A method and apparatus for endurance testing of flash memory chips

Abstract

The application discloses a kind of durability test method and device for flash memory chip, it is related to semiconductor testing technical field.The present application includes information acquisition: the storage type of flash memory chip is obtained and test environment requirement obtains test requirement.The present application ensures the diversity and non-repeatability of test data by establishing real-time updating database, dynamically sets test data type weight and environment parameter in combination with the storage type and working environment of flash memory chip, reflects that test can be dynamically adjusted, realizes accurate durability test, in the process of testing, by cross validation, real-time monitoring chip state and assisting positioning fault point, simultaneously, collect the associated data of state and durability, construct prediction model, to significantly improve test efficiency and accuracy, can be evaluated in advance chip life by model prediction, optimize product design;Auxiliary fault positioning function accelerates defect analysis, reduces research and development cost to provide systematic support for reliability verification and performance improvement of flash memory chip.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of semiconductor testing, and particularly relates to a durability test method and device for a flash memory chip. BACKGROUND

[0002] The flash memory chip is a kind of non-volatile memory capable of storing data in the case of power failure. It is widely used in various electronic devices such as smart phones, solid state disks, USB flash disks, memory cards, etc., for storing operating systems, application programs, personal data and other information. The durability test of the flash memory chip is part of the semiconductor test, which mainly evaluates the performance and reliability of the memory under repeated writing / erasing, environmental changes and other conditions.

[0003] The patent with the publication number CN120236649A discloses a test method, device, system, electronic equipment and storage medium for a flash memory chip, which is applied to an electronic equipment including a flash memory chip. The method comprises the following steps: determining a first test mode of the flash memory chip; obtaining reference test environment parameters and reference test standard parameters corresponding to the first test mode; determining first test environment parameters of the flash memory chip; determining a first deviation degree between the reference test environment parameters and the first test environment parameters; adjusting the reference test standard parameters according to the first deviation degree to obtain target test standard parameters; and testing the flash memory chip according to the target test standard parameters and the first test mode. The application can improve the intelligence of the flash memory chip test and further improve the performance of the flash memory chip.

[0004] At present, most of the flash memory durability tests adopt single or static environmental conditions, which are difficult to reflect the variable actual working scene, leading to deviation of the test from the real use environment. In addition, the traditional test lacks dynamic data updating and real-time monitoring function, which is difficult to reflect the performance degradation trend of the chip in the complex environmental change, and it is also difficult to carry out efficient multi-factor environmental combined action analysis, thereby limiting the comprehensive evaluation of the actual reliability of the chip. Therefore, the present application is proposed. SUMMARY

[0005] The present application aims to provide a durability test method and device for a flash memory chip to solve the problems raised in the background.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a durability test method for a flash memory chip, the method comprising:

[0007] Information acquisition: obtaining the storage type of the flash memory chip and the test environment requirement to obtain the test requirement, the test requirement including the type requirement and the environment requirement;

[0008] characterized in that it comprises:

[0009] Database establishment: establishing a real-time updated test data storage through a set database establishment method to obtain a database;

[0010] Environment setting and determination: setting a test environment according to the working scene of the flash memory chip through an environment processing method to obtain target environment parameters;

[0011] Type setting and determination: setting a test data weight matching type requirement according to the storage type of the flash memory chip through a data processing method to obtain weight information;

[0012] Test performance: performing a durability test on the flash memory chip based on the target environment parameters, the weight information and the database to obtain a durability result;

[0013] Cross-validation: verifying the working state of the flash memory chip through a set cross-validation method to obtain a verification result during the durability test of the flash memory chip, and obtaining the relationship between the real-time state and the durability of the flash memory chip through a set information acquisition method to obtain reflection information;

[0014] Durability prediction: establishing a prediction directory for predicting the durability based on the reflection information through a model establishment method, obtaining a to-be-predicted running state of the flash memory chip, extracting a durability prediction model of the flash memory chip in the prediction directory, and obtaining a durability prediction result based on the to-be-predicted running state through the durability prediction model.

[0015] Further, the database establishment method comprises: presetting a data type, obtaining basic data based on the real-time acquisition of data information, splitting the basic data to obtain a plurality of sub-data, splitting the sub-data to obtain a plurality of data segments, reducing and increasing the sub-data based on the data segments to obtain derivative data, integrating the basic data and the derivative data to obtain overall data, establishing the association between the overall data and the data type, establishing a data sub-library with the data type as the name for storing the overall data, and integrating all the data sub-libraries to obtain the database.

