Method and system for fitting crystal oscillator circuit parameters based on neural network

Abstract

The invention relates to a method for fitting crystal oscillator circuit parameters based on a neural network, and the method comprises the steps: obtaining a frequency impedance curve of a target crystal oscillator circuit through testing, and constructing a training data set; based on a circuit structure formed by a plurality of elements in a target crystal oscillator circuit, constructing a neural network model for fitting element parameters in the target crystal oscillator circuit; taking the test frequency in the training data set as input, taking a circuit impedance value as a real label, and inputting the training data set into the neural network model in batches to output a predicted impedance value; constructing a loss function to train the neural network model; and after training of the neural network model is completed, obtaining fitted element parameters in the target crystal oscillator circuit. According to the method, a crystal oscillator circuit can be fitted by updating neuron parameters by applying a neural network algorithm, and element parameters in the crystal oscillator are obtained within a small error range.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated circuit design, and in particular to a method and system for fitting crystal oscillator circuit parameters based on a neural network. Background Art

[0002] A crystal oscillator is an electronic oscillator that uses the piezoelectric effect of a quartz crystal to generate a stable frequency. Its high frequency stability, low phase noise, and miniaturization make it an indispensable component in modern electronic devices. As the frequency of integrated circuit devices increases, the impact of parasitic noise becomes more and more serious. After exceeding 200MHz, ordinary instruments can no longer accurately measure the size of the key parameters inside the crystal oscillator. The key parameters inside the crystal oscillator play an important guiding role in the design of the oscillator circuit. At present, there is no effective method to fit the crystal oscillator circuit. Summary of the invention

[0003] The purpose of the present invention is to improve the problem in the prior art that the size of key parameters inside a crystal oscillator cannot be accurately measured, and to provide a method and system for fitting crystal oscillator circuit parameters based on a neural network.

[0004] In order to achieve the above-mentioned object of the invention, the embodiment of the present invention provides the following technical solutions:

[0005] A method for fitting crystal oscillator circuit parameters based on a neural network, specifically comprising:

[0006] S1, obtaining the frequency impedance curve of the target crystal oscillator circuit through testing;

[0007] S2, performing preprocessing based on the frequency impedance curve to obtain a training data set including multiple groups of test frequencies and circuit impedance values;

[0008] S3, the target crystal oscillator circuit includes a plurality of components, and a neural network model is constructed according to a circuit structure composed of the plurality of components to fit the component parameters in the target crystal oscillator circuit;

[0009] S4, taking the test frequency in the training data set as input and the circuit impedance value as the true label, inputting the training data set into the neural network model in batches to output the predicted impedance value;

[0010] S5, constructing a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit;

[0011] S6, after the neural network model training is completed, obtaining the component parameters in the fitted target crystal oscillator circuit.

[0012] Further, the S2 includes:

[0013] The frequency impedance curve is subjected to Gaussian smoothing filtering to filter out high frequency noise, thereby obtaining a smooth frequency impedance curve;

[0014] The test frequency and circuit impedance value are extracted based on the smoothed frequency-impedance curve to construct a training data set.

[0015] Further, the target crystal oscillator circuit includes at least one crystal oscillator and one or more elements selected from the group consisting of a resistor, a capacitor, and an inductor; and S3 includes:

[0016] Using component parameters of a plurality of components included in the target crystal oscillator circuit as trainable parameters of a plurality of neurons in the neural network model;

[0017] The trainable parameters of the neuron, i.e., the component parameters, are introduced as variables into the activation function to obtain a functional representation of the component impedance value output by each neuron;

[0018] The circuit structure formed by the multiple elements is used as the connection relationship between neurons to determine the forward calculation process of the neural network model.

[0019] Furthermore, the expression of the activation function in the neural network model is:

[0020]

[0021] Among them, v i represents the component parameter of the ith component, that is, the trainable parameter of the ith neuron, whose distribution satisfies the Gaussian function with a mean of 0 and a standard deviation of 1. max_v i and min_v i The value range of the component parameter.

[0022] Further, the S4 includes:

[0023] Using the test frequency in the training data set as input and the circuit impedance value as the true label, the training data set is input into the neural network model in batches;

[0024] Determine the calculation relationship of the series and / or parallel connection of the circuit composed of each component in the target crystal oscillator circuit as the forward calculation process of the neural network model;

[0025] The predicted impedance value output by the neural network model is determined based on the component impedance value function representation output by each neuron.

