WO2024087848A1

WO2024087848A1 - Simulation test fitting process parameter method, system, device, and storage medium

Info

Publication number: WO2024087848A1
Application number: PCT/CN2023/115029
Authority: WO
Inventors: 杨睿智; 袁军平; 胡锦钊; 常林森; 李帅; 张磊; 郭嘉帅
Original assignee: 深圳飞骧科技股份有限公司
Priority date: 2022-10-25
Filing date: 2023-08-25
Publication date: 2024-05-02
Also published as: CN115495995A; CN115495995B

Abstract

Embodiments of the present invention provide a simulation test fitting process parameter method, a simulation test fitting system, a simulation test fitting device, and a computer readable storage medium. The simulation test fitting process parameter method comprises: cleaning test original data to obtain samples; performing fitting on the samples to obtain a single admittance sample; performing fitting on sampled samples to obtain a plurality of sampled admittance samples; and carrying out distributed computing and fitting on all the samples to obtain all admittance samples; merging a plurality of groups of parameter results corresponding to resonators having a same geometric parameter into one group for processing; training a machine learning model and obtaining an adjustment parameter prediction model; and verifying the sampled samples by means of the adjustment parameter prediction model. Compared with the prior art, the technical solution of the present invention can realize automatic simulation testing in statistical significance under different processes and test fluctuation, and has good universality, a high computation speed, and high accuracy.

Description

Simulation test fitting process parameter method, system, device and storage medium

Technical Field

The present invention relates to the technical field of acoustic simulation test fitting design, and in particular to a simulation test fitting process parameter method, a simulation test fitting system, a simulation test fitting device and a computer-readable storage medium applied to a surface acoustic wave filter.

Background technique

In recent years, with the increasing application of SAW filters, various SAW filters are applied to different scenarios. In response to different needs, the design of SAW filters is becoming more and more important. Whether the design of SAW filters is reliable depends largely on the accuracy of the physical simulation model of SAW filters.

At present, in the prior art, the physical simulation of SAW filters is divided into equivalent circuit model (MBVD), coupled mode model (COM) and finite element method (FEM). Among them, the equivalent circuit model (MBVD) greatly simplifies the actual physics, resulting in low accuracy; and the finite element method (FEM) cannot be used for the rapid iteration design of SAW devices due to its slow calculation; in terms of balancing calculation accuracy and speed, the coupled mode model (COM) is more balanced than the other two methods, so it is also more widely used in design iteration. Specifically, the coupled mode model (COM) is a phenomenological model. The coupled mode model (COM) uses a linear model plus some adjustment parameters to approximate the real nonlinear physics. Therefore, these adjustment parameters are the key to determining whether the calculated results are similar to the measured results. There are many factors that affect the adjustment parameters, which have different degrees of correlation with the frequency, the geometric parameters of the device, the material parameters of the piezoelectric substrate, and the processes of different foundries.

However, the coupled mode model (COM) of the SAW filter in the related art has the following problems in the application of adjustment parameters in the simulation and test during the design iteration: First, the existing model cannot be universally applicable. The coupled mode model in academia is public, but the adjustment parameters are not well understood. The parameters are not universal. Academic papers often only need to prove their effectiveness on one or two examples, but for the actual industrial design simulation needs, it is necessary to ensure accuracy on a universal basis, and the difficulty is incomparable. It is no exaggeration to say that the accuracy of simulation testing directly determines the design cycle and cost. Second, in the actual mass production of SAW filters, with the combination of process fluctuations and test fluctuations, universal simulation testing is relatively difficult. Because the device production and manufacturing of SAW filters will face inevitable process fluctuations, and the same design will get different actual results in manufacturing; and the resonator admittance of the SAW filter needs to be tested at the wafer level, and the objective test fluctuations cannot be avoided. Third, many simulation tests of SAW filters in related technologies rely on manual attempts to find adjustment parameters. The efficiency of manual parameter search is very low. The process of manually adjusting the adjustment parameters repeatedly generally depends on the number of samples that need to be simulated and tested, and the time spent by humans is generally in weeks to months. It cannot be automatically expanded to large quantities of data, it cannot realize automatic processing of large quantities of simulation test data, and it is impossible to quickly obtain an automated method that matches different foundry processes in a short time. Fourth, manual methods can only find adjustment parameters that are applicable within a certain range, so they must be discrete, which will lead to low accuracy. If manual adjustment methods are used, neither universality can be guaranteed (where manual observation of laws is limited, and the laws are generally discrete piecewise functions) nor accuracy over a large range can be guaranteed (where manual observation of laws is limited). How to obtain a continuous model through limited simulation test data and ensure the reliability of calculations when there is no simulation test data is a technical problem that needs to be solved.

Therefore, it is necessary to provide a new method, system and device to solve the above technical problems.