[0016] Further, the environment processing method comprises: obtaining a basic scene of the flash memory chip, the basic scene comprising a plurality of sub-scenes, the sub-scenes comprising a plurality of factor items, presetting a detection time period and a fluctuation factor, generating a tuning node based on the detection time period, adding the fluctuation factor to the factor items in the sub-scenes based on the tuning node to obtain result factor parameters, integrating the result factor parameters to obtain a result scene, establishing a scene library for storing the result scene, and obtaining the target environment parameters based on the environment requirement and the scene library by selecting the result scene corresponding to the environment requirement in the scene library.

[0017] Further, the data processing method comprises: obtaining a target storage type based on a type requirement and a storage type of the flash memory chip, splitting the target storage type to obtain a plurality of sub-storage types, obtaining a type quantity based on the plurality of sub-storage types, sorting the sub-storage types according to storage importance based on the type requirement to obtain sorting information, presetting a weight distribution, the weight distribution comprising a basic weight and a decreasing factor, assigning a weight to the sub-storage types based on the sorting information, the basic weight and the decreasing factor to obtain a target weight, and integrating all the sub-storage types and the target weight to obtain weight information.

[0018] Further, the cross-validation method comprises: presetting fixed data and a detection node, the fixed data being write-in data, writing the write-in data into the flash memory chip based on the detection node and writing out to obtain write-out data, judging whether the write-in data and the write-out data are consistent to obtain a judgment result, obtaining a first result when the judgment result feedback is that the write-in data and the write-out data are inconsistent, extracting the part of the write-in data and the write-out data that are inconsistent to obtain difference information, the first result feedback being that the working state of the flash memory chip is abnormal, obtaining a second result when the judgment result feedback is that the write-in data and the write-out data are consistent, the second result feedback being that the working state of the flash memory chip is normal, and integrating the difference information, the first result and the second result to obtain a verification result.

[0019] Further, the information acquisition method comprises: presetting a monitoring node, recording a test time in a process of performing durability test on the flash memory chip, recording a real-time state of the flash memory chip to obtain a reference state when the test time reaches the monitoring node, recording a data write-in and write-out times of the flash memory chip to obtain a preliminary result, combining the durability result with the preliminary result to obtain a subsequent durability result of the reference state, establishing an association relationship between the reference state and the subsequent durability result, and integrating the reference state, the subsequent durability result and the association relationship to obtain reflection information.

[0020] Further, the model establishment method comprises:

[0021] Training data processing: obtaining a plurality of result information based on type classification reflection information of the flash memory chip, splitting the result information to obtain the reference state, the subsequent durability result and the association relationship, extracting the reference state and the subsequent durability result with the association relationship to obtain to-be-processed data, pre-processing the to-be-processed data to obtain sub-training data, and integrating the sub-training data under the same result information to obtain training data;

[0022] Model selection and training: selecting a deep learning model as a model base, importing the training data into the model base for learning by the model base to obtain an initial model;

[0023] Model adjustment and output: preset verification state and verification result, import verification state into initial model to obtain prediction result, compare prediction result with verification result to obtain comparison result, optimize initial model based on comparison result to obtain durability prediction model;

[0024] Model correlation establishment: establish the correlation between the flash memory chip and the durability prediction model, and establish a prediction directory for storing the flash memory chip, the durability prediction model and the correlation.

[0025] A durability testing device for a flash memory chip uses the durability testing method for a flash memory chip described above.

[0026] Compared with the prior art, the beneficial effects of the present application are:

[0027] The durability testing method and device for a flash memory chip ensure the diversity and non-repeatability of test data by establishing a real-time updated database, dynamically set test data type weight and environment parameters according to the storage type and working environment of the flash memory chip, embodying dynamic adjustment of the test, and realizing accurate durability testing. In the testing process, the state of the chip is monitored in real time and the fault point is located through cross-validation, and the correlation data of the state and the durability is collected at the same time, and a prediction model is constructed, so as to significantly improve the test efficiency and accuracy. The service life of the chip can be evaluated in advance through model prediction, and the product design is optimized; the auxiliary fault positioning function accelerates the defect analysis and reduces the research and development cost, providing systematic support for the reliability verification and performance improvement of the flash memory chip.