[0026] Further, the S5 includes:

[0027] Constructing a loss function based on the predicted impedance value output by the neural network model and the circuit impedance value as the true label;

[0028] The formula for constructing the loss function is:

[0029] σ=mean{(re label (s)-re pred (s)) 2 +(im label (s)-im pred (s)) 2},

[0030] Among them, label (s) and im label (s) represents the real and imaginary parts of the circuit impedance value of the target crystal oscillator circuit corresponding to the frequency s, re pred (s) and im pred (s) represents the real and imaginary parts of the predicted impedance value corresponding to the frequency s output by the neural network model;

[0031] According to the prediction error of the neural network model obtained by the loss function, when the prediction error is greater than a threshold, the optimization algorithm is used to update and optimize the trainable parameters of each neuron in the neural network model. The optimization process of the trainable parameters of the neurons is the fitting process of the component parameters in the target crystal oscillator circuit.

[0032] On the other hand, an embodiment of the present invention further provides a system for fitting crystal oscillator circuit parameters based on a neural network, specifically comprising:

[0033] A data processing module, used for obtaining a frequency impedance curve of a crystal oscillator circuit through testing, and performing preprocessing based on the frequency impedance curve to obtain a training data set including multiple groups of test frequencies and circuit impedance values;

[0034] A model building module, for constructing a neural network model to fit the component parameters in the target crystal oscillator circuit according to the circuit structure composed of the multiple components in the target crystal oscillator circuit;

[0035] A model optimization module is used to take the test frequency in the training data set as input and the circuit impedance value as the true label, input the training data set in batches into the neural network model to output the predicted impedance value; construct a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit;

[0036] The parameter acquisition module is used to obtain the component parameters in the fitted target crystal oscillator circuit after the neural network model training is completed.

[0037] On the other hand, an embodiment of the present invention also provides a computer-readable storage medium including computer-readable instructions, which, when executed, enable a processor to perform the operations in the method described in the embodiment of the present invention.

[0038] On the other hand, an embodiment of the present invention also provides an electronic device, including: a memory, storing program instructions; a processor, connected to the memory, executing the program instructions in the memory, and implementing the steps in the method described in the embodiment of the present invention.

[0039] Compared with the prior art, the beneficial effects achieved by the method and system of the present invention include: utilizing the excellent computing function of the neural network algorithm, transforming the fitting of the crystal oscillator circuit into the training of the neural network, applying the neural network algorithm to fit the crystal oscillator circuit, and fitting the component parameters in the circuit by training the weights of the neurons in the neural network model, thereby obtaining the key component parameters inside the crystal oscillator within a smaller error range. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 is an exemplary flow chart of the method for fitting crystal oscillator circuit parameters based on a neural network described in an embodiment;

[0042] Figure 2 is a schematic diagram of the input and output process of the neuron corresponding to the element i described in the embodiment;

[0043] Figure 3 is a schematic circuit diagram of an exemplary target crystal oscillator circuit described in the embodiment;

[0044] Figure 4 It is an exemplary structural diagram of the system for fitting crystal oscillator circuit parameters based on neural network described in the embodiment. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0046] Figure 1 It is an exemplary flow chart of the method for fitting crystal oscillator circuit parameters based on neural network described in an embodiment.

[0047] like Figure 1 As shown, the method for fitting crystal oscillator circuit parameters based on a neural network specifically includes:

[0048] S1, obtain the frequency impedance curve of the target crystal oscillator circuit through testing.

[0049] In some embodiments, the target crystal oscillator circuit may refer to a crystal oscillator circuit to be fitted, and the frequency impedance curve of the target crystal oscillator circuit may be obtained by instrument testing to obtain a large amount of test data, including the input frequency and the circuit impedance value obtained by the test. For example, the impedance data of a crystal oscillator circuit in the range of 100MHz to 1GHz is obtained by instrument testing, with a total of 90,000 data points. Because the center frequency of the device is 500MHz, the data points in the range of 100MHz to 700MHz are intercepted and made into a standard data set of the pytorch framework, with 3750 data points in each batch, for a total of 16 batches.

[0050] S2, preprocessing is performed based on the frequency impedance curve to obtain test data, and a fitting training data set is constructed.

[0051] Furthermore, S2 specifically includes: performing Gaussian smoothing filtering on the frequency impedance curve to filter out high-frequency noise and obtain a smooth frequency impedance curve; extracting the test frequency and circuit impedance value from the smooth frequency impedance curve to construct a training data set. The circuit impedance value is represented by Z label (s) = re label (s)+im label (s), re label and label Represents the real and imaginary parts of the circuit impedance value of the target crystal oscillator circuit corresponding to the complex frequency s.

