Hard disk prediction method, device, equipment, medium, product and training method

Abstract

The present application provides a hard disk prediction method, apparatus, device, medium, product, and training method, which can be applied to the field of hard disk technology. The hard disk prediction method includes: obtaining the type, performance data, and usage status data of the hard disk to be predicted; processing the usage status data to obtain the health decay moment of the hard disk to be predicted during use, wherein the health decay moment indicates the moment when the health decay degree of the hard disk to be predicted reaches a preset threshold; obtaining a health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted; and obtaining a prediction result of the hard disk to be predicted based on the health decay moment, health score, and performance data, wherein the prediction result indicates the health decay trend and performance decay trend of the hard disk to be predicted.

Description

Technical Field

[0001] The present application relates to the field of hard disk technology, and specifically to a hard disk prediction method, device, equipment, medium, product, and training method. Background Art

[0002] In computers and servers, data storage typically relies on hard drives. Hard drives retain information even after a power outage, making them suitable for long-term data storage. In practice, operating systems, applications, documents, images, and videos all need to be stored on hard drives. Therefore, as an essential storage component in computers and servers, hard drives are particularly important. However, hard drives experience performance degradation over time, leading to limited accuracy in predicting their performance and health. Summary of the Invention

[0003] In view of the above problems, the present application provides a hard disk prediction method, device, equipment, medium, product and training method.

[0004] According to the first aspect of the present application, a hard disk prediction method is provided, including: obtaining the type, performance data and usage status data of the hard disk to be predicted; processing the usage status data to obtain the health decay moment of the hard disk to be predicted during use, wherein the health decay moment indicates the moment when the health decay degree of the hard disk to be predicted reaches a preset threshold; obtaining a health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted; obtaining a prediction result of the hard disk to be predicted based on the health decay moment, the health score and the performance data, wherein the prediction result indicates the health decay trend and performance decay trend of the hard disk to be predicted.

[0005] The second aspect of the present application provides a model training method, including: using an initial model to process sample usage status data of a sample hard disk to obtain a sample health decay moment of the sample hard disk during use, wherein the sample health decay moment indicates the moment when the health decay degree of the sample hard disk reaches a preset threshold; using the initial model to obtain a sample health score of the sample hard disk based on the sample type and sample usage status data of the sample hard disk; using the initial model to obtain a sample prediction result of the sample hard disk based on the sample health decay moment, the sample health score and the sample performance data of the sample hard disk, wherein the sample prediction result indicates a health decay prediction trend and a performance decay prediction trend of the sample hard disk; based on a target loss function, using the sample prediction result and the sample label, training the initial model to obtain a trained target model, wherein the sample label indicates an actual trend of health decay and an actual trend of performance decay of the sample hard disk.

[0006] The third aspect of the present application provides a hard disk prediction device, including: an acquisition module for acquiring the type, performance data and usage status data of the hard disk to be predicted; a first acquisition module for processing the usage status data to obtain the health decay moment of the hard disk to be predicted during use, wherein the health decay moment indicates the moment when the health decay degree of the hard disk to be predicted reaches a preset threshold; a second acquisition module for obtaining a health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted; a third acquisition module for obtaining a prediction result of the hard disk to be predicted based on the health decay moment, health score and performance data, wherein the prediction result indicates the health decay trend and performance decay trend of the hard disk to be predicted.

[0007] The fourth aspect of the present application provides an electronic device, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0008] The fifth aspect of the present application further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0009] The sixth aspect of the present application further provides a computer program product, comprising a computer program or instructions, which implement the steps of the above method when executed by a processor.

[0010] According to an embodiment of the present application, the health decay moment of the hard disk to be predicted is obtained through the usage status data of the hard disk to be predicted, the health score of the hard disk to be predicted is obtained through the type and usage status data of the hard disk to be predicted, and the prediction result representing the health decay trend and performance decay trend of the hard disk to be predicted is obtained through the health decay moment, health score and performance data. This fully combines the health-related data and performance-related data of the hard disk to be predicted, so that more accurate prediction results can be obtained based on the relationship between performance and health, avoiding the impact caused by sudden decay of the health and performance of the hard disk to be predicted. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above contents and other objects, features and advantages of the present application will become more apparent through the following description of the embodiments of the present application with reference to the accompanying drawings, in which:

[0012] Figure 1 An application scenario diagram of the hard disk prediction method, apparatus, equipment, medium, product, and training method according to an embodiment of the present application is shown.

[0013] Figure 2 A flowchart of a hard disk prediction method according to an embodiment of the present application is shown.

[0014] Figure 3 A flowchart of a model training method according to an embodiment of the present application is shown.

[0015] Figure 4 A structural block diagram of a hard disk prediction device according to an embodiment of the present application is shown.

[0016] Figure 5 The figure shows a structural block diagram of a model training device according to an embodiment of the present application.

[0017] Figure 6 A block diagram of an electronic device suitable for implementing a hard disk prediction method and a model training method according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present application. In the detailed description below, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present application. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present application.

[0019] The terms used herein are only for describing specific embodiments and are not intended to limit this application. The terms "comprise," "include," etc. used herein indicate the presence of the features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0020] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0021] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted in accordance with the meaning commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0022] With the rapid growth in data storage demand, server hard drives, particularly non-volatile memory express solid state drives (NVME SSDs), are gaining widespread adoption in data centers and other fields due to their high read / write speeds and low latency. As the core storage medium for NVME SSDs, the performance and reliability of NAND gate (NAND) flash memory are crucial to the entire storage system. NAND flash memory's media characteristics, such as the number of program / erase (P / E) cycles, determine the upper limit of program and erase operations per flash cell, while the uncorrectable bit error rate (UBER) reflects the error risk associated with data storage. However, over time, NAND flash memory can experience performance degradation and increased failure risk due to constant read / write operations. Therefore, predicting the performance and health of hard drives like NVME SSDs is crucial.

