A Failure Prediction Method for Solid State Hard Drive

Abstract

The present invention relates to a fault prediction method for a solid state hard disk, and relates to the field of hard disk life prediction, including: obtaining a reallocated sector count to correct multiple erase balance parameters, and performing erase balance change analysis to obtain erase balance change parameters; predicting the fault time according to the erase balance change parameters and the latest erase characteristic diagram to obtain the remaining fault time; calculating and obtaining the amplified write wear coefficient according to multiple reallocation amplification coefficients, multiple write amplification coefficients and multiple ordinary write coefficients, correcting the remaining fault time, and obtaining the corrected remaining fault time as a fault prediction result. This application can solve the technical problem that the traditional solid state hard disk life prediction method fails to fully consider the dynamic use and wear of the solid state hard disk, and has insufficient life prediction accuracy and reliability; it can significantly improve the precision, accuracy and reliability of the hard disk remaining life prediction, thereby timely identifying potential fault risks.

Description

Technical Field

[0001] The present invention relates to the field of hard disk life prediction, and in particular to a fault prediction method for a solid state hard disk. Background Art

[0002] With the continuous development and widespread application of solid-state drive (SSD) technology, SSD as a high-performance storage device has gradually replaced the important position of traditional mechanical hard disks in personal computers, servers, data centers and enterprise-level storage systems. SSD has the advantages of fast data reading and writing speed, low power consumption, high stability, etc., so it is widely used in various scenarios that require high-performance storage.

[0003] At present, traditional SSD life prediction methods are mainly based on static data and a single health monitoring indicator, which is difficult to fully reflect the complex situations encountered by the hard disk during actual use, and there are technical problems such as insufficient accuracy and reliability in life prediction. Summary of the invention

[0004] The present invention aims to solve the technical problem that traditional solid-state hard disk life prediction methods fail to fully consider the dynamic usage and wear conditions of the solid-state hard disk, and have insufficient accuracy and reliability in life prediction, and provides a solid-state hard disk fault prediction method to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] In a first aspect, the present invention provides a fault prediction method for a solid-state hard disk, comprising: obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in a historical time, constructing multiple erase characteristic maps, performing erase balance analysis, and obtaining multiple erase balance parameters; obtaining a reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correcting the multiple erase balance parameters, obtaining multiple corrected erase balance parameters, performing erase balance change analysis, and obtaining erase balance change parameters; predicting the fault time from the use of the solid-state hard disk to fault damage based on the erase balance change parameters and the latest erase characteristic map, and obtaining the remaining fault time; obtaining multiple write amplification coefficients of multiple write amplifications and multiple ordinary writes of the solid-state hard disk in a historical time, configuring multiple reallocation amplification coefficients based on the reallocation sector count, calculating and obtaining an amplified write wear coefficient based on the multiple reallocation amplification coefficients, the multiple write amplification coefficients, and the multiple ordinary write coefficients, correcting the remaining fault time, and obtaining the corrected remaining fault time as a fault prediction result.

[0007] In a second aspect, the present invention provides a fault prediction system for a solid-state hard disk, including: an erase balance analysis module, used to obtain the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in a historical time, construct multiple erase feature maps, perform erase balance analysis, and obtain multiple erase balance parameters; an erase balance change analysis module, used to obtain the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correct the multiple erase balance parameters, obtain multiple corrected erase balance parameters, perform erase balance change analysis, and obtain erase balance change parameters; a fault time prediction module, used to predict the number of erase times of multiple blocks recorded by the solid-state hard disk according to the erase feature map; In addition to the balanced change parameters and the latest erase characteristic diagram, the failure time of the solid-state hard disk until it is used to fail is predicted to obtain the remaining time of the failure; the remaining time of the failure correction module is used to obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in the historical time, and according to the reallocated sector count, multiple reallocation amplification coefficients are configured to obtain multiple reallocation amplification coefficients, and according to the multiple reallocation amplification coefficients, the amplified write wear coefficient is calculated to obtain the amplified write wear coefficient, and the remaining time of the failure is corrected to obtain the corrected remaining time of the failure as the failure prediction result.

[0008] In a third aspect, the present invention also provides an electronic device comprising: at least one processor; a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the steps of any one of the methods described in the first aspect above.

[0009] In a fourth aspect, a computer-readable storage medium is provided, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed, the steps of the method described in any one of the first aspects are implemented.

