A solid state hard disk control method, device and solid state hard disk system

Abstract

The present invention discloses a control method and device for a solid-state hard disk and a solid-state hard disk system, and relates to the technical field of solid-state hard disk control. The method comprises obtaining a real-time operation vector according to real-time read and write data corresponding to each storage unit, obtaining the trend distance between the real-time operation vector of each storage unit and a boundary vector, marking the unit that is about to fail according to the trend distance between the real-time operation vector corresponding to each storage unit and the boundary vector, collecting and analyzing the historical fault data of each storage unit in the solid-state hard disk, combining the operation time with the correlation coefficient of each type of data, determining the boundary vector of the storage unit, and marking the unit that is about to fail according to the trend distance between the real-time operation vector and the boundary vector, so as to take measures in advance, such as data migration or replacement of storage units, to avoid data loss and damage, and improve the reliability and data security of the solid-state hard disk.

Description

Technical Field

[0001] The present invention belongs to the technical field of solid state hard disk control, and specifically relates to a solid state hard disk control method, a solid state hard disk control device and a solid state hard disk system. Background Art

[0002] With the development of large-scale integrated circuit technology, the speed and capacity of Flash memory have been greatly improved, making it possible to use Flash memory to replace hard disks to achieve large-capacity data storage. Solid-state hard disk refers to a large-capacity storage device that uses chips as storage media, conforms to interface transmission protocols, and has the same working mode as traditional hard disks. Compared with traditional hard disks, solid-state hard disks have the advantages of light weight, fast data access speed, impact resistance, and no noise due to the lack of mechanical parts.

[0003] Patent publication number CN110275676A discloses a control method, device and solid-state hard disk system for a solid-state hard disk. The present invention calculates the thermal time constant of the solid-state hard disk and calculates the remaining write time of the solid-state hard disk based on the thermal time constant, so that the host can predict when thermal throttling will start based on the remaining write time, so that before starting thermal throttling, the tasks of the solid-state hard disk that starts thermal throttling can be assigned to other solid-state hard disks for processing, thereby maximizing data throughput.

[0004] However, during the use of solid-state hard drives, failure of storage units may lead to data loss and degradation of device performance. Traditional fault prediction methods mainly rely on the monitoring of a single indicator, such as read / write error rate or the number of bad blocks, but these methods often cannot fully reflect the health of storage units, resulting in insufficient accuracy and timeliness of fault prediction. Based on this, a control method, device and solid-state hard drive system are proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a control method, device and solid-state hard disk system for a solid-state hard disk, which solves the technical problem that the failure of a storage unit may lead to data loss and degradation of device performance, and that traditional fault prediction methods mainly rely on the monitoring of a single indicator, resulting in insufficient accuracy and timeliness of fault prediction.

[0006] A control method for a solid state hard disk comprises the following steps:

[0007] Step 1: Collect various types of data from multiple historical fault data of each storage unit in the solid-state hard disk; the historical fault data includes the number of read and write times, read and write error rate, number of erase times, total amount of read and write data, and number of bad blocks corresponding to each storage unit in the solid-state hard disk;

[0008] Step 2: Obtain the boundary values ​​corresponding to each type of data in each storage unit;

[0009] Step 3: Analyze the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time;

[0010] Step 4: According to the relationship coefficient between each type of data and the running time and the boundary value corresponding to each storage unit, obtain the boundary vector corresponding to each storage unit;

[0011] Step 5: obtaining the trend distance between the real-time operation vector of each storage unit and the boundary vector according to the real-time read and write data corresponding to each storage unit;

[0012] Step 6: Mark the failing units according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector.