[0023] In one example, stability can be assessed through long-term random read operations, but the drive's health (such as P / E cycles and UBER) cannot be correlated with performance curves. Consequently, there is a lack of dynamic feedback on media health, making it impossible to assess the impact of media aging on performance. In another example, the drive's health can be understood by observing its usage status, but this only allows for static status monitoring and does not dynamically adjust test parameters (such as queue depth and block size) based on the drive's health, resulting in insufficient test coverage.

[0024] In view of this, an embodiment of the present application provides a hard disk prediction method, including: obtaining the type, performance data and usage status data of the hard disk to be predicted; processing the usage status data to obtain the health decay moment of the hard disk to be predicted during use, wherein the health decay moment indicates the moment when the health decay degree of the hard disk to be predicted reaches a preset threshold; obtaining a health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted; obtaining a prediction result of the hard disk to be predicted based on the health decay moment, health score and performance data, wherein the prediction result indicates the health decay trend and performance decay trend of the hard disk to be predicted.

[0025] Figure 1 An application scenario diagram of the hard disk prediction method, apparatus, equipment, medium, product, and training method according to an embodiment of the present application is shown.

[0026] like Figure 1As shown, the application scenario 100 according to this embodiment may include a hard disk 101 and a server 102. Data between the hard disk 101 and the server 102 may be transmitted wirelessly through a network or wiredly through cables.

[0027] Hard disk 101 may be a flash memory medium that provides storage. Server 102 may be a server that provides various services. The backend management server may analyze and process the received data and provide feedback to the terminal device regarding the processing results (e.g., prediction results based on the type, performance data, and usage status data of the hard disk to be predicted).

[0028] It should be noted that the hard disk prediction method provided in the embodiment of the present application can generally be executed by the server 102. Accordingly, the hard disk prediction device provided in the embodiment of the present application can generally be set in the server 102. The hard disk prediction method provided in the embodiment of the present application can also be executed by a server or server cluster that is different from the server 102 and can communicate with the hard disk 101 and / or the server 102. Accordingly, the hard disk prediction device provided in the embodiment of the present application can also be set in a server or server cluster that is different from the server 102 and can communicate with the hard disk 101 and / or the server 102.

[0029] It should be understood that Figure 1 The number of hard disks and servers in the embodiment is merely illustrative. Any number of hard disks and servers may be provided as required.

[0030] Figure 2 A flowchart of a hard disk prediction method according to an embodiment of the present application is shown.

[0031] like Figure 2 As shown, the hard disk prediction method of this embodiment includes operations S210 to S240.

[0032] In operation S210 , the type, performance data, and usage status data of the hard disk to be predicted are obtained.

[0033] In operation S220, the usage status data is processed to obtain a predicted health degradation moment of the hard disk during use, wherein the health degradation moment indicates a moment when the health degradation degree of the hard disk reaches a preset threshold.

[0034] In operation S230 , a health score of the hard disk to be predicted is obtained according to the type and usage status data of the hard disk to be predicted.

[0035] In operation S240, a prediction result of the hard disk to be predicted is obtained based on the health decay time, health score and performance data, wherein the prediction result indicates the health decay trend and performance decay trend of the hard disk to be predicted.

[0036] According to an embodiment of the present application, the type, performance data, and usage status data of the hard disk to be predicted can be obtained through an open interface. The type of the hard disk to be predicted can be, for example, a single-level cell (SLC) type, a multi-level cell (MLC) type, a triple-level cell (TLC) type, or a quad-level cell (QLC) type. The performance data of the hard disk to be predicted can be the bandwidth and latency of the hard disk to be predicted. The usage status data of the hard disk to be predicted can be the read and write times, uncorrectable bit error rate, and error correction count of the hard disk to be predicted.

[0037] According to an embodiment of the present application, the usage status data can be converted into a curve graph in chronological order, and the health decay moment of the hard disk to be predicted during use can be obtained based on the changes in the curve graph. The health decay moment can be the moment when the curve in the curve graph drops by more than a preset amplitude per unit time, which means the moment when the health decay degree of the hard disk to be predicted reaches a preset threshold.

[0038] According to embodiments of the present application, the type of hard drive being predicted can be correlated with its health, and usage data can also influence its health. Therefore, based on the type and usage data of the hard drive being predicted, a health check can be performed on the hard drive to obtain a health score.

[0039] According to an embodiment of the present application, the health decay time can be the time when the health of the hard drive to be predicted decreases, the health score can be the current health of the hard drive to be predicted, and the performance data can be the current performance of the hard drive to be predicted. Based on the health decay time, health score, and performance data, the health and performance of the hard drive to be predicted can be comprehensively predicted to obtain prediction results for both the health and performance of the hard drive to be predicted. The prediction results can indicate the health decay trend and performance decay trend of the hard drive to be predicted.

[0040] According to an embodiment of the present application, the health decay moment of the hard disk to be predicted is obtained through the usage status data of the hard disk to be predicted, the health score of the hard disk to be predicted is obtained through the type and usage status data of the hard disk to be predicted, and the prediction result representing the health decay trend and performance decay trend of the hard disk to be predicted is obtained through the health decay moment, health score and performance data. This fully combines the health-related data and performance-related data of the hard disk to be predicted, so that more accurate prediction results can be obtained based on the relationship between performance and health, avoiding the impact caused by sudden decay of the health and performance of the hard disk to be predicted.