Summary of the invention

The purpose of the present invention is to overcome the above-mentioned technical problems and to provide a simulation test fitting process parameter method, a simulation test fitting system, a simulation test fitting device and a computer-readable storage medium which can solve the problem of realizing automated simulation testing in a statistical sense under different processes and test fluctuations and have good universality, fast calculation speed and high accuracy.

In a first aspect, an embodiment of the present invention provides a simulation test fitting process parameter method, which is applied to a surface acoustic wave filter, wherein the surface acoustic wave filter includes a plurality of resonators; the method includes the following steps:

Step S1, obtaining test raw data of a plurality of resonators, cleaning the test raw data according to a preset cleaning rule and obtaining cleaned samples; the test raw data includes geometric parameters and test files for matching the resonators, and the cleaning rule is to remove inconsistent parts according to the statistical consistency of the test raw data;

Step S2, setting constraints for the preset global optimization method, and then fitting the sample into a single admittance sample by the global optimization method after setting the constraints; the constraints are the type of the selected global optimization method and the relevant parameters in the global optimization method, and the admittance sample, the type of the global optimization method, and the relevant parameters in the global optimization method correspond to each other one by one;

Step S3, sampling all the samples, and then fitting the sampled samples according to the step S2 to obtain multiple sampled admittance samples, and then judging whether the distribution effect of the type of the global class optimization method and the relevant parameters in the global class optimization method in the multiple sampled admittance samples is less than a preset indicator threshold:

If yes, proceed to step S4; if no, adjust the constraint and return to step S2;

Step S4, performing distributed calculation on all the samples and fitting to obtain all the admittance samples;

Step S5, obtaining multiple groups of parameter results according to the data in all the admittance samples, combining the multiple groups of parameter results corresponding to the resonators with the same geometric parameters into one group for processing, and then outputting the processed multiple groups of parameter results;

Step S6, preselecting a machine learning model, using the geometric parameters of the resonator in step S1 as input data of the machine learning model, using the multiple groups of parameter results output in step S5 as output data of the machine learning model, training the machine learning model and obtaining a trained prediction adjustment parameter model;

Step S7: Sampling all the samples in step S2, and then The samples after sampling are verified by the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.

Preferably, in step S2, the types of global optimization methods include particle swarm optimization algorithm and evolutionary computation algorithm.

Preferably, in step S2, the method for fitting the sample into a single admittance sample is: using a coupled mode model method to calculate the admittance of the resonator, and adjusting the parameters of the coupled mode model in the calculation so that the admittance calculated by the coupled mode model method is the same as the admittance of the actual test.

Preferably, in step S3, the test original data includes a first geometric parameter, a first test file, a second geometric parameter and a second test file; wherein, the first geometric parameter is a plurality of different geometric parameters in the tape-out test data of the same batch of the surface acoustic filter; the first test file is a plurality of different test files in the tape-out test data of the same batch of the surface acoustic filter; the second geometric parameter is the same geometric parameter on different wafers and at different positions in the tape-out test data of the surface acoustic filter; and the second test file is the same test file on different wafers and at different positions in the tape-out test data of the surface acoustic filter.

Preferably, in step S3, the distribution effect of the plurality of admittance samples is an average value of statistical indicators of deviation indicators of the resonator, and the deviation indicators of the resonator include oscillation point frequency deviation, anti-resonance point frequency deviation and static capacitance deviation.

Preferably, the step S5 specifically includes:

Step S51, evaluating the fitting effect of all the parameter results according to a preset reference index, and removing the admittance samples corresponding to the parameter results exceeding the index threshold in the fitting effect;

Step S52, calculating the average value of a plurality of groups of parameter results corresponding to the resonators with the same geometric parameters in the parameter results, and taking the calculated average value data as a group of parameter results.

Preferably, before step S6, the step further includes:

Step S60: determining whether the amount of the input data and/or the output data is large or not The judgment is made based on the preset data volume threshold:

If yes, the machine learning model is selected as a deep learning model;

If not, the machine learning model is selected as a decision tree model.

In a second aspect, an embodiment of the present invention further provides a simulation test fitting system, wherein the simulation test fitting system applies the above-mentioned simulation test fitting process parameter method provided by an embodiment of the present invention;

The simulation test fitting system comprises a simulation test fitter, a trainer and a verifier connected in sequence;

The simulation test fitter is used to implement the steps S1 to S5 of the simulation test fitting process parameter method; specifically: to implement the step S1, obtain the test raw data of multiple resonators, clean the test raw data according to the preset cleaning rules and obtain the cleaned samples; the test raw data includes the geometric parameters and test files for matching the resonators, and the cleaning rules are to remove the inconsistent parts according to the statistical consistency of the test raw data; it is also used to implement the step S2, set constraints for the preset global class optimization method, and then fit the sample into a single admittance sample by the global class optimization method after setting the constraints; the constraints are the type of the selected global class optimization method and the related parameters in the global class optimization method, and the admittance samples, the type of the global class optimization method, and the global class optimization method are selected according to the parameters related to the global class optimization method. The method is used to correspond to the parameters related to the optimization method one by one; it is also used to implement the step S3, sample all the samples, and then fit the sampled samples according to the step S2 to obtain multiple sampled admittance samples, and then judge whether the type of the global optimization method and the distribution effect of the parameters related to the global optimization method in the multiple sampled admittance samples are less than the preset index threshold: if so, enter step S4; if not, adjust the constraint and return to step S2; it is also used to implement the step S4, according to all the samples Distributed calculation and fitting to obtain all the admittance samples; it is also used to implement the step S5, according to the data in all the admittance samples, obtain multiple groups of parameter results, merge the multiple groups of parameter results corresponding to the resonator with the same geometric parameters into one group for processing, and then output the processed multiple groups of parameter results;

