Machine learning methods, systems, and computer media for high-throughput piezoelectric coefficient detection

Abstract

The present application relates to the technical field of artificial intelligence, in particular to a machine learning method and system for high-throughput piezoelectric coefficient detection and a computer medium, the machine learning method for high-throughput piezoelectric coefficient detection comprising the following steps: providing a piezoelectric film, obtaining sample points at multiple positions in the film, generating piezoelectric coefficient samples corresponding to the sample points as input data for machine learning; establishing a coordinate system based on the sample points, obtaining the coordinates and piezoelectric coefficient values of the piezoelectric coefficient samples in the coordinate system; obtaining multiple initial cluster centers and initial cluster center coordinates in the coordinate system in a preset manner; clustering in a preset manner based on the above results to obtain new cluster centers and new cluster center coordinates; converging the cluster centers in a preset iteration manner to obtain converged cluster centers and converged cluster center coordinates; planning a preset detection probe movement path based on the converged cluster centers and the converged cluster center coordinates, and taking the path as output data for machine learning. The detection efficiency is greatly improved.

Description

[Technical Field]

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a machine learning method, system, and computer medium for high-throughput piezoelectric coefficient detection. [Background Technology]

[0002] In recent years, piezoelectric thin film materials have been widely used in fingerprint recognition, measurement, infrared, security alarms, healthcare, information engineering, office automation, marine development, geological exploration, and other technical fields. Among these, piezoelectric thin films used in ultrasonic fingerprint sensors have high requirements for various technical indicators, such as the value and distribution of the piezoelectric coefficient. Achieving the required performance of these films is extremely time-consuming and labor-intensive, thus placing high demands on the speed and accuracy of piezoelectric coefficient measurement.

[0003] Most existing piezoelectric coefficient testing equipment uses manual testing methods, requiring manual peeling of the film for inspection. This not only damages the film, increasing production and testing costs, but also consumes manpower and is too slow. While modified non-destructive testing equipment achieves the goal of non-destructive testing, the point-by-point testing method is more suitable for experiments and remains too slow for industrial production testing. [Summary of the Invention]

[0004] To improve the efficiency of piezoelectric coefficient detection, this invention provides a machine learning method, system, and computer medium for high-throughput piezoelectric coefficient detection.

[0005] The present invention provides a machine learning method for high-throughput piezoelectric coefficient detection, comprising the following steps: providing a piezoelectric thin film; acquiring sample points at multiple locations within the piezoelectric thin film; generating piezoelectric coefficient samples corresponding to the sample points, which serve as input data for machine learning; establishing a coordinate system based on the edge lines of the sample points; acquiring the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in the coordinate system; acquiring multiple initial cluster centers and their coordinates in the coordinate system using a preset acquisition method; clustering the piezoelectric coefficient samples, their values, initial cluster centers, and their coordinates using a preset clustering method to obtain new cluster centers and their coordinates; converging the cluster centers using a preset iteration method to obtain converged cluster centers and their coordinates; and planning a preset detection probe movement path based on the converged cluster centers and their coordinates, which serves as output data for machine learning.

[0006] Preferably, the initial cluster center coordinates are the same as or different from the piezoelectric coefficient sample coordinates.

[0007] Preferably, obtaining the initial cluster center and initial cluster center coordinates in a preset acquisition method specifically includes the following steps: obtaining the number of initial cluster centers based on a preset target optimization rate; uniformly and randomly dispersing the initial cluster centers in the coordinate system, and obtaining the coordinates of the dispersed initial cluster centers.

[0008] Preferably, before clustering using a preset clustering method, a clustering boundary value is set. The preset clustering method for clustering the initial cluster center and its coordinates is performed according to the following formula:

[0009]

[0010] Among them, each piezoelectric coefficient sample coordinates , Here is the piezoelectric coefficient of this piezoelectric coefficient sample, and the subscript k is the initial number of cluster centers. The coordinates of each initial cluster center are... , The piezoelectric coefficient value of the initial cluster center. To determine the indicators, In this context, 'a' represents the clustering boundary value.

[0011] Preferably, the clustering of the initial cluster center and its coordinates must meet the following clustering conditions, which include the following judgment process: determining whether the difference between the piezoelectric coefficient of the piezoelectric coefficient sample and the piezoelectric coefficient of the initial cluster center is within the clustering boundary value range; if it is outside the clustering boundary value range, then wait for the next clustering in the iteration process; if it is within the clustering boundary value range, then compare the magnitude of the judgment index corresponding to each initial cluster center and obtain the minimum value among the judgment indices; mark the minimum value among the judgment indices, and assign the initial cluster center corresponding to the minimum value to the new cluster of this iteration.