[0041] According to an embodiment of the present application, a prediction result of the hard disk to be predicted is obtained based on the health decay time, health score and performance data, including: determining the degree of correlation between the health decay time, health score and performance data based on the health decay time, health score and performance data; and obtaining the prediction result of the hard disk to be predicted using a predetermined network based on the degree of correlation between the health decay time, health score and performance data.

[0042] According to an embodiment of the present application, the degree of correlation between the health decay moment, the health score, and the performance data can be determined based on the changes in the health decay moment, the health score, and the performance data at different moments. For example, the health decay moments are moment a and moment b, and the health score at moment a is 96, the performance data still characterizes the hard disk to be predicted as normal, and the health score at moment b is 85, and the performance data characterizes the hard disk to be predicted as good. It can be seen that the degree of correlation between the health decay moment and the health score is large, the degree of correlation between the health decay moment and the performance data is medium, and the degree of correlation between the health score and the performance data is medium. Therefore, the correlation between the health decay trend and the performance decay trend in the prediction result can also be medium.

[0043] According to the embodiments of the present application, by analyzing the correlation between health-related parameters and performance-related parameters, a more accurate health degradation trend and performance degradation trend of the hard disk to be predicted can be further obtained.

[0044] According to an embodiment of the present application, the usage status data is processed to obtain the health decay moment of the hard disk to be predicted during use, including: using a bidirectional long short-term memory network to process the usage status data to obtain forward timing information and reverse timing information based on the usage status data; fusing the forward timing information and the reverse timing information to obtain fused timing information; based on the fused timing information, using an attention mechanism to obtain the health decay moment of the hard disk to be predicted during use.

[0045] According to an embodiment of the present application, a bidirectional long short-term memory (BiLSTM) network may include a forward LSTM network and a reverse LSTM network. By combining the forward LSTM network and the reverse LSTM network, the long-term dependency in the input data is captured. Long-term dependency is the important association between elements that are far apart in the input data. Specifically, in an embodiment of the present application, the usage status data is processed using a bidirectional long short-term memory network, and the forward time series information and reverse time series information in the usage status data can be captured. Among them, the forward time series information can be the information obtained by processing the usage status data from the time series start point to the time series end point through the forward LSTM network, and the reverse time series information can be the information obtained by processing the usage status data from the time series end point to the time series start point through the reverse LSTM network. The forward time series information includes historical information, while the reverse time series information includes future information. Therefore, by fusing the forward time series information and the reverse time series information, the fusion information that includes both historical information and future information can be obtained.

[0046] According to the embodiments of the present application, the attention mechanism can dynamically assign weights to allow the network to focus on the key parts of the sequence. Therefore, the attention mechanism can be used to focus on the key parts of the fused time series information, namely the health degradation moments of the hard disk to be predicted during use.

[0047] According to the embodiments of the present application, the long-term dependencies in the usage status data can be obtained through the bidirectional long short-term memory network, and the key health decay moments can be focused on through the attention mechanism, thereby improving the prediction accuracy of the hard disk to be predicted.

[0048] According to an embodiment of the present application, a health score of the hard disk to be predicted is obtained based on the type and usage status data of the hard disk to be predicted, including: obtaining an arrangement order of target indicators related to the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted, wherein the arrangement order represents the degree of influence of the target indicators on the health of the hard disk to be predicted; and obtaining the health score of the hard disk to be predicted using a health scoring network based on the arrangement order of the target indicators.

[0049] The order of target indicators related to the predicted hard disk can be determined based on the type of the predicted hard disk and the changes in usage status data at different times. For example, a gradient boosting decision tree (GBDT) can be used to determine the contribution of target features to the predicted hard disk, thereby determining the order of target indicators.

[0050] The target indicator may include at least one of the hard drive's read / write cycle count, uncorrectable bit error rate (BER), error correction count, and temperature. The read / write cycle count records the number of read / write cycles of the hard drive, thereby assessing the wear level of the hard drive. The UBER (Uncorrectable Bit Error Rate) can reflect the data integrity risk of the hard drive. The error correction count measures the frequency of error corrections in the hard drive, reflecting the media reliability of the hard drive. The target indicators may be arranged in the order of read / write cycle count, error correction count, uncorrectable bit error rate, and temperature.

[0051] According to an embodiment of the present application, weights may be assigned to target indicators according to their arrangement order, and then a health score network may be used to obtain a health score of the hard disk to be predicted.

[0052] According to an embodiment of the present application, the health scoring network can be established based on a random forest, which can improve the prediction performance of the machine learning model.

[0053] According to the embodiments of the present application, by analyzing and obtaining the arrangement order of the target indicators, it is possible to avoid the influence of indicators with low relevance on the health score, thereby improving the accuracy of the health score.

[0054] According to an embodiment of the present application, the prediction result includes at least one of the expected lifespan, performance degradation trend, and failure risk classification of the hard disk to be predicted.

[0055] According to an embodiment of the present application, the hard disk prediction method may further include: generating life warning prompt information for the hard disk to be predicted when the expected life is less than the expected life threshold; lowering the processing level of the hard disk to be predicted when the performance degradation trend is downward and the health score of the hard disk to be predicted indicates that the hard disk to be predicted is in an unhealthy state; and performing a risk scan on the hard disk to be predicted when the failure risk is classified as high risk.

[0056] According to an embodiment of the present application, the expected lifespan may be the remaining read and write times of the hard disk to be predicted, the performance degradation trend may be the speed of performance degradation of the hard disk to be predicted, and the failure risk classification may be the different risk levels of the hard disk to be predicted.