[0052] In some embodiments, the target crystal oscillator circuit impedance value obtained through instrument testing will have certain noise, causing the originally smooth curve to jitter, affecting the convergence of subsequent algorithms, so the test data needs to be processed first. The processing method of the present invention is to perform Gaussian smoothing filtering to filter out high-frequency noise, so that the frequency impedance curve can become smooth.

[0053] S3, the target crystal oscillator circuit includes multiple components, and according to the circuit structure composed of the multiple components, a neural network model is constructed to fit the component parameters in the target crystal oscillator circuit.

[0054] In some embodiments, the target crystal oscillator circuit may be composed of different components. Specifically, the target crystal oscillator circuit includes multiple components, specifically including at least one crystal oscillator, and one or more components of a resistor component, a capacitor component, and an inductor component.

[0055] The S3 may specifically include the following steps:

[0056] S31, using the component parameters of the multiple components included in the target crystal oscillator circuit as the trainable parameters of the multiple neurons in the neural network model. For example, the component parameters may include a resistance parameter R, a capacitance parameter C, and an inductance parameter L.

[0057] S32, introducing the trainable parameters of the neuron, namely the component parameters, as variables into the activation function to obtain a functional representation of the component impedance value output by each neuron.

[0058] S33, using the circuit structure formed by the multiple elements as the connection relationship between neurons to determine the forward calculation process of the neural network model.

[0059] In some embodiments, the element impedance value function representation of a neuron output in S32 can be as follows: Figure 2 As shown, it represents the input and output process of a neuron corresponding to an element i. The calculation formula is as follows:

[0060] z i =sf(v i ),

[0061] Where s represents the complex frequency of the input, that is, the complex domain representation of frequency f is s=j·2πf; f(v i ) represents the element parameter v corresponding to the i-th element i The activation function of z i Represents the impedance corresponding to the i-th element.

[0062] Furthermore, the expression of the activation function in the neural network model is:

[0063]

[0064] Among them, v i represents the component parameter of the ith component, that is, the trainable parameter of the ith neuron, whose distribution satisfies the Gaussian function with a mean of 0 and a standard deviation of 1. max_v i and min_v i is the value range of the component parameter of the i-th component.

[0065] S4, taking the test frequency in the training data set as input and the circuit impedance value as the true label, inputting the training data set into the neural network model in batches to output the predicted impedance value.

[0066] The S4 specifically includes the following steps:

[0067] S41, using the test frequency in the training data set as input and the circuit impedance value as the true label, and inputting the training data set into the neural network model in batches;

[0068] S42, determining the calculation relationship of the series and / or parallel connection of the circuit composed of each component in the target crystal oscillator circuit as the forward calculation process of the neural network model;

[0069] S43, determining the predicted impedance value output by the neural network model according to the element impedance value function representation output by each neuron, wherein the predicted impedance value is represented by Z pred (s) = re pred +im pred ,re pred and pred The real and imaginary parts of the predicted impedance value output by the neural network model corresponding to the complex frequency s.

[0070] In some embodiments, Figure 3 is a schematic circuit diagram of an exemplary target crystal oscillator circuit described in the embodiment, Figure 3 The process of the neural network model performing forward calculation and outputting predicted resistance value is described as an example.

[0071] like Figure 3 The target crystal oscillator circuit shown is a crystal oscillator circuit composed of multiple resistors, inductors, capacitors and a crystal. In the embodiment of the present invention, the impedance calculation process of the circuit is used as the forward calculation process of the neural network model.

[0072] Specifically, according to Figure 3The target crystal oscillator circuit shown may include four primary circuits A, B, C, and D. Each primary circuit is composed of a number of components. The impedance value corresponding to each primary circuit may be calculated as follows:

[0073]

[0074] B(s)=v5+sv6,

[0075]

[0076] Among them, A(s), B(s), C(s), and D(s) represent the impedance values ​​of the four primary circuits A, B, C, and D respectively, among which circuit A represents the circuit representation of the crystal oscillator crystal; v1 represents the resistance parameter of the first component resistor, v2 represents the capacitance parameter of the second component capacitor, v3 represents the inductance parameter of the third component inductor, v4 represents the crystal parameter of the fourth component quartz crystal, v5 represents the resistance parameter of the fifth component resistor, v6 represents the inductance parameter of the sixth component inductor, v7 represents the resistance parameter of the seventh component resistor, v8 represents the inductance parameter of the eighth component inductor, v9 represents the capacitance parameter of the ninth component capacitor, and v 10 The resistance parameter of the tenth element resistor, v 11 The capacitance parameter of the eleventh element capacitor, v 12 represents the inductance parameter of the twelfth component inductor; s represents the complex frequency of the input, that is, the complex domain representation of the frequency f is s=j·2πf.