[0012] Preferably, the initial cluster center and its coordinates are clustered using a preset clustering method, which specifically includes the following steps: obtaining a preset range with the initial cluster center as the center, and obtaining the standard deviation with the distance from the piezoelectric coefficient sample to the nearest initial cluster center as the radius; based on the initial cluster center and the piezoelectric coefficient sample, obtaining a set of piezoelectric coefficient samples that meet the standard deviation, as a new cluster.

[0013] Preferably, the clustering of the initial cluster center and its coordinates using a preset clustering method further includes the following step of determining the new cluster center: traversing all piezoelectric coefficient samples in the new cluster, so that...

[0014] The point with the smallest value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples; or, all piezoelectric coefficient samples in the new cluster are traversed to calculate the theoretical cluster center value. ,make The point closest to the theoretical cluster center value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples.

[0015] Preferably, the process of convergent cluster centers and their coordinates using a preset iterative method specifically includes the following steps: defining an initial cluster center coordinate as an initial sample, and forming an initial sample set from all initial samples. The initial sample set is then clustered using a preset clustering method to obtain a new set of cluster center coordinates, which is considered primary data. Multiple primary data sets are defined as ensemble data sets, which are then used as new samples. Multiple new initial cluster centers are selected from these new samples. Based on the selected new initial cluster centers, the above steps of clustering and obtaining new cluster centers and their coordinates are repeated until the new initial cluster centers converge, thus obtaining the convergent cluster centers and their coordinates.

[0016] To address the aforementioned technical problems, this invention also provides a high-throughput piezoelectric coefficient machine learning system for implementing the high-throughput piezoelectric coefficient detection machine learning method described above. The system includes an information acquisition module for acquiring piezoelectric coefficient samples at multiple locations in the piezoelectric film, machine learning input data, the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in a coordinate system, multiple initial cluster centers and their coordinates in the coordinate system, new cluster centers and their coordinates after clustering, converged cluster centers and their coordinates, and machine learning output data. A data processing module is also included, which establishes a coordinate system based on the edge lines of the sample points, clusters the initial cluster centers and their coordinates using a preset clustering method, converges the cluster centers using a preset iteration method, and plans a preset detection probe movement path based on the converged cluster centers and their coordinates.

[0017] To solve the above-mentioned technical problems, the present invention also provides a storage medium, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the machine learning method for high-throughput piezoelectric coefficient detection as described above.

[0018] Compared with existing technologies, the machine learning method, system, and computer medium for high-throughput piezoelectric coefficient detection of the present invention have the following advantages:

[0019] 1. The high-throughput piezoelectric coefficient detection machine learning method of the present invention includes the following steps: providing a piezoelectric thin film, acquiring sample points at multiple locations in the piezoelectric thin film, generating piezoelectric coefficient samples corresponding to the sample points, and using the piezoelectric coefficient samples as input data for machine learning; establishing a coordinate system based on the edge lines of the sample points, and acquiring the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in the coordinate system; acquiring multiple initial cluster centers and their coordinates in the coordinate system using a preset acquisition method; clustering the piezoelectric coefficient samples, their piezoelectric coefficient values, initial cluster centers, and their coordinates using a preset clustering method to obtain new cluster centers and their coordinates; converging the cluster centers using a preset iteration method to obtain converged cluster centers and their coordinates; and planning a preset detection probe movement path based on the converged cluster centers and their coordinates, using the preset detection probe movement path as output data for machine learning. The cluster center is the mean value of the piezoelectric coefficient in the cluster. The present invention uses a modified mean value algorithm to divide the clusters obtained by clustering based on the given samples. This reflects the tightness of the samples within the cluster around the central piezoelectric coefficient detection point. It finds the points with extremely high similarity among the samples within the cluster and performs a clustering operation. Thus, it only needs to detect the initial cluster center point and ignores other piezoelectric coefficient samples with similar characteristics, thereby simplifying the detection, reducing the detection time and labor costs, and greatly improving the detection efficiency.

[0020] 2. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which obtains the initial cluster centers and their coordinates in a preset manner, specifically includes the following steps: obtaining the number of initial cluster centers based on a preset target optimization rate; uniformly and randomly dispersing the initial cluster centers in the coordinate system, and obtaining the coordinates of the dispersed initial cluster centers. The number of initial cluster centers is set according to the optimization target that the algorithm needs to satisfy. Uniform and random dispersion is to more efficiently obtain similar characteristics among all piezoelectric coefficient samples, thereby facilitating clustering iteration, simplifying the steps, and making the results more reliable.