[0010] The beneficial effects of the present invention are as follows: by acquiring the erasure times of multiple blocks recorded by the solid-state hard disk at multiple moments in the historical time, constructing multiple erasure characteristic graphs, performing erase balance analysis, and obtaining multiple erase balance parameters; acquiring the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correcting the multiple erase balance parameters, obtaining multiple corrected erase balance parameters, performing erase balance change analysis, and obtaining erase balance change parameters; predicting the failure time of the solid-state hard disk from being used to failure and damage according to the erase balance change parameters and the latest erase characteristic graph, and obtaining the remaining failure time; acquiring the write amplification time of the solid-state hard disk multiple times in the historical time. According to the reallocated sector count, multiple write amplification factors and multiple ordinary write coefficients of multiple ordinary writes are configured to obtain multiple reallocation amplification factors, and according to the multiple reallocation amplification factors, multiple write amplification factors and multiple ordinary write coefficients, an amplified write wear coefficient is calculated to obtain the amplified write wear coefficient, and the remaining failure time is corrected to obtain the corrected remaining failure time as a failure prediction result; that is, by comprehensively considering the use scenarios and wear characteristics of the solid-state hard disk for life prediction, the precision, accuracy and reliability of the hard disk remaining life prediction can be significantly improved, so that potential failure risks can be identified in time, data loss and system downtime can be avoided, and the stability and security of data storage can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic flow chart of a fault prediction method for a solid state hard disk provided by the present invention;

[0012] Figure 2 A structural schematic diagram of a fault prediction system for a solid state hard disk provided by the present invention;

[0013] Figure 3 A schematic diagram of the structure of an electronic device provided by the present invention;

[0014] Figure 4 A schematic diagram of the structure of a computer-readable storage medium provided by the present invention.

[0015] In the accompanying drawings, the components represented by the reference numerals are described as follows:

[0016] Erase balance analysis module 11, erase balance change analysis module 12, fault time prediction module 13, fault remaining time correction module 14, electronic device 500, memory 510, processor 520, first computer program 511, computer readable storage medium 600, second computer program 611. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0018] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0019] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0020] Embodiment 1, as Figure 1 As shown, an embodiment of the present invention provides a fault prediction method for a solid state drive, which specifically includes the following steps:

[0021] S100: Obtain the erasure counts of multiple blocks recorded by the solid state drive at multiple moments in history, construct multiple erasure feature maps, perform erase balance analysis, and obtain multiple erase balance parameters.

[0022] Furthermore, step S100 of the present application also includes:

[0023] S110: Obtain the number of erasure times of multiple blocks recorded in the solid-state hard disk at multiple moments in the historical time, and obtain multiple sets of erasure times at the multiple moments; S120: Use the number of erasure times in the multiple sets of erasure times as the erasure feature values ​​in the erasure feature map, and combine the positions of the multiple blocks in the solid-state hard disk to construct multiple erasure feature maps at the multiple moments, wherein each erasure feature map includes the number of erasure times of multiple blocks at one moment.

[0024] Specifically, during the use of a solid-state drive (SSD), the controller of the hard drive will record the number of erases of each storage block, where the number of erases of each block reflects the number of times the block has been erased. The life of the solid-state drive is closely related to the number of erases, because each erase will cause wear of the block. Use a hard drive management tool (such as software provided by the SSD manufacturer) or a SMART data interface to obtain the number of erases of multiple blocks recorded by the solid-state drive at multiple moments in the historical time, where the number of erases of multiple blocks is recorded at each moment in the historical time (such as within half a year) (for example, data is recorded by day, week, and month), and obtain multiple sets of erases at the multiple moments.

[0025] First, define an erasure feature map, each erasure feature map represents the erasure status of the solid-state drive at a certain moment, each erasure feature map corresponds to a time point, each pixel in the layer represents the number of erasures of a storage block, and an erasure feature map is formed by listing the position of each block (which can be a physical position or a logical position) and the number of erasures at the position. Then, the number of erasures in the multiple erasure number sets is used as the erasure feature value in the erasure feature map, and the multiple erasure feature maps at multiple moments are constructed in combination with the positions of the multiple blocks in the solid-state drive. For example, the position of each storage block (for example, the block number) is used as the coordinate system of the feature map, and the number of erasures is used as the pixel value of each position. For each moment t, the value in the erasure number set is filled according to the position of the block, and the value of each position is the number of erasures of the corresponding storage block. In this way, the erasure number set at each moment constitutes a two-dimensional feature map, and each element in the image represents the number of erasures of the corresponding block; wherein each erasure feature map includes the number of erasures of multiple blocks at a moment. By constructing an erase feature map, support is provided for the next step of erase balance analysis of the solid-state drive, while the convenience and efficiency of erase balance analysis can be improved.

[0026] S130: Performing erasure equalization analysis on the multiple erasure feature maps to obtain multiple erasure equalization parameters.

