A method for constructing a device residual life prediction model and a terminal device

Abstract

The application is suitable for the technical field of data processing, and provides a construction method and device of a device residual life prediction model, comprising: obtaining device use sample data, and performing symbolic regression and horizontal federated learning according to the device use sample data to obtain representative feature data corresponding to the device use sample data; further, the device residual life prediction model can be constructed according to the representative feature data. In this way, in the scenario of symbolic regression, the feature engineering scheme of the device use sample data of symbolic regression is proposed in the horizontal federated learning scenario, different nonlinear use feature data is constructed on the basis of combining the use feature data of different device use sample data, the diversity of the final use feature data is increased, the final model effect is improved from the perspective of deeper mining of the use feature data, and therefore the accuracy of the output of the device residual life prediction model is improved, that is, the accuracy of the device residual life prediction is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of data processing, and particularly relates to a construction method of a device residual life prediction model and a terminal device. BACKGROUND

[0002] With the rise of intelligent devices, people can not only use devices for production, but also use devices for office work and entertainment. However, the habits and frequency of using devices by each person are different, which means that the wear and tear of devices is not always the same. Therefore, the estimation of the residual service life of the traditional device is not accurate enough.

[0003] Based on the above deficiencies, it is necessary to provide a scheme for accurately calculating the residual life of a device to improve the user's satisfaction. SUMMARY

[0004] Therefore, the embodiments of the present application provide a construction method of a device residual life prediction model and a terminal device to solve the problem of how to improve the accuracy of device residual life prediction in the prior art.

[0005] The first aspect of the embodiments of the present application provides a construction method of a device residual life prediction model, comprising:

[0006] obtaining device usage sample data;

[0007] performing symbolic regression and horizontal federated learning according to the device usage sample data to obtain representative feature data corresponding to the device usage sample data;

[0008] constructing a device residual life prediction model according to the representative feature data.

[0009] Optionally, the device usage sample data comprises usage feature data and a label corresponding to the usage feature data; and the performing symbolic regression and horizontal federated learning according to the device usage sample data to obtain representative feature data corresponding to the device usage sample data comprises:

[0010] performing data augmentation processing on the usage feature data to obtain a plurality of augmented usage feature data corresponding to the usage feature data;

[0011] determining representative feature data according to the plurality of augmented usage feature data corresponding to the usage feature data; wherein the representative feature data is the augmented usage feature data with the highest fitness to the label corresponding to the usage feature data among the plurality of augmented usage feature data corresponding to the usage feature data;

[0012] determining whether an iteration condition is met; wherein the iteration condition is that the number of iteration and selection of usage feature data meets a preset threshold.

[0013] If the iteration condition is not met, the representative feature data is taken as the use feature data, and the step of performing data augmentation processing on the use feature data to obtain a plurality of augmented use feature data corresponding to the use feature data is re-executed until the iteration condition is met.

[0014] Optionally, the data augmentation processing is random deformation combination processing.

[0015] Optionally, the determining of the representative feature data according to the plurality of augmented use feature data corresponding to the use feature data comprises:

[0016] calculating fitness of each of the plurality of augmented use feature data with respect to the label corresponding to the use feature data;

[0017] determining, among the fitness of each of the plurality of augmented use feature data, the augmented use feature data with the highest fitness, and taking the use feature data with the highest fitness as the representative feature data.

[0018] Optionally, the use feature data comprises device index data, device use information, and device fault information; and the label corresponding to the use feature data is device remaining life.

[0019] A second aspect of the embodiment of the application provides a device remaining life prediction model construction device, comprising:

[0020] an acquisition unit configured to acquire device use sample data;

[0021] a determination unit configured to perform symbolic regression and horizontal federated learning according to the device use sample data to obtain representative feature data corresponding to the device use sample data;

[0022] a construction unit configured to construct a device remaining life prediction model according to the representative feature data.

[0023] Optionally, the device use sample data comprises use feature data and a label corresponding to the use feature data; and the determination unit is specifically configured to:

[0024] perform data augmentation processing on the use feature data to obtain a plurality of augmented use feature data corresponding to the use feature data;

[0025] determine representative feature data according to the use feature data and the plurality of augmented use feature data corresponding to the use feature data; wherein the representative feature data is the use feature data with the highest fitness with respect to the label corresponding to the use feature data among the use feature data and the plurality of augmented use feature data corresponding to the use feature data.