[0028] At the same time, in the durability testing of the flash memory chip, the write data is allocated weight according to the storage data type (such as high-frequency write log class data and low-frequency write static file), which can more realistically simulate the actual load scene and quickly expose the wear bottleneck of the chip, while optimizing the test resources. This differentiated strategy not only improves the test efficiency (shortens the cycle and reduces redundancy), but also evaluates the reliability difference of different storage units (such as SLC / TLC blocks) and provides verification basis for wear leveling algorithms, and finally outputs results more consistent with the actual application life.

[0029] At the same time, the working state of the flash memory chip is verified through the cross-validation method, and it is determined whether to end the durability testing of the flash memory chip, and the fault position of the flash memory chip can be indirectly determined through the feedback of the verification result, so as to improve the efficiency of optimizing the flash memory chip. Different types of flash memory chips correspond to different durability prediction models, so as to reduce the difficulty of model training and improve the efficiency of using the durability prediction model subsequently. Through the correlation set, the corresponding durability prediction model corresponding to the flash memory chip can be obtained according to the type of the flash memory chip. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 A schematic diagram of the overall process of the present application;

[0031] Figure 2 A schematic diagram of the derived data structure of the present application;

[0032] Figure 3 A schematic diagram of the factor item parameter adjustment of the present application;

[0033] Figure 4 A schematic diagram of the weight information of the present application;

[0034] Figure 5 A schematic diagram of the device structure of the present application. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0036] Flash memory chips use charge storage technology and have certain service life limitations. The main reasons include:

[0037] Limited number of write / erase cycles: Each storage cell has a limited number of cycles (usually several thousand to several ten thousand times) for charge storage, and after exceeding it, the storage cell may fail; data retention capability: After a long time of storage, the storage cell may leak charge, causing data distortion or loss; performance degradation: After a large number of cycles, the read / write speed may slow down and the error rate may increase; environmental impact: external environments such as high temperature, high humidity, and vibration may affect the reliability of the chip; meet quality and standard requirements: ensure the reliability, safety, and stability of the chip in actual use to avoid faults and data loss; durability testing is to verify the flash memory chip under reasonable stress and environmental conditions before it is put into the market, to confirm that it can work stably within the predetermined service life, and to ensure the safety of user data and the reliability of the device.

[0038] As shown in Figures 1-4 The present application provides a technical solution: a durability testing method for a flash memory chip, the method comprising:

[0039] Information acquisition: acquire the storage type of the flash memory chip and the test environment requirements to obtain test requirements, the test requirements including type requirements and environment requirements;

[0040] It should be noted that the type requirement is the type of data stored by the flash memory chip, and the environment requirement is the working environment of the data stored by the flash memory chip. In the specific implementation process, the type requirement and the environment requirement are obtained by the staff testing the flash memory chip.

[0041] It is characterized by comprising:

[0042] Database establishment: a real-time updated test data storage library is established by a set database establishment method to obtain a database;

[0043] It should be noted that a real-time updated database is established by a set database establishment method to ensure that the data used for testing is not repetitive, thereby improving the accuracy of the durability test of the flash memory chip.

[0044] Environment setting and determination: the target environment parameters are obtained by setting the test environment according to the working scene of the flash memory chip through an environment processing method;

[0045] It should be noted that the specific environment for testing is set according to the working scene of the flash memory chip through a set environment processing method, and the target environment parameters are obtained in cooperation with the environment requirement, so as to facilitate subsequent durability testing.

[0046] Type setting and determination: the weight information is obtained by setting the test data weight according to the storage type of the flash memory chip through a data processing method in cooperation with the type requirement;

[0047] It should be noted that the weight information of the test data is obtained by setting the weight of the test data according to the storage type of the flash memory chip through a set data processing method in cooperation with the type requirement, so as to more accurately obtain the durability of the flash memory chip.

[0048] Test: the durability test of the flash memory chip is carried out based on the target environment parameters, the weight information and the database to obtain the durability result;

[0049] It should be noted that the durability result is obtained by carrying out the durability test of the flash memory chip based on the target environment parameters, the weight information and the database, that is, the test data under the weight information is continuously obtained in the database under the target environment parameters and is imported into the flash memory chip for writing and writing operation, and the maximum number of writing and writing operation of the flash memory chip is recorded to obtain the durability result.