[0021] 3. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention requires that the initial cluster center and its coordinates be clustered according to certain conditions. These conditions include the following judgment process: determining whether the difference between the piezoelectric coefficient of the piezoelectric coefficient sample and the piezoelectric coefficient of the initial cluster center is within the clustering boundary value range. If it is outside the range, the next clustering iteration is waited for. If it is within the range, the magnitude of the judgment index corresponding to each cluster center is compared, and the minimum value among the judgment indices is obtained. The minimum value among the judgment indices is marked, and the initial cluster center corresponding to this minimum value is assigned to the new cluster in this iteration. By screening the piezoelectric coefficient samples, after multiple iterations, piezoelectric coefficient samples that meet the same criteria, i.e., have extremely high similarity, can be clustered together. The judgment index represents the error range; the smaller the judgment index, the smaller the error. Therefore, the minimum value of the judgment index, i.e., the most accurate point, is used as the new cluster center. Subsequent calculations can reduce computational errors, thereby ensuring the reliability and accuracy of the results.

[0022] 4. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which clusters the initial cluster center and its coordinates using a preset clustering method, specifically includes the following steps: obtaining the standard deviation of piezoelectric coefficient samples within a preset range and the initial cluster center, centered on the initial cluster center; and obtaining a set of piezoelectric coefficient samples that meet the standard deviation based on the initial cluster center and the piezoelectric coefficient samples, which serves as a new cluster. The initial cluster centers are uniformly distributed on the piezoelectric film, enabling faster acquisition of the standard deviation, i.e., the clustering basis, using this method. The set of piezoelectric coefficient samples that meet the standard deviation is then grouped into the same cluster, further improving clustering efficiency.

[0023] 5. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which uses a preset iterative method to converge cluster centers and obtain converged cluster centers and their coordinates, specifically includes the following steps: defining an initial cluster center coordinate as an initial sample; all initial samples form an initial sample set; clustering the initial sample set using a preset clustering method to obtain a new set of cluster center coordinates, which is primary data; defining multiple primary data sets to form ensemble data; using the ensemble data as new samples; selecting multiple new initial cluster centers from the new samples; and repeating the above clustering steps to obtain new cluster centers and their coordinates based on the selected new initial cluster centers, until the new initial cluster centers converge, thus obtaining the converged cluster centers and their coordinates. This method iterates multiple times to converge the cluster centers, thereby obtaining the optimal result while reducing computational requirements and improving detection efficiency.

[0024] 6. This invention also provides a high-throughput piezoelectric coefficient machine learning system for implementing the high-throughput piezoelectric coefficient detection machine learning method described above. The system includes an information acquisition module for acquiring piezoelectric coefficient samples at multiple locations in the piezoelectric film, machine learning input data, the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in a coordinate system, multiple initial cluster centers and their coordinates in the coordinate system, new cluster centers and their coordinates after clustering, converged cluster centers and their coordinates, and machine learning output data. A data processing module is also provided, which establishes a coordinate system based on the edge lines of the sample points, clusters the initial cluster centers and their coordinates using a preset clustering method, converges the cluster centers using a preset iteration method, and plans a preset detection probe movement path based on the converged cluster centers and their coordinates. The high-throughput piezoelectric coefficient machine learning system has the same beneficial effects as the aforementioned high-throughput piezoelectric coefficient detection machine learning method, and will not be elaborated upon here.

[0025] 7. The present invention also provides a storage medium, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the machine learning method for high-throughput piezoelectric coefficient detection as described above. The storage medium has the same beneficial effects as the machine learning method for high-throughput piezoelectric coefficient detection described above, and will not be elaborated further here. [Attached Image Description]

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. 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 effort.

[0027] Figure 1 This is a flowchart of the steps of a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0028] Figure 2 This is a flowchart of step S3 in a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0029] Figure 3 This is a graph showing the relationship between cluster centers and the number of sample points in a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0030] Figure 4 This is the step flow of step S4 in the machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention. Figure 1 .

[0031] Figure 5 This is the step flow of step S4 in the machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention. Figure 2 .

[0032] Figure 6 This is a schematic diagram of the cluster distribution in the coordinate system of a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0033] Figure 7 This is a flowchart of step S5 in a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0034] Figure 8 This is a schematic diagram of the movement path of the preset detection probe in a machine learning method for high-throughput piezoelectric coefficient detection provided in the first embodiment of the present invention.

[0035] Figure 9 This is a schematic diagram of a high-throughput piezoelectric coefficient machine learning system provided in the second embodiment of the present invention.

[0036] Figure 10 This is a schematic diagram of a storage medium provided in the third embodiment of the present invention.

[0037] Explanation of reference numerals in the attached diagram:

[0038] 1. A high-throughput machine learning system for piezoelectric coefficients; 2. Storage media;

[0039] 11. Information acquisition module; 12. Data processing module; 21. Memory; 22. Processor; 23. Computer program.