[0027] Furthermore, step S130 of the present application also includes:

[0028] S131: In the first erasure feature map at the first moment in the multiple erasure feature maps, select the first erasure feature value, and randomly extract multiple random erasure feature values; S132: Calculate the difference ratio between the first erasure feature value and the mean of the multiple random erasure feature values ​​as the first eigenvalue equalization parameter; S133: Continue to calculate and obtain multiple eigenvalue equalization parameters of the multiple erasure feature values ​​in the first erasure feature map, and calculate the mean to obtain the first erasure equalization parameter; S134: Continue to perform erasure equalization analysis on the other multiple erasure feature maps to obtain multiple erasure equalization parameters.

[0029] Specifically, an erasure balance analysis is then performed on the multiple erasure feature maps. First, any moment is randomly selected from the multiple erasure feature maps as the first moment, and the first erasure feature map of the first moment is obtained; then, the erasure feature value of any block is randomly selected from the first erasure feature map and set as the first erasure feature value, and a preset number of multiple random erasure feature values ​​other than the first erasure feature value are randomly selected from the first erasure feature map. The preset number can be set according to the actual scenario. For example, the preset number is set to 30% of the total number of blocks. Assuming the total number of blocks is 100, 30 erasure feature values ​​other than the first erasure feature value are randomly selected from the first erasure feature map and set as random erasure feature values.

[0030] Then, the mean of the multiple random erased feature values ​​is calculated to obtain the random erased feature mean, and the difference ratio between the first erased feature value and the random erased feature mean is calculated, wherein the difference ratio is the ratio of the feature difference of the first erased feature value minus the random erased feature mean to the random erased feature mean, and the difference ratio is set as the first feature value equalization parameter. Then, the same method is used to continue to calculate and obtain multiple feature value equalization parameters of multiple erased feature values ​​in the first erased feature map, that is, the feature value equalization parameters of all erased feature values ​​in the first erased feature map are calculated, and the mean of the multiple feature value equalization parameters is calculated to obtain the first erased equalization parameter. Further erased feature maps are erased and analyzed to obtain multiple erased equalization parameters, wherein the erased equalization parameters are used to reflect the erased balance of the solid state drive at different times.

[0031] S200: Obtaining the reallocated sector count of the solid state drive through the SMART module of the solid state drive, correcting the multiple erase balance parameters to obtain multiple corrected erase balance parameters, performing erase balance change analysis to obtain erase balance change parameters.

[0032] Furthermore, step S200 of the present application also includes:

[0033] S210: Obtain the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk; S220: Obtain the cumulative usage time of the solid-state hard disk, and obtain the average reallocated sector count of other similar solid-state hard disks when the cumulative usage time reaches the cumulative usage time; S230: Calculate the ratio of the reallocated sector count and the average reallocated sector count, multiply it by the multiple erase balance parameters for correction calculation, and obtain multiple corrected erase balance parameters.

[0034] Specifically, in a solid-state drive (SSD), the reallocated sector count is a SMART data item that reflects the number of times bad blocks (unusable storage blocks) have occurred in the hard disk and have been replaced with spare blocks. The reallocated sector count is one of the important indicators of the health status of the SSD, and is used to detect whether the SSD has damaged bad blocks and their replacement status. Next, the reallocated sector count of the solid-state drive is obtained through the SMART module of the solid-state drive. The reallocated sector count indicates the number of sectors in the hard disk that have been marked as bad blocks. These bad blocks are replaced by spare blocks. Whenever the SSD detects that a storage sector cannot be read or written correctly, it will migrate the data to the spare block and update the status of the area. The more reallocated sector counts there are, the more uneven the erase wear in the solid-state drive is, resulting in damage to some sectors.

[0035] Next, the accumulated usage time of the solid-state drive (SSD) is obtained through the SMART data of the solid-state drive (SSD). The accumulated boot time represents the total running time of the hard drive since it was enabled, in hours, such as 300 hours. Further, multiple reallocated sector counts of multiple similar solid-state drives when the accumulated usage time is reached are calculated, and the average of the multiple reallocated sector counts is calculated to obtain the average reallocated sector count. The ratio of the reallocated sector count to the average reallocated sector count is further calculated, and the ratio is multiplied by the multiple erase balance parameters for correction calculation, and the product of the ratio and the erase balance parameter is used as the corrected erase balance parameter to obtain multiple corrected erase balance parameters.

[0036] S240: Performing erasure equalization change analysis according to the multiple modified erasure equalization parameters to obtain erasure equalization change parameters.

[0037] Furthermore, step S240 of the present application also includes:

[0038] S241: Calculate the variance of the plurality of modified erasure equalization parameters as an erasure equalization variation parameter.