[0013] As a further solution of the present invention: the specific method of obtaining the limit values ​​corresponding to each type of data in each storage unit is:

[0014] S1: Randomly select one of the storage units without replacement as the target unit;

[0015] S2: Randomly select a category from each category of data without replacement as the target category data, mark the target category data in multiple historical fault data of the target unit as Rr, obtain the number b of values ​​in Rr that meet the preset screening condition X, when the number b is greater than the preset threshold Y1, define the mean Rp of Rr as the boundary value A11 corresponding to the target category data in the target unit, when the number b is less than the preset threshold Y1, define the mean of the maximum and minimum values ​​in Rr as the boundary value A11 corresponding to the target category data in the target unit, where r refers to different historical fault data, and the value is r=1, 2, ..., a, where a refers to the number of historical fault data, a is a positive integer and a>1, a≥b≥1;

[0016] S3: Repeat the above step S2 to obtain the boundary values ​​A11, A21, A31, A41 and A51 corresponding to each type of data in the target unit;

[0017] S4: Repeat the above steps S1-S3 to obtain the boundary values ​​A1j, A2j, A3j, A4j and A5j corresponding to each type of data in each storage unit, where j refers to different storage units, j=1, 2, ..., c, where c refers to the number of data categories, c is a positive integer and c is greater than 1.

[0018] As a further solution of the present invention: the preset screening condition X is: |Rr-Rp|≥Y2, wherein Rp is the mean value of Rr, and Y2 is the preset threshold value.

[0019] As a further solution of the present invention: the specific method of obtaining the relationship coefficient between various types of data and the running time is:

[0020] S01: Randomly select a category from various types of data without replacement as analysis data;

[0021] S02: multiple preset running times Tt are set, one is randomly selected from the multiple preset running times Tt as the analysis time, and the average of the maximum and minimum values ​​of the historical values ​​Ki of the analysis data of each storage unit of multiple solid state drives with the same specifications under the analysis time is used as the calibration value E11 of the analysis data under the analysis time, where i is a different analysis data value;

[0022] S03: Repeat the above step S01 to obtain the calibration values ​​E1t corresponding to the analysis data at different preset running times, where t refers to different preset running times, t=1, 2, ..., e, where e refers to the number of preset running times, e is a positive integer and e is greater than 1; according to the calibration values ​​E1t corresponding to different preset running times and the analysis data at different preset running times, a two-dimensional curve graph of the analysis data with the running time is obtained, and the two-dimensional curve graph is analyzed to obtain the relationship coefficient H1 between the analysis data and the running time;

[0023] S04: Repeat the above steps S01-S03, and then obtain the relationship coefficients H1, H2, H3, H4 and H5 between various types of data and the running time.

[0024] As a further solution of the present invention: the specific method of obtaining the relationship coefficient between the analysis data and the running time is:

[0025] Different preset running time Tt is used as the horizontal coordinate, and the calibration value E1t corresponding to different preset running time is used as the vertical coordinate, so as to obtain the data points Wt (Tt, E1t) corresponding to the analysis data under different preset running time, mark the connecting line between each two adjacent data points as a connecting line Lq, and obtain the slope Kq corresponding to each connecting line according to the coordinates of the two data points that constitute each connecting line. The number of positive and negative values ​​in the slope Kq is analyzed to obtain the relationship coefficient H1 between the analysis data and the running time, where q is a different connecting line, q=1, 2, ..., p, where p refers to the number of connecting lines, p is a positive integer and p is equal to the preset running time number e minus 1.

[0026] As a further solution of the present invention: the specific way of analyzing the number of positive and negative values ​​in the slope Kq is:

[0027] The number of positive and negative values ​​in the slope Kq is marked as U+ and U - , get U + and U - The ratio Up to the number of connecting lines p + and Up - , when Up + When it is greater than or equal to one-half, the mean absolute value of the slope Kq is used as the relationship coefficient H1 between the analysis data and the running time. - When it is greater than or equal to one-half, the product of the absolute value of the mean of the slope Kq and minus one is taken as the relationship coefficient H1 between the analysis data and the running time. + and Up - When both are not greater than or equal to one half, 0 is used as the relationship coefficient H1 between the analytical data and the running time.

[0028] As a further solution of the present invention: the specific method of obtaining the boundary vectors corresponding to each storage unit is:

[0029] The products of the boundary values ​​corresponding to each type of data in each storage unit and the relationship coefficients between each type of data and the running time are obtained, and vectorized processing is performed to obtain the boundary vectors Jj (JA1j, JA2j, JA3j, JA4j, JA5j) corresponding to each storage unit.

