: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images