Detailed Implementation Methods

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0041] Please see Figure 1 The first embodiment of the present invention provides a machine learning method for high-throughput piezoelectric coefficient detection, comprising the following steps:

[0042] S1: Provide a piezoelectric thin film, obtain sample points at multiple locations in the piezoelectric thin film, generate piezoelectric coefficient samples corresponding to the sample points, and use the piezoelectric coefficient samples as input data for machine learning;

[0043] S2: Establish a coordinate system based on the edge of the sample point, and obtain the coordinates of the piezoelectric coefficient sample in the coordinate system and the piezoelectric coefficient value;

[0044] S3: Obtain multiple initial cluster centers and their coordinates in the coordinate system using a preset acquisition method;

[0045] S4: Based on the coordinates of the piezoelectric coefficient samples, the piezoelectric coefficient values, the initial cluster center and the coordinates of the initial cluster center, the clusters are divided into new cluster centers and their coordinates using a preset clustering method.

[0046] S5: Use a preset iteration method to converge the cluster center, and obtain the converged cluster center and its coordinates;

[0047] S6: Based on the convergence cluster center and the coordinates of the convergence cluster center, a preset detection probe moving path is planned, and this preset detection probe moving path is used as the output data of machine learning.

[0048] It should be noted that the machine learning method for high-throughput piezoelectric coefficient detection provided in the specific embodiments of the present invention is applicable to both the trained piezoelectric coefficient samples and the piezoelectric coefficient samples to be detected. The trained piezoelectric coefficient samples and the piezoelectric coefficient samples to be detected need to have the same distribution pattern on the piezoelectric film. Furthermore, the established coordinate system is a two-dimensional coordinate system, which is used to intuitively obtain the coordinates of each sample point, piezoelectric coefficient sample, initial cluster center, new cluster center, and convergent cluster center on the plane.

[0049] In this embodiment, a high-throughput D33 non-destructive testing (NDT) device is used to perform NDT on sample points. Specifically, "high-throughput" refers to the automated operation of the experimental process, enabling the testing of a large number of samples in a short time. The D33 coefficient of a piezoelectric material refers to the fact that when the MEMS piezoelectric vibration energy harvester operates in D33 mode, the stress is triaxial. When a piezoelectric material is subjected to tensile or compressive stress applied in a triaxial direction, the resulting mechanical deformation and electric field are both triaxial; that is, both voltage and stress are in a triaxial direction. The triaxial direction refers to the normal stress acting along the z-axis. The testing equipment is not limited here; in this invention, the high-throughput D33 NDT device is merely one of the preferred devices to achieve the expected results. The specific testing equipment should be determined based on actual conditions.

[0050] Furthermore, the cluster center, which is the mean value of the piezoelectric coefficient in the cluster, can be represented by an initial mean vector. This invention uses a modified mean algorithm based on given samples to divide the clusters obtained from clustering, reflecting the tightness of the samples within the cluster around the central piezoelectric coefficient detection point. It finds piezoelectric coefficient samples with extremely high similarity within the cluster and performs a clustering operation. Thus, it only needs to detect the initial cluster center point and ignores other piezoelectric coefficient samples with similar characteristics, thereby simplifying the detection process, reducing detection time and labor costs, and greatly improving detection efficiency.

[0051] For further details, please refer to Figure 2 and Figure 3 Obtaining the initial cluster center and its coordinates using a preset acquisition method specifically includes the following steps:

[0052] S31: Obtain the initial number of cluster centers based on the preset target optimization rate;

[0053] S32: Distribute the initial cluster centers uniformly and randomly in the coordinate system, and obtain the coordinates of the initial cluster centers after dispersion.

[0054] It should be noted that the initial cluster center is independent of the number of sample points, but is positively correlated with the area of ​​the piezoelectric film. The number of cluster center points is the number of sample points after algorithm optimization. Specifically, the initial cluster centers are uniformly distributed in two-dimensional space.

[0055] The number of initial cluster centers is related to the target optimization rate. Optionally, when the preset target optimization rate, i.e. the detection speed optimization, is 85%, if there are 100 test sample points, then the number of initial cluster centers is 15. After optimization by this algorithm, the specific coordinates of the 15 points that can best represent the whole can be known, achieving the most accurate detection effect.

[0056] Understandably, the initial number of cluster centers is set according to the optimization objective that the algorithm needs to satisfy. Uniform random dispersion is to obtain similar characteristics in all piezoelectric coefficient samples more efficiently, thereby facilitating clustering iteration, simplifying the steps, and making the results more reliable.

[0057] Specifically, the initial cluster center coordinates may be the same as or different from the piezoelectric coefficient sample coordinates.

[0058] Understandably, the piezoelectric coefficient samples can be regarded as the test sample points. The initial cluster centers are randomly obtained, so there may be duplicates. If the possibility of duplicates is eliminated, the reliability of the algorithm results will be reduced.