The trainer is used to implement the step S6, preselect a machine learning model, use the geometric parameters of the resonator in the step S1 as input data of the machine learning model, use the multiple groups of parameter results output in the step S5 as output data of the machine learning model, train the machine learning model and obtain a trained prediction adjustment parameter model;

The verifier is used to implement the step S7, sampling all the samples in the step S2, and then verifying the sampled samples through the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.

In a third aspect, an embodiment of the present invention further provides a simulation test fitting device, comprising a processor and a memory, wherein the processor is used to read the program in the memory and execute the steps in the above-mentioned simulation test fitting process parameter method provided in an embodiment of the present invention.

In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the steps in the above-mentioned simulation test fitting process parameter method provided in an embodiment of the present invention are implemented.

Compared with the prior art, the simulation test fitting process parameter method of the present invention implements steps S1 and S2, wherein, in step S1, the test raw data is cleaned and samples are obtained; in step S2, the samples are fitted to a single admittance sample; the implementation of steps S1 and S2 realizes accurate simulation test of a single admittance curve by selecting a reasonable optimization method. The simulation test fitting process parameter method of the present invention then implements steps S3 and S4, wherein, in step S3, the sampled samples are fitted to obtain multiple sampled admittance samples; in step S4, all samples are distributedly calculated and fitted to obtain all admittance samples; the implementation of steps S3 and S4 realizes rapid fitting of a large number of tests through code-level parallelism. The simulation test fitting process parameter method of the present invention then implements step S5, wherein, in step S5, multiple groups of parameter results corresponding to resonators with the same geometric parameters are merged into one group for processing; the implementation of step S5 can be further screened to obtain average data that suppresses process and test fluctuations. The simulation test fitting process parameter method of the present invention further implements step S6, wherein step S6, trains the machine learning model and obtains the prediction adjustment parameter model; and implements step S6 to obtain a reliable continuous model through the machine learning model. The fitting process parameter method is then implemented through step S7, wherein, in step S7, the sampled sample is verified by the prediction adjustment parameter model. Implementing step S7 can improve the accuracy and reliability of the simulation test fitting process parameter method of the present invention. By executing the above steps, the entire process fully automatically realizes the automatic and rapid simulation test fitting of different process characteristic foundries, different piezoelectric substrates, and different types of surface acoustic waves of the surface acoustic wave filter. Under the conditions of the original process and test fluctuations, a statistically accurate simulation test comparison is achieved. More preferably, the simulation test fitting process parameter method of the present invention makes the process of adapting different processes or different foundries very fast, eliminates the process of manually adjusting COM parameters repeatedly, and can ensure universality and ensure data accuracy over a wide range. Therefore, the simulation test fitting process parameter method, simulation test fitting system, simulation test fitting device, and computer-readable storage medium of the present invention can solve the problem of realizing statistically automated simulation testing under different processes and test fluctuations, and have good universality, fast calculation speed, and high accuracy.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

FIG1 is a flowchart of a simulation test fitting process parameter method provided by an embodiment of the present invention;

FIG2 is a flowchart of step S5 of the simulation test fitting process parameter method provided by an embodiment of the present invention;

FIG3 is a flowchart of step S60 of the simulation test fitting process parameter method provided by an embodiment of the present invention;

4 is a graph showing the relationship between the frequency and sampling point curves of the resonance point of the test data before the simulation test fitting is performed in the simulation test fitting process parameter method provided by an embodiment of the present invention;

5 is a graph showing the relationship between the frequency and sampling point curves of the resonance point of the test data after the simulation test fitting is performed in the simulation test fitting process parameter method provided by an embodiment of the present invention;

FIG6 is a schematic diagram of the structure of a simulation test fitting system provided by an embodiment of the present invention;

FIG. 7 is a schematic diagram of the structure of a simulation test fitting device provided in an embodiment of the present invention.

Detailed ways

The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

The terms "including" and "having" and any variations thereof in the specification, claims and drawings of the present application are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification, claims or drawings of the present application are used to distinguish different objects rather than to describe a specific order. Reference to "an embodiment or the present implementation mode" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

The present invention provides a simulation test fitting process parameter method. The simulation test fitting process parameter method is applied to a surface acoustic wave filter. Specifically, the simulation test fitting process parameter method is applied to the electronic design automation (English: Electronic design automation, abbreviation: EDA) software required for the automatic design of the surface acoustic wave filter. The surface acoustic wave filter includes a plurality of the resonators. In this embodiment, the surface acoustic wave filter is a ladder surface acoustic wave filter and is composed of a plurality of the resonators electrically cascaded.