[0057] According to an embodiment of the present application, the life expectancy threshold may be a threshold at which the remaining life of the hard drive being predicted is insufficient to support continued operation of the hard drive being predicted. If the life expectancy is less than the life expectancy threshold, a life expectancy warning message for the hard drive being predicted may be generated to prevent other impacts caused by a sudden failure of the hard drive being predicted. The life expectancy threshold may be, for example, 100 hours.

[0058] When the performance degradation trend is downward and the health score of the hard disk to be predicted indicates that the hard disk to be predicted is in an unhealthy state, the processing level of the hard disk to be predicted can be reduced. For example, the queue depth of the hard disk to be predicted is automatically reduced, which can reduce the processing pressure of the hard disk to be predicted.

[0059] According to an embodiment of the present application, fault line classification can include low risk, medium risk, and high risk. If the fault risk is classified as low risk, the predicted hard drive may not be processed. If the fault risk is classified as medium risk, a risk warning message may be issued. If the fault risk is classified as high risk, the predicted hard drive may be directly scanned for risks. High risk can be determined if the uncorrectable bit error rate exceeds a threshold or the error correction count increases suddenly.

[0060] According to the embodiments of the present application, different prediction results are used to comprehensively manage the hard disk to be predicted, thereby improving the management of the hard disk to be predicted. Problems with the hard disk to be predicted can be discovered in a timely manner, facilitating timely processing and reducing the impact caused by sudden failures of the hard disk to be predicted.

[0061] According to an embodiment of the present application, the hard disk prediction method may further include: determining test data of the hard disk to be predicted in multiple test stages based on the prediction results, so as to perform a stress test on the hard disk to be predicted using the test data to obtain a stress test result.

[0062] According to the embodiments of the present application, since the prediction results represent the health degradation trend and performance degradation trend of the hard disk to be predicted, a stress test plan for the hard disk to be predicted can be formulated based on the health degradation trend and performance degradation trend of the hard disk to be predicted. The test can be divided into multiple test phases, such as a high wear phase, a high error phase, and a steady-state phase. Different test data can be adapted for different phases, and the test data can be used to perform a stress test on the hard disk to be predicted to obtain a stress test result. The stress test result can also be compared with the prediction result. If the comparison results differ significantly, the parameters used in the prediction process of the hard disk to be predicted can be adjusted.

[0063] According to the embodiments of this application, in scenarios with high wear and tear, the queue depth of the hard drive to be predicted can be reduced (for example, from QD64 to QD32), reducing the pressure on the hard drive to be predicted. In scenarios with high bit error rates, high-error packets can be actively injected to trigger the firmware error correction algorithm and record the recovery time.

[0064] According to embodiments of the present application, multi-stage combined testing is also possible. For example, a steady-state stress test can be performed on the hard drive under investigation, using both sequential write and random read operations to continuously monitor performance fluctuations. Transient shock testing can also be performed by simulating voltage fluctuations or sudden temperature changes to verify the stability of the hard drive under investigation.

[0065] According to the embodiments of the present application, test data can be matched more specifically to different test phases based on the prediction results, thereby reducing the preparation process for stress testing and improving the efficiency of stress testing.

[0066] According to an embodiment of the present application, the hard disk prediction method may further include: selecting a target test stage and target test data of the target test stage from multiple test stages based on the prediction results, wherein the expected life and performance of the hard disk to be predicted in the target test stage are lower than those in other test stages, so as to perform a target stress test on the hard disk to be predicted based on the target test data to obtain a target stress test result.

[0067] According to an embodiment of the present application, the multiple test stages of the stress test will include a steady-state stage, that is, a stage in which the performance and life of the hard disk to be predicted are relatively stable, and this stage is a necessary process in the testing process. Therefore, based on the prediction results, the target test stage and the target test data of the target test stage can be selected from the multiple test stages, and the hard disk to be predicted can be pre-processed for accelerated aging based on the target test data, so that the state of the hard disk to be predicted can reach the state of the target test stage as soon as possible, thereby performing a target stress test on the hard disk to be predicted based on the target test data to obtain the target stress test result.

[0068] According to an embodiment of the present application, by selecting a target test phase, the test cycle of the hard disk to be predicted can be shortened.

[0069] According to an embodiment of the present application, the test data includes a data packet for simulating errors, and the data packet is used to trigger an error correction operation of the hard disk to be predicted and record the error correction recovery time of the hard disk to be predicted.

[0070] According to an embodiment of the present application, during the test process, a data packet used to simulate an error may be added to the test data to trigger error correction operations on the hard drive to be predicted and record the error correction recovery time of the hard drive to be predicted. For example, a data packet may be added to the test data after every 5,000 cycles during the test.

[0071] According to an embodiment of the present application, adding a data packet simulating an error into the test data can detect the error correction capability of the hard disk to be predicted, thereby improving the comprehensiveness of the stress test of the hard disk to be predicted.

[0072] Figure 3 A flowchart of a model training method according to an embodiment of the present application is shown.

[0073] like Figure 3 As shown, the model training method includes operations S310 to S340.

[0074] In operation S310, the initial model is used to process the sample usage status data of the sample hard disk to obtain a sample health decay time of the sample hard disk during use, wherein the sample health decay time indicates the time when the health decay degree of the sample hard disk reaches a preset threshold.

[0075] In operation S320 , the sample health score of the sample hard disk is obtained according to the sample type and sample usage status data of the sample hard disk using the initial model.

[0076] In operation S330, the initial model is used to obtain a sample prediction result of the sample hard disk based on the sample health decay time, the sample health score, and the sample performance data of the sample hard disk. The sample prediction result represents the health decay prediction trend and performance decay prediction trend of the sample hard disk.