[0039] Specifically, the variance of the multiple corrected erase balance parameters is then calculated, and the variance calculation result is used as the erase balance change parameter, wherein the smaller the erase balance change parameter is, the more stable the erase balance is, the more uniform the wear of the hard disk is, and the lower the risk of failure is; the larger the erase balance change parameter is, the more uneven the wear or local damage may be on the hard disk at certain moments, and the higher the risk of failure is.

[0040] S300: According to the erase balance variation parameter and the latest erase characteristic diagram, the failure time of the solid state hard disk from use to failure is predicted to obtain the remaining failure time.

[0041] Furthermore, step S300 of the present application also includes:

[0042] S310: Based on the historical usage data of similar solid-state hard disks, process and obtain a set of sample erase balance change parameters and a set of sample erase feature maps, and collect the time from the continued use of solid-state hard disks with different sample erase balance change parameters and sample erase feature maps until they are damaged, to obtain a set of sample fault remaining time; S320: Construct a network structure of a fault prediction channel, use the set of sample erase balance change parameters and the set of sample erase feature maps as input supervised training data, use the set of sample fault remaining time as output supervised training data, and perform supervised training on the fault prediction channel until convergence; S330: Input the erase balance change parameters and the latest erase feature map into the fault prediction channel, and predict the output to obtain the remaining time of the fault.

[0043] Specifically, based on the historical usage data of similar solid-state hard disks, the erase balance change parameters of multiple similar solid-state hard disks at different time points are calculated (obtained by calculating the variance), and the erase count data of the hard disk at multiple times (such as the erase count of each block) are collected to construct an erase feature map; a sample erase balance change parameter set and a sample erase feature map set are obtained. Next, the time from the solid-state hard disks with different sample erase balance change parameters and sample erase feature maps to the time of failure damage is set as the sample failure remaining time, and a sample failure remaining time set is obtained.

[0044] The network structure of the fault prediction channel is constructed based on the BP neural network. The fault prediction channel is a BP neural network model that can be iteratively optimized in machine learning and is obtained through supervised training with sample data. The fault prediction channel includes an input layer, a hidden layer and an output layer, wherein the input data of the input layer is the erasure balance change parameter and the erasure feature map, and the output data of the output layer is the fault remaining time; further, the sample erasure balance change parameter set and the sample erasure feature map set are used as input supervised training data, and the sample fault remaining time set is used as output supervised training data to perform supervised training on the fault prediction channel. During training, the input data is first passed to the hidden layer through the input layer, and the neurons in each hidden layer receive the input data and calculate their The weighted sum is processed through the activation function to obtain the output, and the output of the hidden layer is passed to the output layer for processing. The output layer nodes calculate the final prediction value, that is, the remaining time of the fault; then, by comparing the predicted value of the network with the actual remaining time of the fault, the error is calculated, and the mean square error (MSE) is usually used as the loss function; then the partial derivative of the loss function with respect to each weight and bias is calculated to obtain the gradient, and the weight and bias are updated according to the gradient; by repeating the above forward propagation and back propagation processes, the weights and biases of the network are continuously adjusted until the network obtains a sufficiently small error on the training set until the predetermined convergence conditions are met, wherein the training process can be terminated by setting stop conditions such as the maximum number of iterations and the error threshold to obtain a trained fault prediction channel.

[0045] The latest erasure characteristic map of the solid state drive is obtained, that is, the erasure count data is extracted from the controller or monitoring system of the solid state drive to construct an erasure characteristic map; and the erasure balance change parameter and the latest erasure characteristic map are input into the fault prediction channel, and the prediction output is used to obtain the remaining time of the fault. By constructing a fault prediction channel based on the BP neural network to predict the remaining time of the fault, the accuracy, reliability and precision of the prediction of the remaining life of the solid state drive can be significantly improved.

[0046] S400: Obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in the historical time, configure and obtain multiple reallocation amplification coefficients according to the reallocation sector count, calculate and obtain the amplified write wear coefficient according to the multiple reallocation amplification coefficients, the multiple write amplification coefficients and the multiple ordinary write coefficients, correct the remaining fault time, and obtain the corrected remaining fault time as a fault prediction result.

[0047] Furthermore, step S400 of the present application also includes:

[0048] S410: Obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid state drive in the historical time, wherein each ordinary write coefficient is 1; S420: According to the reallocated sector count, configure to obtain multiple reallocation amplification coefficients, wherein each reallocation amplification coefficient is 2; S430: Integrate the multiple write amplification coefficients, multiple ordinary write coefficients and multiple reallocation amplification coefficients to calculate the amplified write wear coefficient, as shown in the following formula: ; Where K is the amplified write wear factor, P is the number of multiple write amplification factors, Q is the number of multiple normal write factors, M is the reallocated sector count, is the i-th write amplification factor, is the jth common write coefficient, is the lth reallocation amplification coefficient; S440: the fault remaining time is divided by the amplified write wear coefficient to complete the correction calculation and obtain the corrected fault remaining time as the fault prediction result.