[0030] As a further solution of the present invention, the specific method of obtaining the trend distance between the real-time operation vector of each storage unit and the boundary vector is:

[0031] The square root of the sum of the squares of the differences between each data in the operation vector of each storage unit and each data in the boundary vector is taken to obtain the trend distance Qj between the real-time operation vector of each storage unit and the boundary vector.

[0032] As a further solution of the present invention, the specific method of marking the unit that is about to fail is:

[0033] The storage units whose trend distance Qj is greater than the preset value Y3 are marked as failed units.

[0034] A solid state hard disk system, the system implements a control method for a solid state hard disk;

[0035] A data collection module collects various types of data from a plurality of historical fault data of each storage unit in the solid state drive;

[0036] The boundary value acquisition module obtains the boundary values ​​corresponding to each type of data in each storage unit;

[0037] The relationship coefficient acquisition module analyzes the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time;

[0038] A limit vector acquisition module obtains the limit vectors corresponding to each storage unit according to the limit values ​​corresponding to each storage unit;

[0039] A trend distance acquisition module calculates the trend distance between the real-time operation vector of each storage unit and the boundary vector according to the real-time operation vector obtained from the real-time read and write data corresponding to each storage unit;

[0040] The failing unit marking module marks the failing unit according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention collects and analyzes historical failure data of each storage unit in a solid-state hard disk, combines the operating time with the correlation coefficient of each type of data, determines the boundary vector of the storage unit, and marks the unit that is about to fail by the trend distance between the real-time operation vector and the boundary vector. It can effectively predict the failure of each storage unit in the solid-state hard disk, take measures in advance to avoid data loss and equipment damage, such as data migration or replacement of storage units, to avoid data loss and damage, and improve the reliability and data security of the solid-state hard disk. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the framework structure of the method of the present invention;

[0044] Figure 2 It is a schematic diagram of the system framework structure of the present invention. DETAILED DESCRIPTION

[0045] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Example 1: Please refer to Figure 1 , the present application provides a control method for a solid state hard disk, comprising the following steps:

[0047] Step 1: Collect multiple historical fault data of each storage unit in the solid-state hard disk, the historical fault data including the number of read and write times, read and write error rate, number of erase times, total amount of read and write data and number of bad blocks corresponding to each storage unit in the solid-state hard disk;

[0048] Read and write times: the total number of read and write operations of the storage unit when a fault occurs;

[0049] Erasure count: the total number of erase and write operations that a storage unit undergoes when a failure occurs;

[0050] A refined data acquisition module is built at the firmware level of the main control chip of the solid-state drive. The hardware counter resources inside the chip are used to allocate a read / write counter and an erase counter for each storage unit to ensure the accuracy and efficiency of operation counting. Each time a read / write or erase operation is triggered, the hardware counter automatically performs an increment operation, and the data is synchronized to the cache area in real time.

[0051] Read and write error rate: It is the ratio between the number of read and write errors when a storage unit fails and the total number of read and write operations;

[0052] Total amount of read and write data: the total amount of data processed by the storage unit when the storage unit fails;

[0053] In terms of total data volume statistics, a data volume monitoring node is set up on the data transmission path of the storage unit. Digital signal processing technology is used to analyze and accumulate the number of data bytes transmitted in each read and write operation in real time. The data is stored in the cache in the form of an efficient data stream for subsequent processing.

[0054] Number of bad blocks: This is the number of bad blocks in a storage unit when a storage unit fails. The number of bad blocks usually affects read and write performance and data security.

[0055] Bad block detection combines the physical address mapping table at the hardware level with the bad block marking algorithm at the software level. At the hardware level, the physical address space of the storage unit is periodically scanned to identify units that cannot respond normally to read and write operations or have specific electrical anomalies; at the software level, during each read and write operation, if an uncorrectable error is encountered, the corresponding storage unit address is recorded in the bad block marking list, and the bad block count is updated regularly.

[0056] Data such as read and write times, read and write error rates, erase times, total read and write data, and bad block counts can be obtained through SSD management tools, such as SSD controllers and SMART monitoring, or directly through the SSD manufacturer's API interface. The above acquisition methods are existing and mature technologies, so they will not be elaborated here.