[0059] For more details, please see Figure 6 In the figure, different gray areas represent a cluster. A total of 15 clusters are shown in the figure. The area of ​​each cluster and the number of piezoelectric coefficient samples contained are not exactly the same.

[0060] Understandably, this method can further improve clustering efficiency.

[0061] Specifically, before clustering using the preset clustering method, it also includes setting clustering boundary values. The preset clustering method for clustering the initial cluster center and its coordinates is based on the following formula:

[0062]

[0063] Among them, each piezoelectric coefficient sample coordinates , Here is the piezoelectric coefficient of this piezoelectric coefficient sample, and the subscript k is the initial number of cluster centers. The coordinates of each initial cluster center are... , The piezoelectric coefficient value of the initial cluster center. To determine the indicators, In this context, 'a' represents the clustering boundary value.

[0064] For further details, please refer to Figure 4 Clustering of the initial cluster center and its coordinates requires satisfying clustering conditions, which include the following judgment process:

[0065] S411: Determine whether the difference between the piezoelectric coefficient of the sample and the piezoelectric coefficient of the initial cluster center is within the clustering boundary value range. If it is outside the clustering boundary value range, wait for the next clustering in the iteration process.

[0066] S412: If it is within the clustering boundary value range, compare the magnitude of the judgment index corresponding to each cluster center and obtain the minimum value among the judgment indices;

[0067] S42: Mark the minimum value in the judgment index and assign the initial cluster center corresponding to the minimum value to the new cluster in this iteration.

[0068] It should be noted that the judgment index represents the error range; the smaller the judgment index, the smaller the error.

[0069] Understandably, by screening piezoelectric coefficient samples and iterating multiple times, piezoelectric coefficient samples that meet the same criteria, i.e., have a high degree of similarity, can be grouped into the same cluster. Therefore, the minimum value of the judgment index, i.e. the most accurate point, can be used as the new cluster center for subsequent calculations, which can reduce calculation errors and thus ensure the reliability and accuracy of the results.

[0070] For further details, please refer to Figure 5 The specific steps for clustering the initial cluster center and its coordinates using a preset clustering method include the following:

[0071] S43: Obtain the standard deviation of the piezoelectric coefficient samples within a preset range and the initial cluster center, with the initial cluster center as the center;

[0072] S44: Based on the initial cluster center and piezoelectric coefficient samples, obtain a set of piezoelectric coefficient samples that meet the standard deviation, and use them as a new cluster.

[0073] It should be noted that the initial cluster centers are evenly distributed within the coordinate system, which allows for a faster acquisition of the standard deviation, i.e., the basis for clustering. The set of piezoelectric coefficient samples that meet the standard deviation is then grouped into the same cluster.

[0074] Specifically, in a specific embodiment of the present invention, the standard deviation of the piezoelectric coefficient samples within the circular region from the initial cluster center is obtained, with the initial cluster center as the center and the distance from the piezoelectric coefficient sample to the initial cluster center as the radius.

[0075] Furthermore, the process of clustering the initial cluster center and its coordinates using a preset clustering method also includes the following steps for determining the new cluster center:

[0076] Iterate through all piezoelectric coefficient samples in the new cluster, making

[0077] The point with the smallest value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples;

[0078] Alternatively, iterate through all piezoelectric coefficient samples in the new cluster and calculate the theoretical cluster center value. ,make The point closest to the theoretical cluster center value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples.

[0079] It should be noted that when using the formula to obtain the new cluster center, the two formulas cannot be used at the same time. That is, only one formula can be used at a time to obtain the new cluster center. This is to avoid mixing formulas, which would result in inconsistent standards for the obtained new cluster centers and affect the accuracy of the results.

[0080] For further details, please refer to Figure 7 and Figure 8 The process of convergent cluster centers and their coordinates using a preset iterative method includes the following steps:

[0081] S51: Define an initial cluster center coordinate as the initial sample. All initial samples form the initial sample set. The new cluster center coordinate set obtained by clustering the initial sample set in a preset clustering method is the primary data.

[0082] S52: Define multiple primary data sets as ensemble data, use the ensemble data as new samples, and select multiple new initial cluster centers from the new samples;

[0083] S53: Based on the selected new initial cluster center, repeat the above steps of clustering to obtain the new cluster center and the coordinates of the new cluster center until the new initial cluster center converges, and obtain the converged cluster center and the coordinates of the converged cluster center.

[0084] It should be noted that the method for selecting new initial cluster centers in new samples is also to uniformly and randomly disperse them in the coordinate system and obtain the coordinates. The detection probe movement path is the movement path of the device detection probe on the piezoelectric film. The movement path is related to the coordinates of the convergent cluster centers. The movement path obtained by this method is the most efficient detection path.