Please refer to FIG. 1 , which is a flowchart of a simulation test fitting process parameter method provided by an embodiment of the present invention.

The simulation test fitting process parameter method comprises the following steps:

Step S1, obtaining the original test data of the plurality of resonators, and cleaning the resonators according to the preset The rule cleans the test raw data and obtains cleaned samples.

The test raw data includes geometric parameters and test files for matching the resonator, and the cleaning rule is to remove inconsistent parts according to the statistical consistency of the test raw data, wherein the inconsistent parts are parts with irregular statistical rules.

For example, the resonant frequencies calculated for the same resonator in different test files are counted and the variance is calculated. If the variance is small, the part with a variance greater than the set threshold is discarded; if the variance is large, all data of the resonator is discarded. Samples with obvious errors due to testing are automatically removed according to the cleaning rules (for example, the frequency difference between the resonant point and the anti-resonant point of the resonator is limited to not be greater than 1GHz), so as to ensure the correctness of all samples.

Step S2: setting constraints for a preset global optimization method, and then fitting the sample into a single admittance sample by using the global optimization method after setting the constraints.

The constraints are the type of the selected global optimization method and the related parameters in the global optimization method, and the admittance samples, the type of the global optimization method, and the related parameters in the global optimization method correspond one to one.

In this embodiment, in step S2, the types of global optimization methods include particle swarm optimization algorithm and evolutionary computation algorithm.

In this embodiment, in step S2, the method of fitting the sample to obtain a single admittance sample is: using a coupled mode model method to calculate the admittance of the resonator, and adjusting the parameters of the coupled mode model in the calculation so that the admittance calculated by the coupled mode model method is the same as the admittance of the actual test. The fitting process in step S2 relies on the coupled mode model method to calculate the resonator admittance, and the process of fitting a single sample is to find a suitable coupled mode model adjustment parameter so that the admittance calculated by the coupled mode model method is as close as possible to the admittance of the actual test.

The fitting of step S2 needs to find a suitable coupled mode model to adjust the parameter range according to the center frequency of the admittance to be fitted, and try to reduce the possibility of the optimization method falling on different local optimal solutions under the premise of satisfying the optimization space size. Reduce the fluctuation of repeated fitting results, The global optimization method itself also corresponds to some parameters, such as the maximum number of iterations, which need to be determined by repeated attempts to observe the actual effect. Ultimately, the fitting result of a single admittance curve is as close to the test result as possible.

By implementing step S1 and step S2 and selecting a reasonable optimization method, accurate simulation test of a single admittance curve is achieved.

If yes, proceed to step S4; if no, return to step S2 after adjusting the constraint.

In this embodiment, in step S3, the test original data includes a first geometric parameter, a first test file, a second geometric parameter and a second test file; wherein, the first geometric parameter is a plurality of different geometric parameters in the same batch of tape-out test data of the surface acoustic filter; the first test file is a plurality of different test files in the same batch of tape-out test data of the surface acoustic filter; the second geometric parameter is the same geometric parameter on different wafers and at different positions in the tape-out test data of the surface acoustic filter; the second test file is the same test file on different wafers and at different positions in the tape-out test data of the surface acoustic filter. The sample is the test data, and the test data of the same batch of tape-outs contain both the test results of the resonators with different geometric parameters and the test results of the resonators with the same geometric parameters on different wafers and at different positions. The tape-out needs to include the resonators with different geometric parameters in order to meet the diversity of the resonator structure and to provide parameters for simulation test. The test results of the resonators with the same geometric parameters on different wafers and at different positions are to take into account the process and test fluctuations so that the calculated parameters obtained by the simulation test are as close as possible to the actual mass production situation.

In this embodiment, in step S3, the distribution effect of the plurality of admittance samples is the average value of the statistical index of the deviation index of the resonator, and the deviation index of the resonator includes the vibration point frequency deviation, the anti-resonance point frequency deviation and the static capacitance deviation.

The effect of step S3 on all the samples can be evaluated by statistical analysis. The average value of the measurement index is used to measure, such as the statistical average resonance point frequency deviation (for example, the fitting resonance point frequency deviation is compared with the test resonance point frequency deviation), anti-resonance point frequency deviation, static capacitance deviation, etc. If the average values of all the indexes are less than the preset index threshold, the process proceeds to step S4.

Step S4: Perform distributed calculation on all the samples and fit them to obtain all the admittance samples.

In this embodiment, the implementation of step S4 utilizes 40 processes, and it only takes about 5 hours to perform the fitting calculation of 30,000 admittance samples. Therefore, the implementation of step S4 is a fast fitting of a large number of tests.