[0077] Furthermore, the calculation process of the overall impedance value of the target crystal oscillator circuit can be obtained according to the impedance values ​​of the four circuits A, B, C, and D. The calculation formula is as follows:

[0078] Among them, Z pred (s) represents the predicted impedance value of the target crystal oscillator circuit as a whole at the complex frequency s.

[0079] The impedance calculation process of the above circuit is used as the forward calculation process of the neural network model to obtain the predicted impedance value. pred (s) is also used as the final predicted impedance value Z of the neural network model pred (s), including the real part re pred and the imaginary part im pred .

[0080] S5, constructing a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit.

[0081] The S5 specifically includes the following steps:

[0082] S51, constructing a loss function based on the predicted impedance value output by the neural network model and the circuit impedance value as the true label;

[0083] The formula for constructing the loss function is:

[0084] σ=mean{(re label (s)-re pred (s)) 2 +(im label (s)-im pred (s)) 2},

[0085] Wherein, σ represents the prediction error of the neural network model, mean{} represents the average; label (s) and im label (s) represents the real and imaginary parts of the circuit impedance value of the target crystal oscillator circuit corresponding to the frequency s, re pred (s) and im pred (s) represents the real and imaginary parts of the predicted impedance value corresponding to the frequency s output by the neural network model.

[0086] S52, based on the prediction error of the neural network model obtained by the loss function, when the prediction error is greater than a threshold, the optimization algorithm is used to update and optimize the trainable parameters of each neuron in the neural network model until the prediction error is less than the threshold.

[0087] Specifically, the optimization algorithm can be set to the adam algorithm, the learning rate is set to 0.01, and the learning rate scheduler is set to multiply the learning rate by 0.3 after training for 50, 100, 150, 200, and 250 generations, respectively, where each generation refers to training all the data in the data set once. The neural network model is updated once each batch of data is calculated, and a total of 300 generations are trained. The final training results converge, and the neuron parameters of the credible neural network model can be obtained. The neuron parameter set corresponds to the component parameters inside the target crystal oscillator circuit.

[0088] S6, after the neural network model training is completed, obtaining the component parameters in the fitted target crystal oscillator circuit.

[0089] Figure 4 : is an exemplary structural diagram of a system for fitting crystal oscillator circuit parameters based on a neural network in an embodiment, and the system is used to implement the above-mentioned method for fitting a crystal oscillator circuit based on a neural network algorithm, specifically comprising:

[0090] A data processing module 410 is used to obtain a frequency impedance curve of a crystal oscillator circuit through testing and to perform preprocessing based on the frequency impedance curve to obtain a training data set including multiple groups of test frequencies and circuit impedance values;

[0091] A model building module 420 is used for constructing a neural network model to fit the component parameters in the target crystal oscillator circuit according to the circuit structure composed of the multiple components in the target crystal oscillator circuit;

[0092] The model optimization module 430 is used to take the test frequency in the training data set as input and the circuit impedance value as the true label, input the training data set in batches into the neural network model to output the predicted impedance value; construct a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit;

[0093] The parameter acquisition module 440 is used to obtain the component parameters in the fitted target crystal oscillator circuit after the neural network model training is completed.

[0094] An embodiment of the present invention further provides a computer-readable instruction, wherein when the instruction is executed in an electronic device, the program causes the electronic device to execute the operation steps included in the method of the present invention.

[0095] An embodiment of the present invention further provides a storage medium storing computer-readable instructions, wherein the computer-readable instructions enable an electronic device to execute the operation steps included in the method of the present invention.

[0096] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0097] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0098] Those of ordinary skill in the art will appreciate that the modules of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the constituent modules and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0100] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for fitting crystal oscillator circuit parameters based on neural network, characterized in that: include: S1, obtaining the frequency impedance curve of the target crystal oscillator circuit through testing; S2, performing preprocessing based on the frequency impedance curve to obtain a training data set including multiple groups of test frequencies and circuit impedance values; S3, the target crystal oscillator circuit includes a plurality of components, and a neural network model is constructed according to a circuit structure composed of the plurality of components to fit the component parameters in the target crystal oscillator circuit; S4, taking the test frequency in the training data set as input and the circuit impedance value as the true label, inputting the training data set into the neural network model in batches to output the predicted impedance value; S5, constructing a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit; S6, after the neural network model training is completed, obtaining the component parameters in the fitted target crystal oscillator circuit.