[0050] Cross-validation: in the process of durability test of the flash memory chip, the working state of the flash memory chip is verified through a set cross-validation method to obtain a verification result, and the relationship between the real-time state and the durability of the flash memory chip is obtained through a set information acquisition method to obtain reflection information;

[0051] It should be noted that the working state of the flash memory chip is verified by the cross-validation method, and it is determined whether to end the durability test of the flash memory chip. At the same time, the failure position of the flash memory chip can be indirectly determined through the feedback of the verification result, so as to improve the efficiency of optimizing the flash memory chip. The relationship between the durability of the flash memory chip in different states is obtained by the information acquisition method, so as to facilitate the establishment of the durability prediction model.

[0052] Durability prediction: based on the reflection information, a prediction directory for predicting durability is established by a model establishment method, the running state to be predicted of the flash memory chip is obtained, the durability prediction model of the flash memory chip in the prediction directory is extracted, and the durability prediction result is obtained by the durability prediction model based on the running state to be predicted.

[0053] It should be noted that the prediction directory is established by the model establishment method, and the prediction directory includes several durability prediction models of the flash memory chip. The process of obtaining the running state to be predicted of the flash memory chip, that is, obtaining the running state of the flash memory chip by the staff, extracting the durability prediction model of the flash memory chip in the prediction directory, that is, obtaining the durability prediction model according to the model corresponding to the flash memory chip in the prediction directory, and obtaining the durability prediction result by the durability prediction model, so as to facilitate the staff to quickly understand the durability of the flash memory chip.

[0054] As shown in Figure 2 The database establishment method comprises: presetting a data type, obtaining basic data based on the real-time acquisition of data information based on the data type, splitting the basic data to obtain a plurality of sub-data, splitting the sub-data to obtain a plurality of data segments, reducing and increasing the sub-data based on the data segments to obtain derivative data, integrating the basic data and the derivative data to obtain overall data, establishing the association between the overall data and the data type, establishing a data sub-library with the data type as the name for storing the overall data, and integrating all data sub-libraries to obtain the database.

[0055] It should be noted that the process of presetting a data type, that is, presetting a plurality of data types, such as text data, audio data, etc. The basic data is obtained by acquiring data information under the data type, and the basic data is split into a plurality of sub-data. The process of splitting the sub-data to obtain a plurality of data segments and reducing and increasing the sub-data based on the data segments can be used for data derivation of the sub-data, improve the number of data used for testing, and store a plurality of different types of data in real time by the database, and perform data derivation on the stored data to improve the data quantity, as shown in Figure 2 Figure 2 The derivative data in the above formula is the data obtained by increasing and reducing the single data segment of the sub-data.

[0056] As​Figure 3 As shown, the environment processing method comprises: obtaining a working scene of a flash memory chip to obtain a basic scene, the basic scene comprising a plurality of sub-scenes, the sub-scenes comprising a plurality of factor items, presetting a detection time period and a fluctuation factor, generating a tuning point based on the detection time period, adding the fluctuation factor to the factor items in the sub-scenes to obtain a result factor parameter, integrating the result factor parameter to obtain a result scene, establishing a scene library for storing the result scene, and based on the environmental requirements, cooperating with the scene library to select the result scene corresponding to the environmental requirements in the scene library to obtain a target environmental parameter.

[0057] It should be noted that the process of obtaining a working scene of a flash memory chip to obtain a basic scene, i.e., obtaining a basic scene according to the specific working scene of the flash memory chip on the market, the basic scene comprising a plurality of sub-scenes, i.e., obtaining a sub-scene of a flash memory chip, integrating all sub-scenes to obtain a basic scene, the sub-scene comprising a plurality of factor items, the factor items including specific values of humidity, temperature and voltage, etc., the process of presetting a detection time period and a fluctuation factor, the time span of the detection time period does not have specific nature, which is an infinite time period, and the fluctuation factor is a specific fluctuation data corresponding to the unit of the factor item to simulate the real working scene, such as Figure 3 As shown, Figure 3 The table shows the single factor item before and after the fluctuation factor adjustment.