The implementation of step S3 and step S4 achieves fast fitting of a large number of tests through parallelism at the code level.

Step S5, obtaining multiple groups of parameter results based on the data in all the admittance samples, combining the multiple groups of parameter results corresponding to the resonators with the same geometric parameters into one group for processing, and then outputting the processed multiple groups of parameter results.

Please refer to FIG. 2 , which is a flowchart of step S5 of the simulation test fitting process parameter method provided by an embodiment of the present invention. The step S5 specifically includes:

Step S51, evaluate the fitting effect of all the parameter results according to the preset reference index, and remove the admittance samples corresponding to the parameter results exceeding the index threshold in the fitting effect. The fitting effect is the index of all the admittance samples fitted by implementing step S4, such as the resonance point frequency deviation. In this embodiment, the fitting effect is greater than 0.5MHz relative to the resonance point frequency deviation of the test result.

The fluctuation of process and test is inevitable. It is generally believed that the resonator frequency fluctuation caused by process is about 3MHz, while in this embodiment, the admittance fluctuation of the resonator caused by test can exceed 3db. The averaging process in step S52 can actually find the middle value between process and test.

Step S5 can be implemented to further screen and obtain average data that suppresses process and test fluctuations.

Step S6, pre-select a machine learning model, use the geometric parameters of the resonator in step S1 as input data of the machine learning model, use the multiple groups of parameter results output in step S5 as output data of the machine learning model, train the machine learning model and obtain a trained prediction adjustment parameter model.

The machine learning model in step S6 is a commonly used model in the art, such as a vanilla neural network model and a deep learning model.

Step S6 is implemented to obtain a reliable continuous model through the machine learning model.

Please refer to FIG. 3 , which is a flowchart of step S60 of the simulation test fitting process parameter method provided by an embodiment of the present invention.

In this embodiment, before step S6, the following steps are also included:

Step S60: judging whether the amount of the input data and/or the output data is greater than a preset data amount threshold:

If yes, the machine learning model is selected as a deep learning model;

If not, the machine learning model is selected as a decision tree model.

In this embodiment, the deep learning model is a machine learning model corresponding to the MLP method. The decision tree model is a machine learning model corresponding to the Xgboost method.

Step S60 automatically adapts different types of machine learning methods according to the size of the data. For example, when the amount of data is large, deep learning can be used for training; when the amount of data is small, a decision tree model can be used for training. After step S60, step S6 obtains a continuous prediction adjustment parameter model from discrete data.

Step S7: sampling all the samples in step S2, and then verifying the sampled samples through the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.

The verification in step S7 is specifically as follows: bringing the predicted adjustment parameters of the prediction adjustment parameter model back to step S2, adjusting the parameters to update the constraints in step S2, performing resonator admittance calculation and obtaining the calculation results, and comparing the calculation results with the fitting target. If the comparison is consistent, the verification is passed; if the comparison is inconsistent, the verification fails.

Implementing step S7 can improve the accuracy and reliability of the simulation test fitting process parameter method of the present invention.

By executing the above steps S1 to S7, the entire process of the simulation test fitting process parameter method of the present invention fully and automatically realizes the automatic and rapid simulation test fitting of different process characteristic foundries, different piezoelectric substrates, and different types of surface acoustic waves of the surface acoustic wave filter. Under the conditions of the original process and test fluctuations, a statistically accurate simulation test comparison is achieved. More preferably, the simulation test fitting process parameter method of the present invention makes the process of adapting different processes or different foundries very fast, eliminating the process of manually adjusting COM parameters repeatedly, and can ensure universality and ensure data accuracy over a wide range. Therefore, the simulation test fitting process parameter method of the present invention can solve the problem of realizing statistically automated simulation testing under the conditions of different processes and test fluctuations, and has good universality, fast calculation speed and high accuracy.

The following is a set of data to demonstrate the effect of the simulation test fitting process parameter method in practice. This set of data uses the simulation test fitting process parameter method of the present invention to fit 20,000 sets of test data of the 1G frequency band, and samples show the effect before and after fitting.

Please refer to Figures 4 and 5 at the same time. Figure 4 is a curve relationship diagram of the frequency and sampling points of the resonance point of the test data before the simulation test fitting is implemented in the simulation test fitting process parameter method provided in an embodiment of the present invention. Among them, Figure 4 is a curve relationship diagram of the frequency and sampling points of the resonance point when the average resonance point frequency deviation of the original data is 0.36MHz. Figure 5 is a curve relationship diagram of the frequency and sampling points of the resonance point of the test data after the simulation test fitting is implemented in the simulation test fitting process parameter method provided in an embodiment of the present invention. Figure 5 is a curve relationship diagram of the frequency and sampling points of the resonance point when the resonance point frequency deviation is 1.03MHz after the simulation test. As shown in Figures 4 and 5, the simulation test fitting process parameter method of the present invention can solve the problem of realizing automated simulation testing in a statistical sense under different processes and test fluctuations, and has good universality, fast calculation speed and high accuracy.