2. The method for fitting a crystal oscillator circuit based on a neural network algorithm according to claim 1, characterized in that: The S2 includes: The frequency impedance curve is subjected to Gaussian smoothing filtering to filter out high frequency noise, thereby obtaining a smooth frequency impedance curve; The test frequency and circuit impedance value are extracted based on the smoothed frequency-impedance curve to construct a training data set.

3. The method for fitting a crystal oscillator circuit based on a neural network algorithm according to claim 1, characterized in that: The target crystal oscillator circuit includes at least one crystal oscillator and one or more elements selected from a resistor element, a capacitor element, and an inductor element; S3 includes: Using component parameters of a plurality of components included in the target crystal oscillator circuit as trainable parameters of a plurality of neurons in the neural network model; The trainable parameters of the neuron, i.e., the component parameters, are introduced as variables into the activation function to obtain a functional representation of the component impedance value output by each neuron; The circuit structure formed by the multiple elements is used as the connection relationship between neurons to determine the forward calculation process of the neural network model.

4. The method for fitting a crystal oscillator circuit based on a neural network algorithm according to claim 3, characterized in that: The expression of the activation function in the neural network model is: Among them, v i represents the component parameter of the ith component, that is, the trainable parameter of the ith neuron, whose distribution satisfies the Gaussian function with a mean of 0 and a standard deviation of 1. max_v i and min_v i The value range of the component parameter.

5. The method for fitting a crystal oscillator circuit based on a neural network algorithm according to claim 4, characterized in that: The S4 includes: Using the test frequency in the training data set as input and the circuit impedance value as the true label, the training data set is input into the neural network model in batches; Determine the calculation relationship of the series and / or parallel connection of the circuit composed of each component in the target crystal oscillator circuit as the forward calculation process of the neural network model; The predicted impedance value output by the neural network model is determined based on the component impedance value function representation output by each neuron.

6. The method for fitting a crystal oscillator circuit based on a neural network algorithm according to claim 1, characterized in that: The S5 includes: Constructing a loss function based on the predicted impedance value output by the neural network model and the circuit impedance value as the true label; The formula for constructing the loss function is: σ=mean{(re label (s)-re pred (s)) 2 +(im label (s)-im pred (s)) 2 }, Wherein, σ represents the prediction error of the neural network model, mean{} represents the average; label (s) and im label (s) represents the real and imaginary parts of the circuit impedance value of the target crystal oscillator circuit corresponding to the frequency s, re pred (s) and im pred (s) represents the real and imaginary parts of the predicted impedance value corresponding to the frequency s output by the neural network model; According to the prediction error of the neural network model obtained by the loss function, when the prediction error is greater than a threshold, the optimization algorithm is used to update and optimize the trainable parameters of each neuron in the neural network model. The optimization process of the trainable parameters of the neurons is the fitting process of the component parameters in the target crystal oscillator circuit.

7. A system for fitting crystal oscillator circuit parameters based on neural network, characterized in that: A method for fitting a crystal oscillator circuit based on a neural network algorithm according to any one of claims 1 to 6, comprising: A data processing module, used for obtaining a frequency impedance curve of a crystal oscillator circuit through testing, and performing preprocessing based on the frequency impedance curve to obtain a training data set including multiple groups of test frequencies and circuit impedance values; A model building module, for constructing a neural network model to fit the component parameters in the target crystal oscillator circuit according to the circuit structure composed of the multiple components in the target crystal oscillator circuit; A model optimization module is used to take the test frequency in the training data set as input and the circuit impedance value as the true label, input the training data set in batches into the neural network model to output the predicted impedance value; construct a loss function to train the neural network model; the training process of the neural network model is the fitting process of the component parameters in the target crystal oscillator circuit; The parameter acquisition module is used to obtain the component parameters in the fitted target crystal oscillator circuit after the neural network model training is completed.

8. A computer-readable storage medium comprising computer-readable instructions, characterized in that: When the computer readable instructions are executed, the processor performs the operations in any one of the methods of claims 1-6.

9. An electronic device, characterized in that: The device comprises: Memory, which stores program instructions; A processor is connected to the memory and executes program instructions in the memory to implement the steps in any method of claims 1-6.

Images