[0058] As shown, Figure 4 The data processing method comprises: obtaining a storage type of a flash memory chip based on type requirements to obtain a target storage type, splitting the target storage type to obtain a plurality of sub-storage types, obtaining a type number based on the plurality of sub-storage types, sorting the sub-storage types according to the storage importance based on the type requirements to obtain sorting information, presetting a weight distribution, the weight distribution comprising a basic weight and a decreasing factor, assigning weights to the sub-storage types based on the sorting information and the basic weight and the decreasing factor to obtain target weights, and integrating all sub-storage types and target weights to obtain weight information.

[0059] It should be noted that the process of sorting the sub-storage types according to the storage importance based on the type requirements, i.e. sorting the storage types of the flash memory chip provided by the staff according to the storage importance, assigning weights to the sub-storage types according to the sorting result, the basic weight includes a plurality of specific weights, and the decrement factor is a specific value of a fixed value that decreases in turn, for example, when the basic weight is 1 and the decrement factor is 0.1 that decreases by 0.01 in turn, when the type quantity is 6, then the decrement factor decreases by 5 times, and the weights of the sub-storage types in the sorting information are 0.55, 0.1, 0.09, 0.08, 0.07, 0.06 and 0.05, the weight represents the number of imported data, and the specific sub-storage type of the above weight is written into 100 data in the flash memory chip, and the corresponding number is input according to the weight, i.e. 55, 10, 9, 8, 7, 6 and 5, and the process of integrating all sub-storage types and target weights to obtain weight information is as shown in Figure 4 The target weight contains the weight corresponding to the sub-storage type. In the durability test of the flash memory chip, the test data is distributed based on the storage data type provided by the staff, which can improve the accuracy and practicality of the test. By simulating the actual application scenario (such as high-frequency writing of database logs or large-capacity video storage), the weight allocation can specifically expose the failure mechanism of different storage units (such as the high-temperature charge leakage problem of QLC), thereby optimizing the error correction algorithm and storage architecture design. At the same time, dynamically adjusting the test data ratio (such as allocating higher weight to low durability QLC) can accelerate the triggering of the failure critical point, shorten the test period and reduce resource consumption. In addition, the correlation analysis of weight distribution and durability results can provide key training data for the prediction model, helping engineers quickly evaluate chip reliability.

[0060] As shown in Figure 1 The cross-validation method includes: presetting fixed data and detection nodes, the fixed data is write data, writing the write data into the flash memory chip based on the detection nodes and writing out to obtain write-out data, judging whether the write data and the write-out data are consistent to obtain a judgment result, obtaining a first result when the judgment result feedback is that the write data and the write-out data are inconsistent, extracting the part of the write data and the write-out data that are inconsistent to obtain difference information, the first result feedback is that the working state of the flash memory chip is abnormal, obtaining a second result when the judgment result feedback is that the write data and the write-out data are consistent, the second result feedback is that the working state of the flash memory chip is normal, and integrating the difference information, the first result and the second result to obtain a verification result.

[0061] It should be noted that the process of presetting fixed data and detection nodes, i.e., setting fixed data and fixed time points for detecting fixed data, is determined according to actual use, and the process of writing data into a flash memory chip based on the detection nodes and writing out the obtained write-out data, i.e., writing fixed data into a flash memory chip at fixed detection nodes and obtaining write-out data, judging the durability of the flash memory chip by comparing the two, stopping the durability test of the flash memory chip when the judgment result is inconsistent, and obtaining difference information by extracting the part of the write-in data that is inconsistent with the write-out data, so that the damaged part of the flash memory chip can be locked by finding the corresponding part of the difference information for subsequent processing.

[0062] As shown in Figure 1 , the information acquisition method comprises: presetting a monitoring node, recording the test time during the durability test of the flash memory chip, recording the real-time state of the flash memory chip to obtain a reference state when the test time reaches the monitoring node, recording the data write-in and write-out times of the flash memory chip to obtain a preliminary result, combining the durability result with the preliminary result to obtain the subsequent durability result of the reference state, establishing the association relationship between the reference state and the subsequent durability result, and integrating the reference state, the subsequent durability result and the association relationship to obtain reflection information.

[0063] It should be noted that the monitoring node is a specific time point, and when the test time reaches the monitoring node, the real-time state of the flash memory chip is recorded to obtain a reference state, i.e., the state of the flash memory chip can be recorded, and by subsequently combining the durability result with the preliminary result to obtain the subsequent durability result of the reference state, the durability result under the reference state can be obtained to facilitate the establishment of the subsequent model.