[0049] Specifically, first, a plurality of write amplification coefficients of multiple write amplifications and a plurality of ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in a historical time are obtained, wherein each ordinary write coefficient is 1; wherein, write amplification refers to the ratio between the amount actually written by the solid-state hard disk when writing data and the amount requested to be written by the user. Write amplification usually occurs when data is updated, and multiple pages need to be erased and rewritten, resulting in the actual write amount exceeding the original requested write amount. For example, in a solid-state hard disk, data can be written by page, but erasure can only be performed in blocks (usually a block contains multiple pages). Therefore, if the data of a page is to be updated, The entire block of data needs to be read out, modified, and then rewritten. For example, if 100G of data needs to be updated, but the previous 50G of data needs to be erased and rewritten, resulting in 150G of data being written. This is write amplification, and the write amplification factor is 150 / 100=1.5. Among them, the use scenario of hard disk write amplification will lead to an increase in write resources and an increase in the number of erases, thereby reducing the service life; normal write refers to the ratio of the actual amount of data written to the requested amount of data in each write operation without additional write amplification. For normal write operations, this factor is usually 1, which means that the requested amount is written, without additional write or amplification.

[0050] Next, according to the reallocated sector count, multiple reallocation amplification factors are configured, where the reallocation amplification factor represents the additional writing caused by the sector reallocation operation. Assuming that the data of the bad block needs to be migrated to the spare block each time reallocation occurs, this can be regarded as "rewriting" the data, so the reallocation amplification factor is usually set to 2.

[0051] Then construct the amplified write wear coefficient calculation function: In the write wear coefficient calculation function, K is the amplified write wear coefficient, P is the number of multiple write amplification coefficients, Q is the number of multiple ordinary write coefficients, and M is the reallocated sector count. is the i-th write amplification factor, is the jth common write coefficient, is the lth redistribution amplification factor; the amplified write wear factor (K) is calculated by considering the combined effects of write amplification, redistribution, and ordinary write. The calculation method is to perform a weighted summation of all write coefficients and then compare it with the sum of the write coefficients without write amplification. The final ratio is used to correct the remaining time of the fault. For example, assuming that there are 10 write amplification factors (each coefficient is 1.5), 5 redistribution amplification factors (each coefficient is 2), and 85 ordinary write coefficients (each coefficient is 1), the amplified write wear factor is (10 1.5+5 2+85) / 10+5+85=1.1.

[0052] Finally, the remaining time to failure is divided by the amplified write wear coefficient to complete the correction calculation, and the calculation result is used as the corrected remaining time to failure. For example, assuming that the remaining time to failure is 110 hours and the amplified write wear coefficient is 1.1, the corrected remaining time to failure is 100 hours, and the corrected remaining time to failure is used as the failure prediction result. The remaining time to failure is corrected by calculating the amplified write wear coefficient, where the amplified write wear coefficient takes into account the effects of write amplification, redistribution and ordinary writing during the use of the hard disk, which makes the prediction result closer to the actual use of the hard disk, thereby greatly improving the accuracy and reliability of the life prediction of the solid-state hard disk.

[0053] The embodiment of the present invention provides a method for predicting a fault of a solid state drive, which has at least the following technical effects:

[0054] By obtaining the erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in the historical time, multiple erase characteristic graphs are constructed, and erase balance analysis is performed to obtain multiple erase balance parameters; through the SMART module of the solid-state hard disk, the reallocated sector count of the solid-state hard disk is obtained, the multiple erase balance parameters are corrected to obtain multiple corrected erase balance parameters, and erase balance change analysis is performed to obtain erase balance change parameters; according to the erase balance change parameters and the latest erase characteristic graph, the failure time of the solid-state hard disk from being used to failure is predicted to obtain the remaining failure time; multiple write amplifications of the solid-state hard disk during multiple writes in the historical time are obtained. The amplification factor and multiple ordinary write coefficients for multiple ordinary writes are configured according to the reallocated sector count, and the amplified write wear coefficient is calculated according to the multiple reallocated amplification factors, the multiple write amplification factors and the multiple ordinary write coefficients. The remaining time of the fault is corrected to obtain the corrected remaining time of the fault as a fault prediction result; that is, by comprehensively considering the use scenarios and wear characteristics of the solid-state hard disk for life prediction, the precision, accuracy and reliability of the hard disk remaining life prediction can be significantly improved, so that potential failure risks can be identified in time, data loss and system downtime can be avoided, and the stability and security of data storage can be ensured.