The present invention also provides a simulation test fitting system 100. Please refer to FIG6, which is a schematic diagram of the structure of the simulation test fitting system 100 of the present invention. System 100 applies the simulation test fitting process parameter method of the present invention.

Specifically, the simulation test fitting system 100 includes a connected simulation test fitter 1, a trainer 2 and a verifier 3. In this embodiment, the simulation test fitter 1, the trainer 2 and the verifier 3 are all digital processors or software programs.

The simulation test fitter is used to implement the steps S1 to S5 of the simulation test fitting process parameter method.

Specifically, the simulation test fitter 1 is used to implement the step S1, obtain the test raw data of multiple resonators, clean the test raw data according to the preset cleaning rules and obtain the cleaned samples. The test raw data includes geometric parameters and test files for matching the resonators. The cleaning rules are to remove the inconsistent parts according to the statistical consistency of the test raw data.

The simulation test fitter 1 is also used to implement the step S2, set constraints for the preset global optimization method, and then fit the sample to a single admittance sample by the global optimization method after setting the constraints. The constraints are the type of the selected global optimization method and the related parameters in the global optimization method. The admittance samples, the type of the global optimization method, and the related parameters in the global optimization method correspond one to one.

The simulation test fitter 1 is also used to implement the step S3, sampling all the samples, and then fitting the sampled samples according to the step S2 to obtain multiple sampled admittance samples, and then judging whether the distribution effect of the type of the global class optimization method and the relevant parameters in the global class optimization method in the multiple sampled admittance samples is less than a preset indicator threshold: if so, enter step S4; if not, adjust the constraints and return to step S2.

The simulation test fitter 1 is also used to implement the step S4, by performing distributed calculations on all the samples and fitting them to obtain all the admittance samples.

The simulation test fitter 1 is also used to implement the step S5, obtain multiple groups of parameter results based on the data in all the admittance samples, merge the multiple groups of parameter results corresponding to the resonators with the same geometric parameters into one group for processing, and then output the processed multiple groups of parameter results.

The trainer 2 is used to implement the step S6, pre-select a machine learning model, use the geometric parameters of the resonator in the step S1 as input data of the machine learning model, use the multiple groups of parameter results output in the step S5 as output data of the machine learning model, train the machine learning model and obtain the trained prediction adjustment parameter model.

The verifier 3 is used to implement the step S7, sampling all the samples in the step S2, and then verifying the sampled samples through the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.

The simulation test fitting system 100 provided in the embodiment of the present invention can implement various implementation methods and corresponding beneficial effects in the embodiment of the simulation test fitting process parameter method, which will not be described here to avoid repetition.

The present invention further provides a simulation test fitting device 1000. Please refer to Fig. 7, which is a schematic diagram of the structure of the simulation test fitting device 1000 of the present invention.

The simulation test fitting device 1000 includes a processor 1001, a memory 1002, a network interface 1003, and a computer program stored in the memory 1002 and executable on the processor 1001. The processor 1001 is used to read the program in the memory 1002. When the processor 1001 executes the computer program, the steps in the simulation test fitting process parameter method provided in the embodiment are implemented. That is, the processor 1001 executes the steps in the simulation test fitting process parameter method.

Specifically, the processor 1001 is configured to perform the following steps:

Step S1, obtaining the test raw data of a plurality of resonators, cleaning the test raw data according to a preset cleaning rule and obtaining a cleaned sample. The test raw data includes geometric parameters and a test file for matching the resonator. The cleaning rule is to remove the inconsistent parts according to the statistical consistency of the test raw data.

Step S2: set constraints for the preset global optimization method, and then fit the sample into a single admittance sample by the global optimization method after setting the constraints. The constraints are the type of the selected global optimization method and the related parameters in the global optimization method. The relevant parameters in the method correspond one to one.

If yes, proceed to step S4; if no, adjust the constraint and return to step S2. Step S4: Perform distributed calculation on all the samples and fit them to obtain all the admittance samples.

The simulation test fitting device 1000 provided in the embodiment of the present invention can implement various implementation methods and corresponding beneficial effects in the embodiment of the simulation test fitting process parameter method, which will not be described here to avoid repetition.

It should be noted that FIG. 7 only shows 1001-1003 with components, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the simulation test fitting device 1000 here is a device that can automatically perform numerical calculations and/or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), programmable gate arrays (FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

The memory 1002 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 1002 can be an internal storage unit of the simulation test fitting device 1000, such as a hard disk or memory of the simulation test fitting device 1000. In other embodiments, the memory 1002 can also be an external storage device of the simulation test fitting device 1000, such as a plug-in hard disk equipped on the simulation test fitting device 1000, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 1002 can also include both the internal storage unit of the simulation test fitting device 1000 and its external storage device. In this embodiment, the memory 1002 is generally used to store an operating system and various application software installed in the simulation test fitting device 1000, such as program codes of the simulation test fitting process parameter method of the simulation test fitting device 1000. In addition, the memory 1002 can also be used to temporarily store various data that have been output or are to be output.