[0026] determining whether an iteration condition is satisfied, wherein the iteration condition is that a preset threshold is satisfied by a number of times of iterative screening using the feature data;

[0027] If the iteration condition is not satisfied, the representative feature data is used as the feature data for use, and the step of performing data augmentation on the feature data for use is re-executed until the iteration condition is satisfied.

[0028] Optionally, the determining unit is further configured to:

[0029] calculate fitness of the feature data for use and the label corresponding to the feature data for use;

[0030] calculate fitness of each of the plurality of augmented feature data and the label corresponding to the feature data for use;

[0031] determine the feature data for use with the highest fitness from among the fitness of the feature data for use and the fitness of each of the plurality of augmented feature data, and use the feature data for use with the highest fitness as the representative feature data.

[0032] A third aspect of the embodiment of the present application provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method according to any one of the first aspect when executing the computer program.

[0033] A fourth aspect of the embodiment of the present application provides a computer readable storage medium, and the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the method according to any one of the first aspect.

[0034] Compared with the prior art, the embodiment of the present application has the beneficial effect that the method for constructing a device residual life prediction model provided by the present application obtains device use sample data, performs symbolic regression and horizontal federated learning according to the device use sample data, obtains representative feature data corresponding to the device use sample data, and further constructs a device residual life prediction model according to the representative feature data. In this way, the present application proposes a feature engineering scheme for the device use sample data of symbolic regression in the scene of horizontal federated learning based on symbolic regression, constructs different nonlinear use feature data based on the use feature data of different device use sample data, increases the diversity of the final use feature data, improves the final model effect from the perspective of mining use feature data at a deeper level, and thus improves the accuracy of the output of the device residual life prediction model, that is, improves the accuracy of the device residual life prediction. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0036] Figure 1 is the implementation flowchart of the method for constructing a device residual life prediction model provided by the embodiment of the present application;

[0037] Figure 2 is the schematic diagram of the construction device of the device residual life prediction model provided by the embodiment of the present application;

[0038] Figure 3 is the schematic diagram of the construction terminal device of the device residual life prediction model provided by the embodiment of the present application. DETAILED DESCRIPTION

[0039] In the following description, specific details are set forth in order to provide a thorough understanding of the embodiments of the present application, but these specific details do not limit the present application. It should be apparent to those skilled in the art that other embodiments of the present application can be implemented without these specific details. In other cases, well-known systems, devices, circuits and methods have not been described in detail in order not to obscure the description of the present application with unnecessary detail.

[0040] In order to illustrate the technical solutions of the present application, the following will be described through specific embodiments.

[0041] Reference is made to Figure 1, a method for constructing a device remaining life prediction model is shown, wherein the method can be completely applied to a terminal device (such as a mobile device, a notebook, an electronic communication watch, etc.), or can be completely applied to a server, or can be partially applied to a terminal device and partially applied to a server. In this embodiment, the method may, for example, include the following steps:

[0042] S101: Obtain device usage sample data.

[0043] In an implementation manner, the device usage sample data may include usage feature data and a label corresponding to the usage feature data. It should be noted that the device usage sample data may include multiple groups of usage feature data and a label corresponding to each group of usage feature data. The usage feature data includes device index data (such as device name, device model, device manufacturer, device part model, etc.), device usage information (such as device usage time length, device usage frequency, device usage environment, etc.), device fault information (such as the number of faults of the device within a preset time length, the number of maintenance times, etc.); and the label corresponding to the usage feature data is the device remaining life.

[0044] S102: Perform symbolic regression and horizontal federated learning according to the device usage sample data to obtain representative feature data corresponding to the device usage sample data.

[0045] Symbolic regression is a supervised learning method that attempts to discover a hidden mathematical formula to predict the target variable using the feature variable. The advantage of symbolic regression is that it can establish a symbolic model for a nonlinear system without relying on prior knowledge or models.

[0046] Horizontal federated learning: also known as feature-aligned federated learning, that is, the data features of participants in horizontal federated learning are aligned; this method is suitable for a case where the data features of participants overlap more and the sample IDs overlap less, for example, customer data of two different regions; wherein, the word "horizontal" comes from "horizontal partitioning (a.k.a. sharding)" of data; this method jointly performs federated learning on multiple rows of samples with the same features of multiple participants, that is, the training data of each participant is horizontally partitioned, which is called horizontal federated learning, and horizontal federated learning increases the total number of training samples.