[0077] In operation S340, based on the target loss function, the initial model is trained using the sample prediction results and the sample labels to obtain a trained target model, wherein the sample labels indicate the actual trends of health degradation and performance degradation of the sample hard disks.

[0078] According to an embodiment of the present application, the sample usage status data and the sample performance data may be data of the sample hard disk at a first moment, and the sample tag may be data of the sample hard disk at a second moment, where the first moment is before the second moment.

[0079] According to an embodiment of the present application, sample types, sample performance data, and sample usage status data of sample hard disks may be input into the initial model as training data.

[0080] According to an embodiment of the present application, the sample usage status data can be converted into a curve graph in chronological order, and the sample health decay moment of the sample hard disk during use can be obtained based on the changes in the curve graph.

[0081] The type of the sample hard drive may be related to the health of the sample hard drive, and the sample usage status data may also reflect the health of the sample hard drive. Therefore, the health of the sample hard drive may be tested based on the sample type and sample usage status data of the sample hard drive to obtain a sample health score of the sample hard drive.

[0082] According to an embodiment of the present application, the sample health decay time, the sample health score and the sample performance data are combined to obtain sample prediction results of the sample hard disk in terms of health and performance.

[0083] According to an embodiment of the present application, a target loss function is established based on the initial model. Since the sample labels indicate the actual health degradation trend and performance degradation trend of the sample hard disk, and the sample prediction results indicate the predicted health degradation trend and performance degradation trend of the sample hard disk, the difference between the sample prediction results and the sample labels can be used to adjust the parameters of the initial model to obtain a trained target model.

[0084] According to an embodiment of the present application, the target model can be used to predict the hard disk to be predicted. Since the target model is trained based on sample performance data representing the performance of the sample hard disk, and sample usage status data and sample labels representing the health of the sample hard disk, the obtained target model can fully combine the correlation between health and performance to improve the accuracy of hard disk prediction.

[0085] According to an embodiment of the present application, the initial model includes an initial bidirectional long short-term memory network, an initial health score network and an initial reservation network. Based on the target loss function, the initial model is trained using sample prediction results and sample labels to obtain a trained target model, including: based on the target loss function, obtaining a loss value according to the sample prediction results and sample labels; based on the loss value, adjusting the parameters of the initial bidirectional long short-term memory network, the parameters of the initial health score network and the parameters of the initial reservation network until the loss value converges to obtain a trained target model.

[0086] According to an embodiment of the present application, the structure of the initial bidirectional long short-term memory network can refer to the structure of the long short-term memory network. The initial bidirectional long short-term memory network is the network before parameter adjustment. The sample usage status data of the sample hard disk can be processed by the initial bidirectional long short-term memory network to obtain the sample health decay moment of the sample hard disk during use.

[0087] According to an embodiment of the present application, the structure of the initial health scoring network can refer to the structure of the health scoring network. The initial health scoring network is the network before parameter adjustment. The sample health score of the sample hard disk can be obtained through the initial health scoring network based on the sample type and sample usage status data of the sample hard disk.

[0088] According to an embodiment of the present application, the structure of the initial predetermined network can refer to the structure of the predetermined network. The initial predetermined network is the network before parameter adjustment. The sample prediction result of the sample hard disk can be obtained through the initial predetermined network based on the sample health decay time, the sample health score and the sample performance data of the sample hard disk.

[0089] According to an embodiment of the present application, the sample prediction result may also include at least one of the sample life expectancy, the sample performance degradation trend, and the sample failure risk classification. The sample life expectancy can be expressed by the following formula:

[0090] (1)

[0091] in, represents the sample life expectancy, Indicates the maximum read and write cycles of the sample hard disk. Indicates the current read and write cycle of the sample hard disk, UBER indicates the uncorrectable bit error rate of the sample hard disk, Indicates the error correction count curve of this hard disk, where k and b are parameters.

[0092] According to an embodiment of the present application, as an example, when the hard disk is of TLC type, k=0.15, b=0.02. When the hard disk is of QLC type, k=0.25, b=0.03.

[0093] According to embodiments of the present application, training data can be obtained through accelerated aging testing of sample hard drives. For example, continuous writing can be performed at high temperature (85°C), recording the total number of P / E read / write cycles from 0 to failure, and recording the performance degradation trajectory of the sample hard drive before failure.

[0094] The objective loss function of the initial model can be as follows:

[0095] (2)

[0096] MAE(RUL) represents lifespan loss, MSE represents performance loss, and CrossEntropy represents failure risk classification loss. α, β, and γ are all parameters. For example, the values ​​of α, β, and γ can be 0.5, 0.3, and 0.2, respectively.

[0097] According to an embodiment of the present application, based on the target loss function, a loss value can be obtained according to the sample prediction results and sample labels; based on the loss value, the parameters of the initial bidirectional long short-term memory network, the parameters of the initial health score network, and the parameters of the initial predetermined network can be adjusted until the loss value converges, and the trained target model is obtained, that is, the long short-term memory network, the health score network, and the predetermined network of the target model are obtained.

[0098] According to the embodiments of the present application, by establishing a target model between the health of the hard disk and performance degradation, the early warning capability of potential hard disk failures is provided, thereby ensuring the working status of the hard disk.

[0099] The present application also provides a test environment for simulating a hard disk for stress testing. The test environment includes a hard disk to be tested, a monitoring tool, and an environmental simulation device. The environmental simulation device can be a temperature-adjustable constant temperature box and a programmable power supply with a ±5% voltage fluctuation. This test environment provides a temperature and voltage-controllable test environment, which enables more precise control of the stress testing process of the hard disk to be tested.

[0100] Based on the above hard disk prediction method, this application also provides a hard disk prediction device. Figure 4 The device is described in detail.