[0055] Embodiment 2, as Figure 2 As shown, based on the same inventive concept of a solid state drive fault prediction method provided in Embodiment 1, an embodiment of the present invention further provides a solid state drive fault prediction system, including:

[0056] The erase balance analysis module 11 is used to obtain the number of erases of multiple blocks recorded by the solid-state hard disk at multiple moments in the historical time, construct multiple erase characteristic maps, perform erase balance analysis, and obtain multiple erase balance parameters; the erase balance change analysis module 12 is used to obtain the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correct the multiple erase balance parameters, obtain multiple corrected erase balance parameters, perform erase balance change analysis, and obtain erase balance change parameters; the failure time prediction module 13 is used to predict the failure time of the solid-state hard disk from being used to failure damage according to the erase balance change parameters and the latest erase characteristic map, and obtain the remaining failure time; the remaining failure time correction module 14 is used to obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary writes of the solid-state hard disk in the historical time, configure and obtain multiple reallocation amplification coefficients according to the reallocation sector count, calculate and obtain the amplified write wear coefficient according to the multiple reallocation amplification coefficients, multiple write amplification coefficients and multiple ordinary write coefficients, correct the remaining failure time, and obtain the corrected remaining failure time as the failure prediction result.

[0057] Furthermore, the fault prediction system for a solid-state hard disk also includes: obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in historical time, and obtaining multiple sets of erase times at the multiple moments; using the number of erase times in the multiple sets of erase times as the erase feature values ​​in the erase feature map, and combining the positions of the multiple blocks in the solid-state hard disk to construct multiple erase feature maps at the multiple moments, wherein each erase feature map includes the number of erase times of multiple blocks at one moment; performing erase balance analysis on the multiple erase feature maps to obtain multiple erase balance parameters.

[0058] Furthermore, the fault prediction system for a solid-state hard disk also includes: selecting a first erasure characteristic value in the first erasure characteristic map at the first moment in the multiple erasure characteristic maps, and randomly extracting multiple random erasure characteristic values; calculating the difference ratio between the first erasure characteristic value and the mean of the multiple random erasure characteristic values ​​as a first characteristic value equalization parameter; continuing to calculate multiple characteristic value equalization parameters of the multiple erasure characteristic values ​​in the first erasure characteristic map, and calculating the mean to obtain a first erasure equalization parameter; continuing to perform erasure equalization analysis on the other multiple erasure characteristic maps to obtain multiple erasure equalization parameters.

[0059] Furthermore, the fault prediction system for a solid-state hard disk also includes: obtaining a reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk; obtaining the cumulative usage time of the solid-state hard disk, and obtaining the average reallocated sector count of other similar solid-state hard disks when the cumulative usage time reaches the cumulative usage time; calculating the ratio of the reallocated sector count to the average reallocated sector count, multiplying it by the multiple erase balance parameters for correction calculation, and obtaining multiple corrected erase balance parameters; performing erase balance change analysis based on the multiple corrected erase balance parameters to obtain erase balance change parameters.

[0060] Furthermore, the fault prediction system for a solid state hard disk further includes: calculating the variance of the multiple modified erase balance parameters as an erase balance variation parameter.

[0061] Furthermore, the fault prediction system for a solid-state hard disk also includes: based on historical usage data of similar solid-state hard disks, processing to obtain a sample erase balance change parameter set and a sample erase feature map set, and collecting the time when solid-state hard disks with different sample erase balance change parameters and sample erase feature maps continue to be used until fault damage, to obtain a sample fault remaining time set; constructing a network structure of a fault prediction channel, using the sample erase balance change parameter set and the sample erase feature map set as input supervised training data, using the sample fault remaining time set as output supervised training data, and performing supervised training on the fault prediction channel until convergence; inputting the erase balance change parameters and the latest erase feature map into the fault prediction channel, and predicting the output to obtain the fault remaining time.

[0062] Furthermore, the fault prediction system of a solid-state hard disk further includes: obtaining multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in a historical time, wherein each ordinary write coefficient is 1; configuring to obtain multiple reallocation amplification coefficients according to the reallocated sector count, wherein each reallocation amplification coefficient is 2; integrating the multiple write amplification coefficients, the multiple ordinary write coefficients and the multiple reallocation amplification coefficients, and calculating to obtain the amplified write wear coefficient, as shown in the following formula: ; Where K is the amplified write wear factor, P is the number of multiple write amplification factors, Q is the number of multiple normal write factors, M is the reallocated sector count, is the i-th write amplification factor, is the jth common write coefficient, is the lth reallocation amplification coefficient; the fault remaining time is divided by the amplified write wear coefficient to complete the correction calculation and obtain the corrected fault remaining time as the fault prediction result.