The processor 1001 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 1001 is generally used to control the overall operation of the simulation test fitting device 1000. In this embodiment, the processor 1001 is used to run the program code or process data stored in the memory 1002, such as running the program code of the simulation test fitting process parameter method of the simulation test fitting device 1000.

The network interface 1003 may include a wireless network interface or a wired network interface, and the network interface 1003 is generally used to establish a communication connection between the simulation test fitting device 1000 and other electronic devices.

The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, wherein the program instructions When executed by the processor 1001, the steps in the simulation test fitting process parameter method as described above are implemented.

A person skilled in the art can understand that all or part of the processes in the simulation test fitting process parameter method of the simulation test fitting device 1000 of the embodiment can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of each method. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

The embodiments mentioned in the embodiments of the present invention are for the convenience of description. The above disclosure is only the preferred embodiment of the present invention, and of course it cannot be used to limit the scope of the rights of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Compared with the prior art, the simulation test fitting process parameter method of the present invention implements steps S1 and S2, wherein, in step S1, the test raw data is cleaned and samples are obtained; in step S2, the samples are fitted to a single admittance sample; the implementation of steps S1 and S2 realizes accurate simulation test of a single admittance curve by selecting a reasonable optimization method. The simulation test fitting process parameter method of the present invention then implements steps S3 and S4, wherein, in step S3, the sampled samples are fitted to obtain multiple sampled admittance samples; in step S4, all samples are distributedly calculated and fitted to obtain all admittance samples; the implementation of steps S3 and S4 realizes rapid fitting of a large number of tests through code-level parallelism. The simulation test fitting process parameter method of the present invention then implements step S5, wherein, in step S5, multiple groups of parameter results corresponding to resonators with the same geometric parameters are merged into one group for processing; the implementation of step S5 can be further screened to obtain average data that suppresses process and test fluctuations. The simulation test fitting process parameter method of the present invention is then implemented through step S6, wherein step S6, trains the machine learning model and obtains the prediction adjustment parameter model; and step S6 is implemented to obtain a reliable continuous model through the machine learning model. The simulation test fitting process parameter method of the present invention is then implemented through step S7, wherein step S7, the sampled sample is verified through the prediction adjustment parameter model. Implementing step S7 can improve the simulation of the present invention. The accuracy and reliability of the true test fitting process parameter method. By executing the above steps, the entire process fully automates the automatic and rapid simulation test fitting of different process characteristic foundries, different piezoelectric substrates, and different types of surface acoustic waves of the surface acoustic wave filter. Under the conditions of the original process and test fluctuations, a statistically accurate simulation test comparison is achieved. More preferably, the simulation test fitting process parameter method of the present invention makes the process of adapting to different processes or different foundries very fast, eliminating the process of manually adjusting COM parameters repeatedly, and can ensure universality and ensure data accuracy over a wide range. Therefore, the simulation test fitting process parameter method, simulation test fitting system, simulation test fitting device, and computer-readable storage medium of the present invention can solve the problem of realizing statistically automated simulation testing under different processes and test fluctuations, and have good universality, fast calculation speed, and high accuracy.

The above description is only an implementation mode of the present invention. It should be pointed out that, for ordinary technicians in this field, improvements can be made without departing from the creative concept of the present invention, but these all belong to the protection scope of the present invention.

Claims

A simulation test fitting process parameter method is applied to a surface acoustic wave filter, wherein the surface acoustic wave filter includes a plurality of resonators; the method is characterized in that the method comprises the following steps:

Step S1, obtaining test raw data of a plurality of resonators, cleaning the test raw data according to a preset cleaning rule and obtaining cleaned samples; the test raw data includes geometric parameters and test files for matching the resonators, and the cleaning rule is to remove inconsistent parts according to the statistical consistency of the test raw data;

Step S2, setting constraints for the preset global optimization method, and then fitting the sample into a single admittance sample by the global optimization method after setting the constraints; the constraints are the type of the selected global optimization method and the relevant parameters in the global optimization method, and the admittance sample, the type of the global optimization method, and the relevant parameters in the global optimization method correspond to each other one by one;

Step S3, sampling all the samples, and then fitting the sampled samples according to the step S2 to obtain multiple sampled admittance samples, and then judging whether the distribution effect of the type of the global class optimization method and the relevant parameters in the global class optimization method in the multiple sampled admittance samples is less than a preset indicator threshold:

If yes, proceed to step S4; if no, adjust the constraint and return to step S2;

Step S4, performing distributed calculation on all the samples and fitting to obtain all the admittance samples;

Step S5, obtaining multiple groups of parameter results according to the data in all the admittance samples, combining the multiple groups of parameter results corresponding to the resonators with the same geometric parameters into one group for processing, and then outputting the processed multiple groups of parameter results;