[0057] Step 2: Analyze multiple historical fault data of each storage unit, and obtain the boundary values ​​corresponding to each type of data in each storage unit according to the analysis results. The specific method is as follows:

[0058] S1: Randomly select one of the storage units without replacement as the target unit;

[0059] S2: Randomly select one category from each category of data without replacement as the target category data;

[0060] The target class data in the multiple historical fault data of the target unit are marked as Rr respectively; wherein r refers to different historical fault data, and the value is r=1, 2, ..., a, wherein a refers to the number of historical fault data, a is a positive integer and a>1;

[0061] Get the value Rb of each occluded area Rr that meets the preset filtering condition X, where b is the number of values ​​in Rr that meet the preset filtering condition X, a≥b≥1, and compare the number b with the preset threshold Y1. When the number b is greater than the preset threshold Y1, it means that the number of values ​​in Rr that meet the preset filtering condition X is large, and then the mean of Rr is representative, and the mean Rp of Rr is defined as the boundary value A11 corresponding to the target class data in the target unit. When the number b is less than the preset threshold Y1, it means that the number of values ​​in Ri that meet the preset filtering condition X is small, and the mean of Rr is not representative, and then the mean of the maximum and minimum values ​​in Rr is defined as the boundary value A11 corresponding to the target class data in the target unit, that is, H1=(Rin+Rax) / 2, where Rax and Rin are the maximum and minimum values ​​in Rr, respectively. Here, the preset screening condition X is specifically: |Rr-Rp|≥Y2, where Rp is the mean value of Rr, Y2 is the preset threshold, and the specific values ​​of the preset thresholds Y1 and Y2 are formulated by relevant personnel according to needs;

[0062] S3: Repeat the above step S2 to obtain the boundary values ​​A11, A21, A31, A41 and A51 corresponding to each type of data in the target unit;

[0063] S4: Repeat the above steps S1-S3 to obtain the boundary values ​​A1j, A2j, A3j, A4j and A5j corresponding to each type of data in each storage unit, where j refers to different storage units, j=1, 2, ..., c, where c refers to the number of data categories, c is a positive integer and c is greater than 1.

[0064] Step 3: Analyze the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time. The specific method is as follows:

[0065] S01: Randomly select a category from various types of data without replacement as analysis data;

[0066] S02: setting multiple preset running times Tt, and randomly selecting one from the multiple preset running times Tt as the analysis time, where t refers to different preset running times, t=1, 2, ..., e, where e refers to the number of preset running times, e is a positive integer and e is greater than 1;

[0067] Obtain the historical value Ki of the analysis data of each storage unit of multiple solid state hard disks with the same specifications under the analysis time, and use the average of the maximum and minimum values ​​in Ki as the calibration value E11 of the analysis data under the analysis time, where i is a different analysis data value;

[0068] S03: Repeat the above step S01 to obtain the calibration values ​​E1t corresponding to the analysis data under different preset running time;

[0069] According to different preset running time and the calibration values ​​E1t corresponding to the analysis data under different preset running time, a two-dimensional curve graph of the analysis data with the running time is obtained, and the two-dimensional curve graph is analyzed to obtain the relationship coefficient H1 between the analysis data and the running time. The specific method is:

[0070] Different preset running time Tt is used as the horizontal coordinate, and the calibration value E1t corresponding to different preset running time is used as the vertical coordinate, so as to obtain the data points Wt (Tt, E1t) corresponding to the analysis data under different preset running time, and mark the connecting line between each two adjacent data points as a connecting line Lq, and obtain the slope Kq corresponding to each connecting line according to the coordinates of the two data points constituting each connecting line, where q is a different connecting line, q=1, 2, ..., p, where p refers to the number of connecting lines, p is a positive integer and p is equal to the number of preset running time e minus 1;

[0071] The number of positive and negative values ​​in the slope Kq is analyzed to obtain the relationship coefficient H1 between the analytical data and the running time, as follows:

[0072] The number of positive and negative values ​​in the slope Kq is marked as U + and U - , get U + and U - The ratio Up to the number of connecting lines p + and Up - , when Up + When it is greater than or equal to one-half, the mean absolute value of the slope Kq is used as the relationship coefficient H1 between the analysis data and the running time. -When it is greater than or equal to one-half, the product of the absolute value of the mean of the slope Kq and minus one is taken as the relationship coefficient H1 between the analysis data and the running time. + and Up - If both are not greater than or equal to one half, 0 is taken as the relationship coefficient H1 between the analytical data and the running time;

[0073] The specific method of obtaining the slope Kq corresponding to the connecting lines is:

[0074] That is, the ratio between the absolute value of the difference between the ordinate of the next data point on each connecting line and the ordinate of the previous data point and the absolute value of the difference between their corresponding abscissas, i.e., the preset running time, is used as the slope Kq corresponding to each connecting line;

[0075] It should be noted that the first endpoint of the connecting line refers to the data point close to the origin among the two data points constituting the connecting line, and the second endpoint refers to the data point located at the right endpoint of the connecting line;

[0076] S04: repeat the above steps S01-S03, and then the relationship coefficients between various types of data and running time are H1, H2, H3, H4 and H5;

[0077] Step 4: According to the relationship coefficient between each type of data and the running time and the boundary value corresponding to each type of data in each storage unit, the boundary vector corresponding to each storage unit is obtained. The specific method is as follows:

[0078] The products of the boundary values ​​corresponding to each type of data in each storage unit and the relationship coefficients between each type of data and the running time are obtained, and vectorized processing is performed to obtain the boundary vectors Jj (JA1j, JA2j, JA3j, JA4j, JA5j) corresponding to each storage unit.

[0079] Step 5: When performing data read and write operations, obtain the real-time read and write data corresponding to each storage unit, which includes the real-time read and write times, real-time read and write error rate, real-time erase times, total real-time read and write data, and real-time bad block quantity. According to the real-time read and write data corresponding to each storage unit, obtain the real-time operation vector corresponding to each storage unit, and calculate the trend distance between the real-time operation vector of each storage unit and the boundary vector. The specific method is as follows:

[0080] The real-time read and write times, real-time read and write error rates, real-time erase times, total real-time read and write data, and real-time bad block numbers corresponding to each storage unit are vectorized to obtain the real-time operation vectors Zj (Z1j, Z2j, Z3j, Z4j, Z5j) corresponding to each storage unit.

[0081] The specific calculation formula of trend distance is: ;

[0082] Calculate and obtain the trend distance Qj between the real-time operation vector of each storage unit and the boundary vector;

[0083] Step 6: Mark the failing units according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector, so as to take measures in advance, such as data migration or replacement of storage units to control the solid-state drive, and prevent the failing units from continuing to read and write data, resulting in data loss and damage. The specific method is as follows:

[0084] The storage unit whose trend distance Qj is greater than the preset value Y3 is marked as a failed unit. The specific value of the preset value Y3 is formulated by relevant personnel according to actual needs.

[0085] By collecting and analyzing the historical failure data of each storage unit in the SSD, combining the operating time with the correlation coefficient of various data, determining the boundary vector of the storage unit, and marking the failing unit through the trend distance between the real-time operation vector and the boundary vector, the failure of each storage unit in the SSD can be effectively predicted, and measures can be taken in advance to avoid data loss and equipment damage, such as data migration or replacement of storage units, to avoid data loss and damage, thereby improving the reliability and data security of the SSD.

[0086] Example 2: As Example 2 of the present invention, please refer to Figure 2 , providing a solid state hard disk system, the system is used to implement a solid state hard disk control method disclosed above, specifically comprising;

[0087] A data collection module collects various types of data from a plurality of historical fault data of each storage unit in the solid state drive;

[0088] The boundary value acquisition module obtains the boundary values ​​corresponding to each type of data in each storage unit;

[0089] The relationship coefficient acquisition module analyzes the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time;

[0090] A limit vector acquisition module obtains the limit vectors corresponding to each storage unit according to the limit values ​​corresponding to each storage unit;

[0091] A trend distance acquisition module calculates the trend distance between the real-time operation vector of each storage unit and the boundary vector according to the real-time operation vector obtained from the real-time read and write data corresponding to each storage unit;