[0047] In the embodiment, the use feature data can be subjected to data augmentation processing first to obtain a plurality of augmented use feature data corresponding to the use feature data (for example, 3000 augmented use feature data can be augmented). It should be noted that when there are multiple groups of use feature data, one or more groups of use feature data can be extracted for data augmentation processing. In an implementation manner, the data augmentation processing is random deformation combination processing. For example, the use feature data can be subjected to variation (i.e., random deformation combination). For example, when a group of use feature data is "A manufacturer + use for one year + failure twice", A manufacturer / use for one year / failure twice can be combined, or (A manufacturer + model) * use for one year / equipment working environment.

[0048] Then, the representative feature data can be determined according to the plurality of augmented use feature data corresponding to the use feature data. The representative feature data is the augmented use feature data with the highest fitness to the label corresponding to the use feature data among the plurality of augmented use feature data corresponding to the use feature data. Specifically, the fitness of each of the plurality of augmented use feature data to the label corresponding to the use feature data is calculated. For example, the correlation coefficient of the augmented use feature data to the label corresponding to the use feature data can be calculated, the IV value can be calculated in the case of binary classification of the label (Y value), that is, the prediction ability of the variation of the augmented use feature data for the Y value is evaluated); then, the augmented use feature data with the highest fitness is determined from the fitnesses respectively corresponding to the plurality of augmented use feature data, and the augmented use feature data with the highest fitness is taken as the representative feature data.

[0049] It is determined whether the iteration condition is met. The iteration condition is that the iteration screening number of the use feature data meets a preset threshold. For example, the preset threshold can be 300. It should be noted that the preset threshold can be set according to actual needs.

[0050] If the iteration condition is not met, the representative feature data is taken as the use feature data, the step of obtaining the plurality of augmented use feature data corresponding to the use feature data by subjecting the use feature data to data augmentation processing is re-executed until the iteration condition is met. In this way, the winner (i.e., the representative feature data with strong prediction ability for the Y value enters the next round of inheritance); the entrants in the next round (the winners in the last round: the representative feature data) are subjected to cross variation or self variation (the specific variation logic can be referred to the variation logic of symbolic regression), and then the fitness of the Y value is judged, and then the inheritance and variation are performed again. The total number of variation rounds (i.e., the preset threshold) is set in advance. After the preset threshold number of rounds, the last winner (i.e., the augmented use feature data) is output as the output representative feature data.

[0051] S103: constructing a device remaining life prediction model according to the representative feature data.

[0052] After obtaining the representative feature data, that is, for the representative feature data with strong prediction ability for the label corresponding to the use feature data (that is, the device remaining life), a device remaining life prediction model can be constructed according to the representative feature data. For example, the device remaining life prediction model can be constructed according to the representative feature data based on logistic regression xgboost. That is, the device remaining life prediction model can be constructed according to the relationship between the representative feature data and the label corresponding to the representative feature data (that is, the label of the use feature data corresponding to the representative feature data). In this way, after receiving a use feature data to be predicted, the use feature data to be predicted can be predicted according to the constructed device remaining life prediction model to obtain the label corresponding to the use feature data to be predicted, that is, the device remaining life.

[0053] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution. The execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0054] Compared with the prior art, the embodiment of the present application has the beneficial effect that the device remaining life prediction model construction method provided by the present application obtains device use sample data, and performs symbolic regression and horizontal federated learning according to the device use sample data to obtain representative feature data corresponding to the device use sample data. Further, a device remaining life prediction model can be constructed according to the representative feature data. In this way, the present application proposes a feature engineering scheme for symbolic regression of device use sample data in the horizontal federated learning scene based on the symbolic regression scene. Different nonlinear use feature data is constructed based on the use feature data of different device use sample data, the diversity of the final use feature data is increased, the model effect is improved from a deeper level by mining the use feature data from a more deep level, and the accuracy of the device remaining life prediction model output is improved, that is, the accuracy of the device remaining life prediction is improved.