[0085] Specifically, the motion path of the detection probe is based on the established two-dimensional coordinate system. First, the vertical coordinate remains unchanged while the horizontal coordinate increases sequentially; then, the horizontal coordinate remains unchanged while the vertical coordinate increases sequentially; then, the vertical coordinate remains unchanged while the horizontal coordinate decreases sequentially; and finally, the horizontal coordinate remains unchanged while the vertical coordinate increases sequentially. This process forms a complete loop and serves as the basis for the detection probe's path planning. The movement distance is then adjusted based on the cluster center coordinates to form a motion path plan that adapts to the entire coordinate system, ensuring that all piezoelectric coefficient values, initial cluster centers, new cluster centers, and new initial cluster centers can be fully clustered.

[0086] Understandably, this method iterates multiple times to bring the cluster centers together, thereby obtaining the optimal result while reducing computational requirements and improving detection efficiency.

[0087] For further details, please refer to Figure 9 The second embodiment of the present invention provides a high-throughput piezoelectric coefficient machine learning system 1, used to implement the high-throughput piezoelectric coefficient detection machine learning method as described above, including an information acquisition module 11: used to acquire piezoelectric coefficient samples at multiple locations in the piezoelectric film, machine learning input data, the coordinates of the piezoelectric coefficient samples in the coordinate system and the piezoelectric coefficient values, multiple initial cluster centers and their coordinates in the coordinate system, new cluster centers and their coordinates after clustering, converged cluster centers and their coordinates, and machine learning output data; and a data processing module 12: used to establish a coordinate system based on the edge lines of the sample points, cluster the initial cluster centers and their coordinates using a preset clustering method, converge the cluster centers using a preset iteration method, and plan a preset detection probe movement path based on the converged cluster centers and their coordinates.

[0088] The machine learning system for high-throughput piezoelectric coefficient has the same beneficial effects as the machine learning method for high-throughput piezoelectric coefficient detection described above, and will not be elaborated here.

[0089] For further details, please refer to Figure 10 The third embodiment of the present invention provides a storage medium 2, including a memory 21, a processor 22, and a computer program 23 stored on the memory 21 and executable on the processor 22. When the processor 22 executes the computer program 23, it implements the machine learning method for high-throughput piezoelectric coefficient detection as described above. The storage medium 1 has the same beneficial effects as the machine learning method for high-throughput piezoelectric coefficient detection described above, and will not be described in detail here.

[0090] It is understood that, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. Computer-readable storage media include, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0091] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0092] In the embodiments provided by this invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0093] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Those skilled in the art should also recognize that the embodiments described in the specification are optional embodiments, and the actions and modules involved are not necessarily essential to the invention.

[0094] In various embodiments of the present invention, it should be understood that the sequence number of each process does not necessarily imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0095] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It is particularly important to note that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0096] Compared with existing technologies, the machine learning method, system, and computer medium for high-throughput piezoelectric coefficient detection of the present invention have the following advantages:

[0097] 1. The high-throughput piezoelectric coefficient detection machine learning method of the present invention includes the following steps: providing a piezoelectric thin film, acquiring sample points at multiple locations in the piezoelectric thin film, generating piezoelectric coefficient samples corresponding to the sample points, and using the piezoelectric coefficient samples as input data for machine learning; establishing a coordinate system based on the edge lines of the sample points, and acquiring the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in the coordinate system; acquiring multiple initial cluster centers and their coordinates in the coordinate system using a preset acquisition method; clustering the piezoelectric coefficient samples, their piezoelectric coefficient values, initial cluster centers, and their coordinates using a preset clustering method to obtain new cluster centers and their coordinates; converging the cluster centers using a preset iteration method to obtain converged cluster centers and their coordinates; and planning a preset detection probe movement path based on the converged cluster centers and their coordinates, using the preset detection probe movement path as output data for machine learning. The cluster center is the mean value of the piezoelectric coefficient in the cluster. The present invention uses a modified mean value algorithm to divide the clusters obtained by clustering based on the given samples. This reflects the tightness of the samples within the cluster around the central piezoelectric coefficient detection point. It finds the points with extremely high similarity among the samples within the cluster and performs a clustering operation. Thus, it only needs to detect the initial cluster center point and ignores other piezoelectric coefficient samples with similar characteristics, thereby simplifying the detection, reducing the detection time and labor costs, and greatly improving the detection efficiency.

[0098] 2. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which obtains the initial cluster centers and their coordinates in a preset manner, specifically includes the following steps: obtaining the number of initial cluster centers based on a preset target optimization rate; uniformly and randomly dispersing the initial cluster centers in the coordinate system, and obtaining the coordinates of the dispersed initial cluster centers. The number of initial cluster centers is set according to the optimization target that the algorithm needs to satisfy. Uniform and random dispersion is to more efficiently obtain similar characteristics among all piezoelectric coefficient samples, thereby facilitating clustering iteration, simplifying the steps, and making the results more reliable.