[0101] Figure 4 A structural block diagram of a hard disk prediction device according to an embodiment of the present application is shown.

[0102] like Figure 4 As shown, the hard disk prediction device 400 of this embodiment includes an acquisition module 410 , a first obtaining module 420 , a second obtaining module 430 and a third obtaining module 440 .

[0103] The acquisition module 410 is used to acquire the type, performance data, and usage status data of the hard disk to be predicted. In one embodiment, the acquisition module 410 can be used to perform the operation S210 described above, which will not be repeated here.

[0104] A first obtaining module 420 is configured to process the usage status data to obtain a predicted health degradation moment of the hard disk during use. The health degradation moment indicates the moment when the health degradation of the hard disk reaches a preset threshold. In one embodiment, the first obtaining module 420 may be configured to perform operation S220 described above and will not be further described here.

[0105] The second obtaining module 430 is used to obtain the health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted. In one embodiment, the second obtaining module 430 can be used to perform the operation S230 described above, which will not be repeated here.

[0106] The third obtaining module 440 is configured to obtain a prediction result for the hard drive to be predicted based on the health decay time, the health score, and the performance data. The prediction result indicates the health decay trend and performance decay trend of the hard drive to be predicted. In one embodiment, the third obtaining module 440 can be used to perform operation S240 described above and will not be further described here.

[0107] According to an embodiment of the present application, the third obtaining module 440 for obtaining the prediction result of the hard disk to be predicted based on the health decay time, health score and performance data includes: a first obtaining unit, for determining the degree of correlation between the health decay time, health score and performance data based on the health decay time, health score and performance data; a second obtaining unit, for obtaining the prediction result of the hard disk to be predicted using a predetermined network based on the degree of correlation between the health decay time, health score and performance data.

[0108] According to an embodiment of the present application, a first obtaining module 420 for processing usage status data to obtain the health decay moment of the hard disk to be predicted during use includes: a third obtaining unit for processing the usage status data using a bidirectional long short-term memory network to obtain forward timing information and reverse timing information based on the usage status data; a fourth obtaining unit for fusing the forward timing information and the reverse timing information to obtain fused timing information; and a fifth obtaining unit for obtaining the health decay moment of the hard disk to be predicted during use using an attention mechanism based on the fused timing information.

[0109] According to an embodiment of the present application, the second obtaining module 430 for obtaining the health score of the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted includes: a sixth obtaining unit, for obtaining the arrangement order of target indicators related to the hard disk to be predicted based on the type and usage status data of the hard disk to be predicted, wherein the arrangement order represents the degree of influence of the target indicators on the health of the hard disk to be predicted; and a seventh obtaining unit, for obtaining the health score of the hard disk to be predicted using a health score network based on the arrangement order of the target indicators.

[0110] According to an embodiment of the present application, the target indicator includes at least one of the read / write cycle, uncorrectable bit error rate, error correction count, and temperature of the hard disk to be predicted.

[0111] According to an embodiment of the present application, the prediction result includes at least one of the expected lifespan, performance degradation trend, and failure risk classification of the hard disk to be predicted.

[0112] According to an embodiment of the present application, the hard disk prediction device 400 also includes: a generation module for generating life warning prompt information for the hard disk to be predicted when the expected life is less than the expected life threshold; a reduction module for reducing the processing level of the hard disk to be predicted when the performance degradation trend is declining and the health score of the hard disk to be predicted indicates that the hard disk to be predicted is in an unhealthy state; and a scanning module for performing a risk scan on the hard disk to be predicted when the failure risk is classified as high risk.

[0113] According to an embodiment of the present application, the hard disk prediction device 400 also includes: a determination module, which is used to determine the test data of the hard disk to be predicted in multiple test stages based on the prediction results, so as to use the test data to perform stress testing on the hard disk to be predicted and obtain stress test results.

[0114] According to an embodiment of the present application, the hard disk prediction device 400 also includes: a selection module for selecting a target test stage and target test data of the target test stage from multiple test stages based on the prediction results, wherein the expected life and performance of the hard disk to be predicted in the target test stage are lower than those in other test stages, so as to perform a target stress test on the hard disk to be predicted according to the target test data to obtain a target stress test result.

[0115] According to an embodiment of the present application, the test data includes a data packet for simulating errors, and the data packet is used to trigger an error correction operation of the hard disk to be predicted and record the error correction recovery time of the hard disk to be predicted.

[0116] According to an embodiment of the present application, any multiple modules among the acquisition module 410, the first acquisition module 420, the second acquisition module 430, and the third acquisition module 440 can be combined into a single module, or any one of these modules can be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules can be combined with at least part of the functionality of other modules and implemented in a single module. According to an embodiment of the present application, at least one of the acquisition module 410, the first acquisition module 420, the second acquisition module 430, and the third acquisition module 440 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented in hardware or firmware through any other reasonable means of circuit integration or packaging, or can be implemented in any one of the three implementation methods of software, hardware, and firmware, or any appropriate combination of any of these. Alternatively, at least one of the acquisition module 410 , the first obtaining module 420 , the second obtaining module 430 and the third obtaining module 440 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0117] Based on the above-mentioned training method of the preset model, this application also provides a model training device. Figure 5 The device is described in detail.

[0118] Figure 5 The figure shows a structural block diagram of a model training device according to an embodiment of the present application.

[0119] like Figure 5As shown, the model training device 500 includes a fourth obtaining module 510 , a fifth obtaining module 520 , a sixth obtaining module 530 and a seventh obtaining module 540 .