[0092] A failing unit marking module marks failing units according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector;

[0093] After marking the units that are about to fail, a multi-channel system notification mechanism is established to ensure that the administrator can obtain information about the failed unit marking and subsequent control operations in a timely manner. In addition to displaying notification messages in the SSD management software, notifications are also sent to administrators via email, text messages, etc. For example, when the system detects that a storage unit is marked as a failed unit, a notification email is immediately sent to the administrator's email address, and a text message reminder is sent to the bound mobile phone to inform the location of the failed unit, changes in related performance indicators, and other detailed information, so that the administrator can make decisions in a timely manner.

[0094] Embodiment 3: As embodiment 3 of the present invention, a control device for a solid-state hard disk is provided, and the device is used to apply and implement a control method for a solid-state hard disk and a solid-state hard disk system disclosed above, and serve as an application carrier of the control method and the hard disk system.

[0095] Embodiment 4: As the fourth embodiment of the present invention, when the present application is implemented in detail, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned embodiments 1, 2 and 3.

[0096] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0097] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A control method for a solid state hard disk, characterized in that: The following steps are involved: Step 1: Collect various types of data from multiple historical fault data of each storage unit in the solid-state hard disk; the historical fault data includes the number of read and write times, read and write error rate, number of erase times, total amount of read and write data, and number of bad blocks corresponding to each storage unit in the solid-state hard disk; Read and write times: the total number of read and write operations of the storage unit when a fault occurs; Erase times: the total number of erase and write operations experienced by the storage unit when a fault occurs; Read and write error rate: the ratio between the number of read and write errors and the total number of read and write operations when a fault occurs; Total amount of read and write data: the total amount of data processed by the storage unit when a fault occurs; Step 2: Obtain the boundary values ​​corresponding to each type of data in each storage unit; Step 3: Analyze the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time; Step 4: According to the relationship coefficient between each type of data and the running time and the boundary value corresponding to each storage unit, obtain the boundary vector corresponding to each storage unit; Step 5: obtaining the trend distance between the real-time operation vector of each storage unit and the boundary vector according to the real-time read and write data corresponding to each storage unit; Step 6: Mark the units that are about to fail according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector; The specific method of obtaining the boundary values ​​corresponding to each type of data in each storage unit is: S1: Randomly select one of the storage units without replacement as the target unit; S2: Randomly select a category from each category of data without replacement as the target category data, mark the target category data in multiple historical fault data of the target unit as Rr, obtain the number b of values ​​in Rr that meet the preset screening condition X, when the number b is greater than the preset threshold Y1, define the mean Rp of Rr as the boundary value A11 corresponding to the target category data in the target unit, when the number b is less than the preset threshold Y1, define the mean of the maximum and minimum values ​​in Rr as the boundary value A11 corresponding to the target category data in the target unit, where r refers to different historical fault data, and the value is r=1, 2, ..., a, where a refers to the number of historical fault data, a is a positive integer and a>1, a≥b≥1; S3: Repeat the above step S2 to obtain the boundary values ​​A11, A21, A31, A41 and A51 corresponding to each type of data in the target unit; S4: Repeat the above steps S1-S3 to obtain the boundary values ​​A1j, A2j, A3j, A4j and A5j corresponding to each type of data in each storage unit, where j refers to different storage units, j=1, 2, ..., c, where c refers to the number of data categories, c is a positive integer and c is greater than 1; The specific method of obtaining the relationship coefficient between various data and running time is: S01: Randomly select a category from various types of data without replacement as analysis data; S02: multiple preset running times Tt are set, one is randomly selected from the multiple preset running times Tt as the analysis time, and the average of the maximum and minimum values ​​of the historical values ​​Ki of the analysis data of each storage unit of multiple solid state drives with the same specifications under the analysis time is used as the calibration value E11 of the analysis data under the analysis time, where i is a different analysis data value; S03: Repeat the above step S01 to obtain the calibration values ​​E1t corresponding to the analysis data at different preset running times, where t refers to different preset running times, t=1, 2, ..., e, where e refers to the number of preset running times, e is a positive integer and e is greater than 1; according to the calibration values ​​E1t corresponding to different preset running times and the analysis data at different preset running times, a two-dimensional curve graph of the analysis data with the running time is obtained, and the two-dimensional curve graph is analyzed to obtain the relationship coefficient H1 between the analysis data and the running time; S04: Repeat the above steps S01-S03, and then calculate the relationship coefficients H1, H2, H3, H4 and H5 between various types of data and the running time.