[0099] 3. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention requires that the initial cluster center and its coordinates be clustered according to certain conditions. These conditions include the following judgment process: determining whether the difference between the piezoelectric coefficient of the piezoelectric coefficient sample and the piezoelectric coefficient of the initial cluster center is within the clustering boundary value range. If it is outside the range, the next clustering iteration is waited for. If it is within the range, the magnitude of the judgment index corresponding to each cluster center is compared, and the minimum value among the judgment indices is obtained. The minimum value among the judgment indices is marked, and the initial cluster center corresponding to this minimum value is assigned to the new cluster in this iteration. By screening the piezoelectric coefficient samples, after multiple iterations, piezoelectric coefficient samples that meet the same criteria, i.e., have extremely high similarity, can be clustered together. The judgment index represents the error range; the smaller the judgment index, the smaller the error. Therefore, the minimum value of the judgment index, i.e., the most accurate point, is used as the new cluster center. Subsequent calculations can reduce computational errors, thereby ensuring the reliability and accuracy of the results.

[0100] 4. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which clusters the initial cluster center and its coordinates using a preset clustering method, specifically includes the following steps: obtaining the standard deviation of piezoelectric coefficient samples within a preset range and the initial cluster center, centered on the initial cluster center; and obtaining a set of piezoelectric coefficient samples that meet the standard deviation based on the initial cluster center and the piezoelectric coefficient samples, which serves as a new cluster. The initial cluster centers are uniformly distributed on the piezoelectric film, enabling faster acquisition of the standard deviation, i.e., the clustering basis, using this method. The set of piezoelectric coefficient samples that meet the standard deviation is then grouped into the same cluster, further improving clustering efficiency.

[0101] 5. The machine learning method for high-throughput piezoelectric coefficient detection of the present invention, which uses a preset iterative method to converge cluster centers and obtain converged cluster centers and their coordinates, specifically includes the following steps: defining an initial cluster center coordinate as an initial sample; all initial samples form an initial sample set; clustering the initial sample set using a preset clustering method to obtain a new set of cluster center coordinates, which is primary data; defining multiple primary data sets to form ensemble data; using the ensemble data as new samples; selecting multiple new initial cluster centers from the new samples; and repeating the above clustering steps to obtain new cluster centers and their coordinates based on the selected new initial cluster centers, until the new initial cluster centers converge, thus obtaining the converged cluster centers and their coordinates. This method iterates multiple times to converge the cluster centers, thereby obtaining the optimal result while reducing computational requirements and improving detection efficiency.

[0102] 6. This invention also provides a high-throughput piezoelectric coefficient machine learning system for implementing the high-throughput piezoelectric coefficient detection machine learning method described above. The system includes an information acquisition module for acquiring piezoelectric coefficient samples at multiple locations in the piezoelectric film, machine learning input data, the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in a coordinate system, multiple initial cluster centers and their coordinates in the coordinate system, new cluster centers and their coordinates after clustering, converged cluster centers and their coordinates, and machine learning output data. A data processing module is also provided, which establishes a coordinate system based on the edge lines of the sample points, clusters the initial cluster centers and their coordinates using a preset clustering method, converges the cluster centers using a preset iteration method, and plans a preset detection probe movement path based on the converged cluster centers and their coordinates. The high-throughput piezoelectric coefficient machine learning system has the same beneficial effects as the aforementioned high-throughput piezoelectric coefficient detection machine learning method, and will not be elaborated upon here.

[0103] 7. The present invention also provides a storage medium, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the machine learning method for high-throughput piezoelectric coefficient detection as described above. The storage medium has the same beneficial effects as the machine learning method for high-throughput piezoelectric coefficient detection described above, and will not be elaborated further here.