[0055] Corresponding to the device remaining life prediction model construction method described above, the embodiment of the present application provides a device remaining life prediction model construction device, as shown in Figure 2 The device remaining life prediction model construction device comprises:

[0056] The acquisition unit 201 is configured to acquire device use sample data.

[0057] The determination unit 202 is configured to perform symbolic regression and horizontal federated learning according to the device use sample data to obtain representative feature data corresponding to the device use sample data.

[0058] A constructing unit 203 is configured to construct a device residual life prediction model according to the representative feature data.

[0059] Optionally, the device usage sample data comprises usage feature data and a label corresponding to the usage feature data; and the determining unit 202 is specifically configured to:

[0060] perform data augmentation processing on the usage feature data to obtain a plurality of augmented usage feature data corresponding to the usage feature data;

[0061] determine representative feature data according to the usage feature data and the plurality of augmented usage feature data corresponding to the usage feature data; wherein the representative feature data is usage feature data with the highest fitness to the label corresponding to the usage feature data among the usage feature data and the plurality of augmented usage feature data corresponding to the usage feature data.

[0062] determine whether an iteration condition is met; wherein the iteration condition is that the number of iteration and screening of usage feature data meets a preset threshold;

[0063] if the iteration condition is not met, the representative feature data is taken as usage feature data, and the step of performing data augmentation processing on the usage feature data to obtain a plurality of augmented usage feature data corresponding to the usage feature data is re-executed until the iteration condition is met.

[0064] Optionally, the determining unit 202 is further specifically configured to:

[0065] calculate the fitness of the usage feature data to the label corresponding to the usage feature data;

[0066] calculate the fitness of each of the plurality of augmented usage feature data to the label corresponding to the usage feature data;

[0067] determine usage feature data with the highest fitness among the fitness of the usage feature data, the fitness of each of the plurality of augmented usage feature data, and take the usage feature data with the highest fitness as representative feature data.

[0068] Optionally, the data augmentation processing is random deformation and combination processing.

[0069] Optionally, the usage feature data comprises device index data, device usage information and device fault information; and the label corresponding to the usage feature data is device residual life.

[0070] Figure 3 is a schematic diagram of a device residual life prediction model construction device / terminal device provided by an embodiment of the present application. As shown inFigure 3 As shown in FIG. 1, the device residual life prediction model construction apparatus / terminal device 3 of this embodiment comprises a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30, for example, a device residual life prediction model construction program. The processor 30 implements the steps in each of the above device residual life prediction model construction method embodiments when executing the computer program 32, for example Figure 1 The processor 30 implements the functions of each module / unit in each of the above apparatus embodiments when executing the computer program 32, for example Figure 2 The functions of the modules 201 to 203.

[0071] For example, the computer program 32 can be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 30 to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 32 in the device residual life prediction model construction apparatus / terminal device 3. For example, the computer program 32 can be divided into a synchronization module, a summary module, an acquisition module, and a return module (modules in virtual apparatus), and the specific functions of each module are as follows:

[0072] The device residual life prediction model construction apparatus / terminal device 3 can be a desktop computer, a notebook, a palm computer, a cloud server, and other computing devices. The device residual life prediction model construction apparatus / terminal device can include, but is not limited to, a processor 30, a memory 31. Those skilled in the art can understand that Figure 3 The device residual life prediction model construction apparatus / terminal device 3 is only an example and does not constitute a limitation on the device residual life prediction model construction apparatus / terminal device 3, which can include more or fewer components than those shown, or combine certain components, or different components, for example, the device residual life prediction model construction apparatus / terminal device can also include an input / output device, a network access device, a bus, etc.

[0073] The processor 30 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0074] The memory 31 can be an internal storage unit of the device remaining life prediction model construction apparatus / terminal device 3, such as a hard disk or a memory of the device remaining life prediction model construction apparatus / terminal device 3. The memory 31 can also be an external storage device of the device remaining life prediction model construction apparatus / terminal device 3, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 31 can include both an internal storage unit and an external storage device of the device remaining life prediction model construction apparatus / terminal device 3. The memory 31 is used to store the computer program and other programs and data required by the device remaining life prediction model construction apparatus / terminal device. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0075] It can be clearly understood by those skilled in the art that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the above-described functions. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit, and the integrated unit can be realized in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction, and are not used to limit the protection scope of the present application. The specific working process of the units and modules in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0076] In the above embodiments, the description of each embodiment is focused on, and the part not described or recorded in a certain embodiment can be referred to the relevant description of other embodiments.