[0120] The fourth obtaining module 510 is configured to process the sample usage status data of the sample hard drive using the initial model to obtain a sample health decay moment during the use of the sample hard drive. The sample health decay moment indicates the moment when the health decay degree of the sample hard drive reaches a preset threshold. In one embodiment, the fourth obtaining module 510 can be used to perform operation S310 described above and will not be further described here.

[0121] The fifth obtaining module 520 is used to obtain the sample health score of the sample hard disk based on the sample type and sample usage status data of the sample hard disk using the initial model. In one embodiment, the fifth obtaining module 520 can be used to perform the operation S320 described above, which will not be repeated here.

[0122] The sixth obtaining module 530 is configured to utilize the initial model to obtain a sample prediction result for the sample hard drive based on the sample health decay time, the sample health score, and the sample performance data of the sample hard drive. The sample prediction result represents the predicted health decay trend and performance decay trend of the sample hard drive. In one embodiment, the sixth obtaining module 530 can be used to perform operation S330 described above and will not be further described here.

[0123] The seventh obtaining module 540 is configured to train the initial model based on the target loss function using the sample prediction results and sample labels to obtain a trained target model, wherein the sample labels indicate the actual health degradation trend and performance degradation trend of the sample hard drive. In one embodiment, the seventh obtaining module 540 can be used to perform operation S340 described above and will not be further described here.

[0124] According to an embodiment of the present application, the initial model includes an initial bidirectional long short-term memory network, an initial health score network and an initial reservation network, which is used to train the initial model based on the target loss function using sample prediction results and sample labels to obtain the trained target model. The seventh obtaining module 540 includes: an eighth obtaining unit, which is used to obtain a loss value based on the target loss function according to the sample prediction results and sample labels; a ninth obtaining unit, which is used to adjust the parameters of the initial bidirectional long short-term memory network, the parameters of the initial health score network and the parameters of the initial reservation network based on the loss value until the loss value converges to obtain the trained target model.

[0125] According to an embodiment of the present application, any multiple modules among the fourth obtaining module 510, the fifth obtaining module 520, the sixth obtaining module 530, and the seventh obtaining module 540 may be combined into a single module, or any one of these modules may be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules may be combined with at least part of the functionality of other modules and implemented in a single module. According to an embodiment of the present application, at least one of the fourth obtaining module 510, the fifth obtaining module 520, the sixth obtaining module 530, and the seventh obtaining module 540 may be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or may be implemented in hardware or firmware by any other reasonable means of circuit integration or packaging, or may be implemented in any one of the three implementation methods of software, hardware, and firmware, or any appropriate combination of any of these. Alternatively, at least one of the fourth obtaining module 510 , the fifth obtaining module 520 , the sixth obtaining module 530 and the seventh obtaining module 540 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0126] Figure 6 A block diagram of an electronic device suitable for implementing a hard disk prediction method and a model training method according to an embodiment of the present application is shown.

[0127] like Figure 6 As shown, an electronic device 600 according to an embodiment of the present application includes a processor 601, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 602 or a program loaded from a storage unit 608 into a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present application.

[0128] Various programs and data required for the operation of the electronic device 600 are stored in the RAM 603. The processor 601, ROM 602, and RAM 603 are connected to each other via a bus 604. The processor 601 performs various operations of the method flow according to the embodiment of the present application by executing the programs in the ROM 602 and / or RAM 603. It should be noted that the programs may also be stored in one or more memories other than the ROM 602 and RAM 603. The processor 601 may also perform various operations of the method flow according to the embodiment of the present application by executing the programs stored in the one or more memories.

[0129] According to an embodiment of the present application, electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to bus 604. Electronic device 600 may also include one or more of the following components connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage section 608 including a hard disk; and a communication section 609 including a network interface card such as a LAN card or modem. Communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. Removable media 611, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 610 as needed, so that computer programs read from the removable media can be installed into storage section 608 as needed.

[0130] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently and not be incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiments of this application is implemented.

[0131] According to an embodiment of the present application, a computer-readable storage medium may be a non-volatile computer-readable storage medium, and may include, for example, but not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present application, a computer-readable storage medium may include the ROM 602 and / or RAM 603 described above and / or one or more memories other than ROM 602 and RAM 603.

[0132] The embodiments of the present application also include a computer program product, which includes a computer program containing program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the method provided in the embodiments of the present application.

[0133] The computer program executes the above functions defined in the system / device of the embodiment of the present application when the computer program is executed by the processor 601. According to the embodiment of the present application, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0134] In one embodiment, the computer program may be stored on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may be transmitted and distributed in the form of a signal on a network medium, downloaded and installed via the communication portion 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0135] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from a removable medium 611. When the computer program is executed by the processor 601, the above-mentioned functions defined in the system of the embodiment of the present application are performed. According to the embodiment of the present application, the systems, devices, means, modules, units, etc. described above can be implemented by computer program modules.

[0136] According to an embodiment of the present application, the program code for executing the computer program provided by the embodiment of the present application can be written in any combination of one or more programming languages. Specifically, these computer programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect via the Internet).

[0137] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of the boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0138] Those skilled in the art will appreciate that the features described in the various embodiments of this application may be combined and / or coupled in various ways, even if such combinations or couplings are not explicitly described in this application. In particular, the features described in the various embodiments of this application may be combined and / or coupled in various ways without departing from the spirit and teachings of this application. All such combinations and / or couplings fall within the scope of this application.

[0139] The embodiments of the present application have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present application. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be advantageously used in combination. Without departing from the scope of the present application, those skilled in the art may make various substitutions and modifications, and these substitutions and modifications should all fall within the scope of the present application.