[0104] The above provides a detailed description of a machine learning method, system, and computer medium for high-throughput piezoelectric coefficient detection disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention. Any modifications, equivalent substitutions, and improvements made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A machine learning method for high-throughput piezoelectric coefficient detection, characterized in that: Includes the following steps: A piezoelectric thin film is provided, and sample points at multiple locations in the piezoelectric thin film are obtained. Piezoelectric coefficient samples are generated corresponding to the sample points, and the piezoelectric coefficient samples are used as input data for machine learning. A coordinate system is established based on the edge of the sample point to obtain the coordinates of the piezoelectric coefficient sample in the coordinate system and the piezoelectric coefficient value. Obtain multiple initial cluster centers and their coordinates in the coordinate system using a preset acquisition method; A clustering boundary is set. Based on the coordinates of the piezoelectric coefficient samples, the piezoelectric coefficient values, the initial cluster center, and the coordinates of the initial cluster center, clustering is performed using a preset clustering method. Specifically, it is determined whether the difference between the piezoelectric coefficient of the piezoelectric coefficient sample and the piezoelectric coefficient of the initial cluster center is within the clustering boundary. If it is outside the range, the next clustering iteration is waited for. If it is within the range, the magnitudes of the judgment indicators corresponding to each cluster center are compared, and the minimum value among the judgment indicators is obtained. The minimum value among the judgment indicators is marked, and the cluster center corresponding to this minimum value is assigned to the new cluster in this iteration, resulting in the new cluster center and its coordinates. The judgment indicator is... : Among them, each piezoelectric coefficient sample coordinates , Here is the piezoelectric coefficient of this piezoelectric coefficient sample, and the subscript k is the initial number of cluster centers. The coordinates of each initial cluster center are... , The piezoelectric coefficient value of the initial cluster center. In this context, 'a' represents the clustering boundary value. The cluster centers are converged using a preset iterative method, resulting in the converged cluster centers and their coordinates. The preset detection probe movement path is planned based on the convergence cluster center and the coordinates of the convergence cluster center, and this preset detection probe movement path is used as the output data of machine learning.

2. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: The initial cluster center coordinates may be the same as or different from the piezoelectric coefficient sample coordinates.

3. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: Obtaining the initial cluster center and its coordinates using a preset acquisition method specifically includes the following steps: The initial number of cluster centers is obtained based on the preset target optimization rate; The initial cluster centers are uniformly and randomly distributed in the coordinate system to obtain the coordinates of the initial cluster centers after dispersion.

4. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: The preset motion path of the detection probe is based on the established coordinate system. First, the vertical coordinate remains unchanged while the horizontal coordinate increases sequentially. Then, the horizontal coordinate remains unchanged while the vertical coordinate increases sequentially. Next, the vertical coordinate remains unchanged while the horizontal coordinate decreases sequentially. Then, the horizontal coordinate remains unchanged while the vertical coordinate increases sequentially. Finally, the movement distance is adjusted according to the cluster center coordinates to output data.

5. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: The specific steps for clustering the initial cluster center and its coordinates using a preset clustering method are as follows: The standard deviation of the piezoelectric coefficient samples within a preset range and the initial cluster center is obtained with the initial cluster center as the center. Based on the cluster center and piezoelectric coefficient samples, a set of piezoelectric coefficient samples that meet the standard deviation is obtained as a new cluster.

6. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: Clustering the initial cluster center and its coordinates using a preset clustering method also includes the following steps for determining the new cluster center: Iterate through all piezoelectric coefficient samples in the new cluster, making The point with the smallest value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples; Alternatively, iterate through all piezoelectric coefficient samples in the new cluster and calculate the theoretical cluster center value. ,make The point closest to the theoretical cluster center value is taken as the new cluster center, where m is the total number of piezoelectric coefficient samples.

7. The machine learning method for high-throughput piezoelectric coefficient detection as described in claim 1, characterized in that: To achieve cluster center convergence using a preset iteration method, and to obtain the converged cluster center and its coordinates, the specific steps include the following: Define an initial cluster center coordinate as the initial sample, and all initial samples form the initial sample set. The new cluster center coordinate set obtained by clustering the initial sample set according to the preset clustering method is the primary data. Multiple primary data sets are defined as ensemble data sets. The ensemble data sets are used as new samples, and multiple new initial cluster centers are selected from the new samples. Based on the selected new initial cluster center, repeat the above steps of clustering to obtain the new cluster center and its coordinates until the new initial cluster center converges, thus obtaining the converged cluster center and its coordinates.

8. A machine learning system for high-throughput piezoelectric coefficient detection, used to implement the machine learning method for high-throughput piezoelectric coefficient detection as described in any one of claims 1-7, characterized in that: It includes an information acquisition module: used to acquire piezoelectric coefficient samples at multiple locations in the piezoelectric film, machine learning input data, the coordinates and piezoelectric coefficient values ​​of the piezoelectric coefficient samples in the coordinate system, multiple initial cluster centers and their coordinates in the coordinate system, the new cluster centers and their coordinates after clustering, the converged cluster centers and their coordinates, and the machine learning output data. Data processing module: It is used to establish a coordinate system based on the edge of the sample point, to cluster the initial cluster center and the coordinates of the initial cluster center in a preset clustering method, to converge the cluster center in a preset iteration method, and to plan the preset detection probe movement path based on the converged cluster center and the coordinates of the converged cluster center.

9. A storage medium comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the machine learning method for high-throughput piezoelectric coefficient detection as described in any one of claims 1-7.

Images