Step S6, preselecting a machine learning model, using the geometric parameters of the resonator in step S1 as input data of the machine learning model, using the multiple groups of parameter results output in step S5 as output data of the machine learning model, training the machine learning model and obtaining a trained prediction adjustment parameter model;

Step S7: Sampling all the samples in step S2, and then The samples after sampling are verified by the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.
The simulation test fitting process parameter method according to claim 1 is characterized in that, in step S2, the types of global optimization methods include particle swarm optimization algorithm and evolutionary computing algorithm.
According to the simulation test fitting process parameter method of claim 1, it is characterized in that in the step S2, the method of fitting the sample into a single admittance sample is: using a coupled mode model method to calculate the admittance of the resonator, and adjusting the parameters of the coupled mode model in the calculation so that the admittance calculated by the coupled mode model method is the same as the admittance of the actual test.
According to the simulation test fitting process parameter method according to claim 1, it is characterized in that, in the step S3, the test original data includes a first geometric parameter, a first test file, a second geometric parameter and a second test file; wherein, the first geometric parameter is a plurality of different geometric parameters in the tape-out test data of the same batch of the SAW filters; the first test file is a plurality of different test files in the tape-out test data of the same batch of the SAW filters; the second geometric parameter is the same geometric parameter on different wafers and at different positions in the tape-out test data of the SAW filters; the second test file is the same test file on different wafers and at different positions in the tape-out test data of the SAW filters.
According to the simulation test fitting process parameter method of claim 1, it is characterized in that in the step S3, the distribution effect in the multiple admittance samples is the average value of the statistical index of the deviation index of the resonator, and the deviation index of the resonator includes the vibration point frequency deviation, the anti-resonance point frequency deviation and the static capacitance deviation.
The simulation test fitting process parameter method according to claim 1 is characterized in that the step S5 specifically comprises:

Step S51, evaluating the fitting effect of all the parameter results according to a preset reference index, and removing the admittance samples corresponding to the parameter results exceeding the index threshold in the fitting effect;

Step S52: the corresponding resonators of the same geometric parameters in the parameter results The average value of the plurality of groups of parameter results is calculated, and the calculated average value data is used as a group of parameter results.
The simulation test fitting process parameter method according to claim 1, characterized in that before step S6, it also includes:

Step S60: judging whether the amount of the input data and/or the output data is greater than a preset data amount threshold:

If yes, the machine learning model is selected as a deep learning model;

If not, the machine learning model is selected as a decision tree model.
A simulation test fitting system, characterized in that the simulation test fitting system applies the simulation test fitting process parameter method according to any one of claims 1 to 7;

The simulation test fitting system comprises a simulation test fitter, a trainer and a verifier connected in sequence;

The simulation test fitter is used to implement the steps S1 to S5 of the simulation test fitting process parameter method; specifically: to implement the step S1, obtain the test raw data of multiple resonators, clean the test raw data according to the preset cleaning rules and obtain the cleaned samples; the test raw data includes the geometric parameters and test files for matching the resonators, and the cleaning rules are to remove the inconsistent parts according to the statistical consistency of the test raw data; it is also used to implement the step S2, set constraints for the preset global class optimization method, and then fit the sample into a single admittance sample through the global class optimization method after setting the constraints; the constraints are to select The type of the global optimization method and the related parameters in the global optimization method, the admittance samples, the type of the global optimization method, and the related parameters in the global optimization method correspond one to one; it is also used to implement the step S3, sampling all the samples, and then fitting the sampled samples according to the step S2 to obtain a plurality of sampled admittance samples, and then judging whether the distribution effect of the type of the global optimization method and the related parameters in the global optimization method in the plurality of sampled admittance samples is less than a preset index threshold: if so, enter step S4; if not, adjust the constraints and return to step S2; it is also used to implement the step S4, according to the classification of all the samples The method comprises the following steps: performing Bournard calculation and fitting to obtain all the admittance samples; and implementing the step S5 to obtain multiple groups of parameter results according to the data in all the admittance samples, combining the multiple groups of parameter results corresponding to the resonators having the same geometric parameters into one group for processing, and then outputting the processed multiple groups of parameter results;

The trainer is used to implement the step S6, preselect a machine learning model, use the geometric parameters of the resonator in the step S1 as input data of the machine learning model, use the multiple groups of parameter results output in the step S5 as output data of the machine learning model, train the machine learning model and obtain a trained prediction adjustment parameter model;

The verifier is used to implement the step S7, sampling all the samples in the step S2, and then verifying the sampled samples through the prediction adjustment parameter model to ensure that the fitting results of the sampling meet the preset requirements.
A simulation test fitting device, characterized in that it includes a processor and a memory, wherein the processor is used to read the program in the memory and execute the steps in the simulation test fitting process parameter method as described in any one of claims 1 to 7.
A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, the computer program includes program instructions, and when the program instructions are executed by a processor, the steps in the simulation test fitting process parameter method as described in any one of claims 1 to 7 are implemented.