[0077] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0078] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / terminal device and method can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely schematic, for example, the division of the modules or units is merely a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0079] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0080] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0081] The integrated module / unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0082] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for constructing a prediction model for the remaining useful life of equipment, characterized in that, include: Acquire device usage sample data, which includes multiple sets of usage feature data and a label corresponding to each set of usage feature data, wherein the multiple sets of usage feature data are data corresponding to different devices; Based on the device usage sample data, symbolic regression and lateral federated learning are performed to obtain representative feature data corresponding to the device usage sample data; Based on the representative feature data, construct a prediction model for the remaining lifespan of the equipment; The step of performing symbolic regression and lateral federated learning based on the device usage sample data to obtain representative feature data corresponding to the device usage sample data includes: The usage feature data is subjected to data amplification processing to obtain several amplified usage feature data corresponding to the usage feature data; Based on the several amplification usage feature data corresponding to the usage feature data, representative feature data is determined; wherein, the representative feature data is the amplification usage feature data with the highest fitness to the tag corresponding to the usage feature data among the several amplification usage feature data corresponding to the usage feature data. Determine whether the iteration condition is met; wherein, the iteration condition is that the number of iterations using feature data meets a preset threshold. If the iteration condition is not met, the representative feature data is used as the usage feature data, and the step of performing data augmentation processing on the usage feature data to obtain several augmented usage feature data corresponding to the usage feature data is repeated until the iteration condition is met.

2. The method for constructing a prediction model for remaining equipment life as described in claim 1, characterized in that, The data augmentation process is a randomized combination process.

3. The method for constructing a prediction model for the remaining useful life of equipment as described in claim 2, characterized in that, The step of determining representative feature data based on several amplified usage feature data corresponding to the usage feature data includes: Calculate the fitness of each of the amplification feature data and the tag corresponding to the feature data; Among the fitness values ​​corresponding to the various amplification and usage feature data, the amplification and usage feature data with the highest fitness is determined, and the usage feature data with the highest fitness is used as the representative feature data.

4. A method for constructing a device remaining life prediction model as described in any one of claims 1-3, characterized in that, The usage characteristic data includes equipment indicator data, equipment usage information, and equipment fault information; the tag corresponding to the usage characteristic data is the remaining lifespan of the equipment.

5. An apparatus for constructing a prediction model for the remaining useful life of equipment, characterized in that, include: The acquisition unit is used to acquire device usage sample data, which includes multiple sets of usage feature data and a label corresponding to each set of usage feature data. The multiple sets of usage feature data are data corresponding to different devices. A determining unit is configured to perform symbolic regression and lateral federated learning on the device usage sample data to obtain representative feature data corresponding to the device usage sample data. Specifically, the determining unit is configured to: perform data augmentation on the usage feature data to obtain several augmented usage feature data corresponding to the usage feature data; determine representative feature data based on the usage feature data and the several augmented usage feature data corresponding to the usage feature data; wherein the representative feature data is the usage feature data with the highest fitness among the usage feature data and the several augmented usage feature data corresponding to the usage feature data and the label corresponding to the usage feature data; determine whether an iteration condition is met; wherein the iteration condition is that the number of iterations of the usage feature data meets a preset threshold; if the iteration condition is not met, then the representative feature data is used as the usage feature data, and the step of performing data augmentation on the usage feature data to obtain several augmented usage feature data corresponding to the usage feature data is re-executed until the iteration condition is met. The construction unit is used to construct a device remaining life prediction model based on the representative feature data.

6. The apparatus for constructing a prediction model for the remaining life of equipment as described in claim 5, characterized in that, The determining unit is further specifically used for: Calculate the fitness of the usage feature data with the corresponding labels; Calculate the fitness of each of the amplification feature data and the tag corresponding to the feature data; Among the fitness corresponding to the used feature data and the fitness corresponding to each of the several amplified used feature data, the used feature data with the highest fitness is determined, and the used feature data with the highest fitness is used as the representative feature data.

7. A terminal device, 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 steps of the method as described in any one of claims 1 to 4.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 4.

Images