[0064] As shown in Figure 1 , the model establishment method comprises:

[0065] Training data processing: based on the type classification of the flash memory chip, the reflection information is obtained to obtain a plurality of result information, the result information is split to obtain the reference state, the subsequent durability result and the association relationship, the reference state and the subsequent durability result with the association relationship are extracted to obtain the to-be-processed data, the to-be-processed data is preprocessed to obtain sub-training data, and the sub-training data under the same result information is integrated to obtain training data;

[0066] It should be noted that the process of obtaining a plurality of result information based on the type classification of the flash memory chip, i.e., classifying the reflection information according to the type of the flash memory chip, the process of preprocessing the to-be-processed data to obtain sub-training data, i.e., data cleaning, deduplication and feature enhancement operations on the to-be-processed data, and the process of integrating the sub-training data under the same result information to obtain training data, i.e., obtaining the training data of the model for predicting a single flash memory chip.

[0067] Model selection and training: select a deep learning model as the model base, import the training data into the model base for the model base to learn and obtain an initial model;

[0068] It should be noted that the process of selecting a deep learning model as the model base, i.e., selecting a deep learning model, can select a convolutional neural network in the specific implementation process.

[0069] Model adjustment and output: preset verification state and verification result, import the verification state into the initial model to obtain the prediction result, compare the prediction result with the verification result to obtain the comparison result, and optimize the initial model based on the comparison result to obtain the durability prediction model;

[0070] It should be noted that the verification state corresponds to the verification result, and the verification state is a data set consistent with the to-be-predicted running state parameter. By importing the verification state into the initial model, the prediction result is obtained, the prediction result is compared with the verification result to obtain the comparison result, and the initial model is optimized based on the comparison result to obtain the durability prediction model. This process, i.e., according to the difference between the prediction result and the verification result, optimizes the initial model to improve the robustness of the model.

[0071] Establishing model association: establishing the association between the flash memory chip and the durability prediction model, and establishing a prediction directory for storing the flash memory chip, the durability prediction model and the association.

[0072] It should be noted that different types of flash memory chips correspond to different durability prediction models, so as to reduce the difficulty of model training and improve the efficiency of using the durability prediction model subsequently. Through the set association, the corresponding durability prediction model corresponding to the flash memory chip can be obtained according to the type of the flash memory chip.

[0073] As shown in Figure 5 A durability test device for a flash memory chip, which includes a programmable environmental test box capable of adjusting humidity, temperature, voltage and other parameters, a flash memory test host for writing and writing test data, a data acquisition system for acquiring test data, a weight adjustment engine for adjusting the number of data types, and a test database for collecting durability test results. The durability prediction model is also included, which can predict the durability of the flash memory chip according to its running state.

[0074] It should be noted that the programmable environmental test box capable of adjusting humidity, temperature, voltage and other parameters is used to adjust the humidity, temperature, voltage and other parameters of the flash memory chip environment during the test process. The data acquisition system is set up to collect test data and establish a database. The flash memory test host is set up to access the flash memory chip and write and write data. The test database is set up to store the durability results and test states of the flash memory chip.

[0075] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since various modifications can be made by those skilled in the art, without departing from the spirit and scope of the application, which are defined by the appended claims and their equivalents.

Claims

1. A method for endurance testing of a flash memory chip, the method comprising: information acquisition: obtaining the storage type of the flash memory chip and the test environment requirements to obtain the test requirements, the test requirements including type requirements and environment requirements; characterized in that it comprises: database establishment: establishing a real-time updated test data repository by a set database establishment method to obtain a database; environment setting and determination: setting the test environment according to the working scenario of the flash memory chip by an environment processing method to obtain target environment parameters; type setting and determination: setting the test data weight according to the storage type of the flash memory chip by a data processing method to obtain weight information in combination with the type requirements; test performance: performing endurance testing of the flash memory chip based on the target environment parameters, the weight information and the database to obtain endurance results; cross-validation: verifying the working state of the flash memory chip during the endurance testing of the flash memory chip by a set cross-validation method to obtain verification results, and obtaining the relationship between the real-time state and the endurance of the flash memory chip by a set information acquisition method to obtain reflection information; endurance prediction: establishing a prediction directory for predicting endurance based on the reflection information by a model establishment method, obtaining the to-be-predicted running state of the flash memory chip, extracting an endurance prediction model in the prediction directory based on the flash memory chip, and obtaining an endurance prediction result based on the to-be-predicted running state by the endurance prediction model.