[0063] For example 3, please refer to Figure 3 , Figure 3 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 3As shown, an embodiment of the present invention provides an electronic device 500, including a memory 510, a processor 520, and a first computer program 511 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the first computer program 511, the following steps are implemented: obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in a historical time, constructing multiple erase feature maps, performing erase balance analysis, and obtaining multiple erase balance parameters; obtaining the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correcting the multiple erase balance parameters, obtaining multiple corrected erase balance parameters, and performing erase balance analysis. The erase balance change analysis is performed to obtain the erase balance change parameters; based on the erase balance change parameters and the latest erase characteristic diagram, the failure time of the solid state drive until it is damaged is predicted to obtain the remaining failure time; multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid state drive in the historical time are obtained, and multiple reallocation amplification coefficients are configured according to the reallocated sector count; based on the multiple reallocation amplification coefficients, the multiple write amplification coefficients and the multiple ordinary write coefficients, the amplified write wear coefficient is calculated, the remaining failure time is corrected, and the corrected remaining failure time is obtained as the failure prediction result.

[0064] Example 4, please refer to Figure 4 , Figure 4 A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 4 As shown, this embodiment provides a computer-readable storage medium 600, on which a second computer program 611 is stored. When the second computer program 611 is executed by a processor, the following steps are implemented: obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in the historical time, constructing multiple erase feature maps, performing erase balance analysis, and obtaining multiple erase balance parameters; obtaining the reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correcting the multiple erase balance parameters, obtaining multiple corrected erase balance parameters, performing erase balance change analysis, and obtaining erase balance change Parameters; predict the failure time of the solid-state hard disk from being used to failure and damage according to the erase balance change parameters and the latest erase characteristic diagram, and obtain the remaining failure time; obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in the historical time, and configure and obtain multiple reallocation amplification coefficients according to the reallocated sector count; calculate and obtain the amplified write wear coefficient according to the multiple reallocation amplification coefficients, multiple write amplification coefficients and multiple ordinary write coefficients, correct the remaining failure time, and obtain the corrected remaining failure time as the failure prediction result.

[0065] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0066] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0067] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0068] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0070] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0071] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. A method for predicting a fault of a solid state drive, characterized in that: include: Obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in history, constructing multiple erase feature graphs, performing erase balance analysis, and obtaining multiple erase balance parameters; Obtaining a reallocated sector count of the solid-state hard disk through the SMART module of the solid-state hard disk, correcting the multiple erase balance parameters to obtain multiple corrected erase balance parameters, performing erase balance change analysis to obtain erase balance change parameters; According to the erase balance variation parameter and the latest erase characteristic graph, the failure time of the solid state drive from use to failure is predicted to obtain the remaining failure time, including: According to the historical usage data of similar solid-state hard disks, a sample erase balance change parameter set and a sample erase feature map set are processed and obtained, and the time when solid-state hard disks with different sample erase balance change parameters and sample erase feature maps are continuously used until they fail to damage is collected to obtain a sample failure remaining time set; Constructing a network structure of a fault prediction channel, using the sample erasure balance change parameter set and the sample erasure feature map set as input supervised training data, using the sample fault remaining time set as output supervised training data, and performing supervised training on the fault prediction channel until convergence; Input the erasure balance variation parameter and the latest erasure characteristic map into the fault prediction channel, and output the prediction to obtain the remaining time of the fault; Acquiring multiple write amplification coefficients of multiple write amplifications and multiple common write coefficients of multiple common writes of the solid-state hard disk in a historical time, configuring and obtaining multiple reallocation amplification coefficients according to the reallocation sector count, calculating and obtaining an amplified write wear coefficient according to the multiple reallocation amplification coefficients, the multiple write amplification coefficients and the multiple common write coefficients, correcting the remaining time of the fault, and obtaining a corrected remaining time of the fault as a fault prediction result, including: Acquire multiple write amplification coefficients of multiple write amplifications and multiple common write coefficients of multiple common writes of the solid state drive in a historical period, wherein each common write coefficient is 1; According to the reallocated sector count, a plurality of reallocation magnification factors are configured, wherein each reallocation magnification factor is 2; The multiple write amplification coefficients, multiple common write coefficients and multiple redistribution amplification coefficients are integrated to calculate the amplified write wear coefficient, as shown in the following formula: ； Where K is the amplified write wear factor, P is the number of multiple write amplification factors, Q is the number of multiple normal write factors, M is the reallocated sector count, is the i-th write amplification factor, is the jth common write coefficient, is the lth redistribution magnification factor; The fault remaining time is divided by the amplified write wear coefficient to complete the correction calculation and obtain the corrected fault remaining time as the fault prediction result.