Claims

1. A hard disk prediction method, characterized in that: The method comprises: Obtain the type, performance data, and usage status data of the hard disk to be predicted; Processing the usage status data to obtain a health degradation moment of the hard disk to be predicted during use, wherein the health degradation moment indicates a moment when the health degradation degree of the hard disk to be predicted reaches a preset threshold; Obtaining a health score of the hard disk to be predicted based on the type of the hard disk to be predicted and the usage status data; Obtaining a prediction result of the hard disk to be predicted based on the health decay time, the health score, and the performance data, wherein the prediction result includes at least one of an expected lifespan, a performance decay trend, and a failure risk classification of the hard disk to be predicted; Obtaining a prediction result of the hard disk to be predicted based on the health decay time, the health score, and the performance data includes: determining, based on the health decay time, the health score, and the performance data, a degree of correlation between the health decay time, the health score, and the performance data; The prediction result of the hard disk to be predicted is obtained by using a predetermined network according to the correlation between the health decay time, the health score and the performance data.

2. The method according to claim 1, characterized in that The processing of the usage status data to obtain the predicted health degradation moment of the hard disk during use includes: Processing the usage status data using a bidirectional long short-term memory network to obtain forward time series information and reverse time series information based on the usage status data; fusing the forward time series information and the reverse time series information to obtain fused time series information; According to the fused timing information, an attention mechanism is used to obtain the health decay moment of the hard disk to be predicted during use.

3. The method according to claim 1, characterized in that Obtaining a health score of the hard disk to be predicted based on the type of the hard disk to be predicted and the usage status data includes: Obtaining an arrangement order of target indicators related to the hard disk to be predicted based on the type of the hard disk to be predicted and the usage status data, wherein the arrangement order represents the degree of influence of the target indicators on the health of the hard disk to be predicted; According to the arrangement order of the target indicators, the health score of the hard disk to be predicted is obtained using a health score network.

4. The method according to claim 3, characterized in that The target indicator includes at least one of a read / write cycle, an uncorrectable bit error rate, an error correction count, and a temperature of the hard disk to be predicted.

5. The method according to claim 1, wherein The method further comprises: When the expected life span is less than the expected life span threshold, generating life span warning prompt information of the hard disk to be predicted; When the performance degradation trend is decreasing and the health score of the hard disk to be predicted indicates that the hard disk to be predicted is in an unhealthy state, lowering the processing level of the hard disk to be predicted; When the failure risk is classified as high risk, a risk scan is performed on the hard disk to be predicted.

6. The method according to claim 1, characterized in that The method further comprises: According to the prediction result, test data of the hard disk to be predicted in multiple test phases are determined, so as to perform a stress test on the hard disk to be predicted using the test data to obtain a stress test result.

7. The method according to claim 6, characterized in that The method further comprises: According to the prediction results, a target test stage and target test data of the target test stage are selected from the multiple test stages, wherein the expected life and performance of the hard disk to be predicted in the target test stage are lower than those in other test stages, so as to perform a target stress test on the hard disk to be predicted according to the target test data to obtain a target stress test result.

8. The method according to claim 6, characterized in that The test data includes a data packet for simulating errors, and the data packet is used to trigger an error correction operation of the hard disk to be predicted and record the error correction recovery time of the hard disk to be predicted.

9. A model training method, characterized in that: The method comprises: Using the initial model to process sample usage status data of the sample hard disk, a sample health decay moment of the sample hard disk during use is obtained, wherein the sample health decay moment indicates the moment when the health decay degree of the sample hard disk reaches a preset threshold; Obtaining a sample health score of the sample hard disk using the initial model according to the sample type of the sample hard disk and the sample usage status data; Determining, using the initial model, a degree of correlation between the sample health decay time, the sample health score, and the sample performance data of the sample hard disk based on the sample health decay time, the sample health score, and the sample performance data of the sample hard disk; Obtaining a sample prediction result of the sample hard disk according to the correlation between the sample health decay time, the sample health score, and the sample performance data of the sample hard disk, wherein the sample prediction result represents a health decay prediction trend and a performance decay prediction trend of the sample hard disk; Based on the target loss function, the initial model is trained using the sample prediction results and sample labels to obtain a trained target model, wherein the sample labels indicate the actual health degradation trend and performance degradation trend of the sample hard disk.

10. The method according to claim 9, characterized in that The initial model includes an initial bidirectional long short-term memory network, an initial health score network, and an initial predetermined network. The initial model is trained based on the target loss function using the sample prediction results and sample labels to obtain a trained target model, including: Based on the target loss function, a loss value is obtained according to the sample prediction result and the sample label; Based on the loss value, parameters of the initial bidirectional long short-term memory network, parameters of the initial health score network, and parameters of the initial predetermined network are adjusted until the loss value converges to obtain a trained target model.

11. A hard disk prediction device, characterized in that: The device comprises: An acquisition module is used to obtain the type, performance data and usage status data of the hard disk to be predicted; A first obtaining module is configured to process the usage status data to obtain a health decay moment of the hard disk to be predicted during use, wherein the health decay moment indicates a moment when the health decay degree of the hard disk to be predicted reaches a preset threshold; A second obtaining module is configured to obtain a health score of the hard disk to be predicted based on the type of the hard disk to be predicted and the usage status data; a third obtaining module, configured to obtain a prediction result of the hard disk to be predicted based on the health decay time, the health score, and the performance data, wherein the prediction result includes at least one of an expected lifespan, a performance decay trend, and a failure risk classification of the hard disk to be predicted; The third module includes: a first obtaining unit, configured to determine a correlation degree between the health decay time, the health score, and the performance data according to the health decay time, the health score, and the performance data; The second obtaining unit is configured to obtain the prediction result of the hard disk to be predicted by using a predetermined network according to the health decay time, the health score, and the correlation degree between the performance data.

12. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images