2. A control method for a solid state hard disk according to claim 1, characterized in that: The preset screening condition X is: |Rr-Rp|≥Y2, where Rp is the mean of Rr, and Y2 is the preset threshold.

3. The control method of a solid state hard disk according to claim 1, characterized in that: The specific method of obtaining the relationship coefficient between analytical data and running time is: Different preset running time Tt is used as the horizontal coordinate, and the calibration value E1t corresponding to different preset running time is used as the vertical coordinate, so as to obtain the data points Wt (Tt, E1t) corresponding to the analysis data under different preset running time, mark the connecting line between each two adjacent data points as a connecting line Lq, and obtain the slope Kq corresponding to each connecting line according to the coordinates of the two data points that constitute each connecting line. The number of positive and negative values ​​in the slope Kq is analyzed to obtain the relationship coefficient H1 between the analysis data and the running time, where q is a different connecting line, q=1, 2, ..., p, where p refers to the number of connecting lines, p is a positive integer and p is equal to the preset running time number e minus 1.

4. A control method for a solid state hard disk according to claim 3, characterized in that: The specific way to analyze the number of positive and negative values ​​in the slope Kq is: The number of positive and negative values ​​in the slope Kq is marked as U + and U - , get U + and U - The ratio Up to the number of connecting lines p + and Up - , when Up + When it is greater than or equal to one-half, the mean absolute value of the slope Kq is used as the relationship coefficient H1 between the analysis data and the running time. - When it is greater than or equal to one-half, the product of the absolute value of the mean of the slope Kq and minus one is taken as the relationship coefficient H1 between the analysis data and the running time. + and Up - When both are not greater than or equal to one half, 0 is used as the relationship coefficient H1 between the analytical data and the running time.

5. A control method for a solid state hard disk according to claim 4, characterized in that: The specific method of obtaining the boundary vectors corresponding to each storage unit is: The products of the boundary values ​​corresponding to each type of data in each storage unit and the relationship coefficients between each type of data and the running time are obtained, and vectorized processing is performed to obtain the boundary vectors Jj (JA1j, JA2j, JA3j, JA4j, JA5j) corresponding to each storage unit.

6. A control method for a solid state hard disk according to claim 5, characterized in that: The specific method of obtaining the trend distance between the real-time operation vector of each storage unit and the boundary vector is as follows: The square root of the sum of the squares of the differences between each data in the operation vector of each storage unit and each data in the boundary vector is taken to obtain the trend distance Qj between the real-time operation vector of each storage unit and the boundary vector.

7. A control method for a solid state hard disk according to claim 6, characterized in that: The specific method of marking the failing unit is as follows: The storage units whose trend distance Qj is greater than the preset value Y3 are marked as failed units.

8. A solid state hard disk system, characterized in that: The system implements a control method for a solid state hard disk as described in any one of claims 1 to 7, including: A data collection module collects various types of data from a plurality of historical fault data of each storage unit in the solid state drive; The boundary value acquisition module obtains the boundary values ​​corresponding to each type of data in each storage unit; The relationship coefficient acquisition module analyzes the historical values ​​of various data of the solid state drive under different preset operating time to obtain the relationship coefficient between various data and operating time; A limit vector acquisition module obtains the limit vectors corresponding to each storage unit according to the limit values ​​corresponding to each storage unit; A trend distance acquisition module calculates the trend distance between the real-time operation vector of each storage unit and the boundary vector according to the real-time operation vector obtained from the real-time read and write data corresponding to each storage unit; The failing unit marking module marks the failing unit according to the trend distance between the real-time operation vectors corresponding to each storage unit and the boundary vector.

Images