2. A method for predicting a fault of a solid state drive according to claim 1, characterized in that: Obtaining the number of erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in history, constructing multiple erase feature graphs, performing erase balance analysis, and obtaining multiple erase balance parameters, including: Obtaining the number of times of erasure of multiple blocks recorded at multiple moments in the solid state drive during historical time, and obtaining multiple sets of erasure numbers at the multiple moments; Using the erasure times in the multiple erasure times set as the erasure feature values ​​in the erasure feature map, and combining the positions of the multiple blocks in the solid-state hard disk, constructing the multiple erasure feature maps at the multiple moments, wherein each erasure feature map includes the erasure times of the multiple blocks at a moment; Performing erasure equalization analysis on the multiple erasure feature maps to obtain multiple erasure equalization parameters.

3. A method for predicting a fault of a solid state drive according to claim 2, characterized in that: Performing erasure equalization analysis on the multiple erasure feature maps to obtain multiple erasure equalization parameters, including: In the first erasure feature map at the first moment in the multiple erasure feature maps, select a first erasure feature value, and randomly extract a plurality of random erasure feature values; Calculating a difference ratio between the first erasure eigenvalue and the average of the plurality of random erasure eigenvalues ​​as a first eigenvalue equalization parameter; Continue to calculate and obtain multiple eigenvalue equalization parameters of multiple erasure eigenvalues ​​in the first erasure feature map, and calculate the average to obtain a first erasure equalization parameter; Continue to perform erasure equalization analysis on other multiple erasure feature maps to obtain multiple erasure equalization parameters.

4. The method for predicting a fault of a solid state drive according to claim 1, wherein: Obtaining the reallocated sector count of the solid state drive through the SMART module of the solid state drive, correcting the multiple erase balance parameters to obtain multiple corrected erase balance parameters, performing erase balance change analysis to obtain erase balance change parameters, including: Obtaining a reallocated sector count of the solid state drive through a SMART module of the solid state drive; Obtaining the accumulated usage time of the solid state drive, and obtaining the average reallocated sector count of other similar solid state drives when the accumulated usage time reaches the accumulated usage time; Calculating a ratio of the reallocated sector count to the average reallocated sector count, and multiplying the ratio by the multiple erasure equalization parameters to perform correction calculation to obtain multiple corrected erasure equalization parameters; An erasure equalization variation analysis is performed according to the multiple modified erasure equalization parameters to obtain an erasure equalization variation parameter.

5. A method for predicting a fault of a solid state drive according to claim 4, characterized in that: Performing an erasure equalization change analysis according to the multiple modified erasure equalization parameters to obtain an erasure equalization change parameter includes: The variance of the plurality of modified erasure equalization parameters is calculated as an erasure equalization variation parameter.

6. A fault prediction system for a solid state drive, characterized in that: The steps for implementing the fault prediction method of a solid state drive according to any one of claims 1 to 5 include: An erase balance analysis module is used to obtain the erase times of multiple blocks recorded by the solid-state hard disk at multiple moments in history, construct multiple erase feature graphs, perform erase balance analysis, and obtain multiple erase balance parameters; An erase balance change analysis module, used to obtain the reallocated sector count of the solid state drive through the SMART module of the solid state drive, correct the multiple erase balance parameters, obtain multiple corrected erase balance parameters, perform erase balance change analysis, and obtain erase balance change parameters; A failure time prediction module is used to predict the failure time of the solid state drive from use to failure and damage according to the erase balance change parameter and the latest erase characteristic map, and obtain the remaining failure time; The fault remaining time correction module is used to obtain multiple write amplification coefficients of multiple write amplifications and multiple ordinary write coefficients of multiple ordinary writes of the solid-state hard disk in the historical time, configure multiple reallocation amplification coefficients according to the reallocated sector count, calculate and obtain the amplified write wear coefficient according to the multiple reallocation amplification coefficients, multiple write amplification coefficients and multiple ordinary write coefficients, correct the fault remaining time, and obtain the corrected fault remaining time as the fault prediction result.

7. An electronic device, characterized in that: include: Memory for storing computer software programs; A processor is used to read and execute the computer software program, thereby implementing the steps of a fault prediction method for a solid state drive as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium, characterized in that: The storage medium stores a computer software program, and when the computer software program is executed by the processor, the steps of a fault prediction method for a solid state drive as described in any one of claims 1 to 5 are implemented.

Images