2. The method of claim 1, wherein: The database establishment method comprises: presetting a data type, obtaining basic data based on the data type, splitting the basic data to obtain a plurality of sub-data, splitting the sub-data to obtain a plurality of data segments, reducing and increasing the sub-data based on the data segments to obtain derivative data, integrating the basic data and the derivative data to obtain overall data, establishing the association between the overall data and the data type, establishing a data sub-library with the data type as the name for storing the overall data, and integrating all data sub-libraries to obtain the database.

3. The method of claim 1, wherein: The environment processing method comprises: obtaining a basic scenario of the flash memory chip, the basic scenario comprising a plurality of sub-scenarios, the sub-scenarios comprising a plurality of factor items, presetting a detection time period and a fluctuation factor, generating a tuning node based on the detection time period, adding the fluctuation factor to the factor items in the sub-scenarios to obtain result factor parameters based on the tuning node, integrating the result factor parameters to obtain a result scenario, establishing a scenario library for storing the result scenario, and obtaining the target environment parameters by selecting the result scenario corresponding to the environment requirements in the scenario library based on the environment requirements and the scenario library.

4. The method of claim 1, wherein: The data processing method comprises: obtaining the target storage type of the flash memory chip based on the type requirements, splitting the target storage type to obtain a plurality of sub-storage types, obtaining the type quantity based on the plurality of sub-storage types, sorting the sub-storage types according to the storage importance based on the type requirements to obtain sorting information, presetting a weight distribution, the weight distribution comprising a basic weight and a decrement factor, assigning weights to the sub-storage types based on the sorting information, the basic weight and the decrement factor to obtain target weights, and integrating all sub-storage types and target weights to obtain weight information.

5. The method of claim 1, wherein: The cross-validation method comprises: presetting fixed data and a detection node, the fixed data being write data, writing the write data into a flash memory chip based on the detection node and writing out to obtain write-out data, judging whether the write data and the write-out data are consistent to obtain a judgment result, obtaining a first result when the judgment result feedback is that the write data and the write-out data are inconsistent, extracting a part where the write data and the write-out data are inconsistent to obtain difference information, the first result feedback being that the working state of the flash memory chip is abnormal, obtaining a second result when the judgment result feedback is that the write data and the write-out data are consistent, the second result feedback being that the working state of the flash memory chip is normal, and integrating the difference information, the first result and the second result to obtain a verification result.

6. The method of claim 1, wherein: The information acquisition method comprises: presetting a monitoring node, recording a test time in a process of performing a durability test on the flash memory chip, recording a real-time state of the flash memory chip to obtain a reference state when the test time reaches the monitoring node, recording a data write-in and write-out times of the flash memory chip to obtain a preliminary result, combining a durability result with the preliminary result to obtain a subsequent durability result of the reference state, establishing an association relationship between the reference state and the subsequent durability result, and integrating the reference state, the subsequent durability result and the association relationship to obtain reflection information.

7. The method of claim 1, wherein: the method is performed by a tester. The model establishment method comprises: training data processing: obtaining a plurality of result information based on type classification of the flash memory chip, splitting the result information to obtain the reference state, the subsequent durability result and the association relationship, extracting the reference state and the subsequent durability result having the association relationship to obtain to-be-processed data, pre-processing the to-be-processed data to obtain sub-training data, and integrating the sub-training data under the same result information to obtain training data; model selection and training: selecting a deep learning model as a model base, importing the training data into the model base for learning by the model base to obtain an initial model; model adjustment and output: presetting a verification state and a verification result, importing the verification state into the initial model to obtain a prediction result, comparing the prediction result with the verification result to obtain a comparison result, optimizing the initial model based on the comparison result to obtain a durability prediction model; establishing model association: establishing the association between the flash memory chip and the durability prediction model, and establishing a prediction directory for storing the flash memory chip, the durability prediction model and the association.

8. A durability test apparatus for a flash memory chip, characterized by: A durability test method for a flash memory chip is used.

Images