A memory failure prediction method, device, equipment and medium

Abstract

The present invention proposes a method for predicting memory failure, comprising: establishing a first memory failure prediction model based on a set of memory failure characteristic parameters, the first prediction model being used to predict the possibility of memory failure; establishing a second memory failure prediction model based on the first memory failure prediction model and a memory information sampling time, the second prediction model being used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure; determining the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The present invention also proposes a memory failure prediction device, equipment and medium, which can accurately obtain the remaining time until the memory fails, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

Description

Technical Field

[0001] The present invention relates to the field of memory failure, and in particular to a memory failure prediction method, device, equipment and medium. Background Art

[0002] Memory failure can cause server performance degradation or even downtime, leading to problems such as loss of server operating data.

[0003] There are many types of memory on the market. Different manufacturers have different diagnostic methods for memory failures, but there are relatively few methods for predicting memory failures. In addition, in the existing technology, although it is possible to judge and predict whether the memory is faulty, it is impossible to accurately predict the time when the memory failure will occur, which reduces the reliability of memory failure prediction.

[0004] In view of this problem, the present invention provides a memory failure prediction method, device, equipment and medium to solve the above problem. Summary of the invention

[0005] In order to solve the problems existing in the prior art, the present invention innovatively proposes a memory fault prediction method, device, equipment and medium, which effectively solves the problem of low reliability of memory fault prediction caused by the prior art and effectively improves the reliability of memory fault prediction.

[0006] A first aspect of the present invention provides a method for predicting memory failure, comprising:

[0007] Establishing a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of memory failure;

[0008] Establishing a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure;

[0009] The remaining time until the memory fails is determined based on the time series sliding window and the second memory failure prediction model.

[0010] Optionally, establishing a first memory failure prediction model based on the memory failure characteristic parameter set specifically includes:

[0011] At each time interval T, memory information of different models is collected, and characteristic parameters that can reflect the fault conditions of the memory are selected from the collected memory information of different models to form a memory fault characteristic parameter set;

[0012] Collect data information of multiple different types of faulty memory, focus on a certain memory fault characteristic parameter respectively, take the mode of the memory fault characteristic parameter as the threshold value corresponding to the memory fault characteristic parameter;

[0013] The first memory failure prediction model established is specifically:

[0014] Wherein, F is the possibility of memory failure. When the F value is 0, the memory will fail. The smaller the difference between the F value and 0, the higher the possibility of memory failure. N is the number of memory failure characteristic parameters in the memory failure characteristic parameter set. Y i is the weight of the i-th memory fault characteristic parameter when the memory fails; x i is a value of the i-th memory fault characteristic parameter in the memory fault characteristic parameter set; θ i is the threshold at which the i-th memory fault characteristic parameter in the memory fault characteristic parameter set can cause a memory fault.

[0015] Furthermore, when a memory failure occurs, the weight of the i-th memory failure characteristic parameter is calculated as follows: i =100*y i , where y i is the probability of the influence of the i-th memory fault characteristic parameter on the memory fault, and the method for determining the probability of the influence of the memory fault characteristic parameter on the memory fault is:

[0016] P(a i ) is the i-th memory fault characteristic parameter a i The prior probability of the impact on memory failure, P(x i |a i ) is the i-th memory fault characteristic parameter a i The value of x i The conditional probability of memory failure.

[0017] Optionally, it also includes:

[0018] In the memory information sampling samples, the thresholds of various memory fault characteristic parameters are adjusted according to the difference between the probability F of memory fault occurrence corresponding to different sampling times and 0 after extracting the faulty memory information.

[0019] Optionally, the remaining time until a memory failure occurs is determined based on the time series sliding window and the second memory failure prediction model:

[0020] Set a time series sliding window and get the current moving position of the time series sliding window;

[0021] Determine, according to the current moving position of the time series sliding window, the cumulative average value of the memory fault characteristic parameter information within the current time series sliding window range and the cumulative average value of the probability of memory fault corresponding to the memory fault characteristic parameter information;

[0022] Determine a sampling time corresponding to the cumulative average value of the possibility of memory failure in the current time series sliding window according to the cumulative average value of the possibility of memory failure corresponding to the memory failure characteristic parameter information and the second prediction model;

[0023] The remaining time until the memory fails is determined according to the sampling time corresponding to the cumulative average value of the probability of memory failure in the current time series sliding window and the time when the memory fails.

[0024] Furthermore, the remaining time until the memory fails is the difference between the time when the memory fails and the sampling time corresponding to the cumulative average of the probability of the memory failing in the current time series sliding window.

[0025] Optionally, the time when the memory failure occurs is a time point when the value of the first memory failure prediction model is zero.

[0026] A second aspect of the present invention provides a memory failure prediction device, comprising:

[0027] A first establishing module, establishing a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of memory failure;

[0028] A second establishing module is used to establish a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure;

[0029] A determination module determines the remaining time until a memory failure occurs based on a time series sliding window and a second memory failure prediction model.

[0030] A third aspect of the present invention provides an electronic device, comprising: a memory for storing a computer program; and a processor for implementing the steps of a memory failure prediction method as described in the first aspect of the present invention when executing the computer program.

[0031] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of a memory failure prediction method as described in the first aspect of the present invention are implemented.

[0032] The technical solution adopted by the present invention includes the following technical effects:

[0033] 1. The present invention establishes a first memory failure prediction model for predicting the possibility of memory failure based on a memory failure characteristic parameter set, establishes a second memory failure prediction model for establishing a corresponding relationship between memory information sampling time and the possibility of memory failure based on the first memory failure prediction model and memory information sampling time, and determines the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The remaining time until the memory fails can be accurately obtained, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

[0034] 2. The first memory failure prediction model in the technical solution of the present invention includes the weight of the i-th memory failure characteristic parameter when the memory fails. It not only takes into account the situation where multiple influencing factors work together to cause memory failure, but also reduces the difficulty of predicting memory failure caused by the joint action of multiple influencing factors.

[0035] 3. In the technical solution of the present invention, in the memory information sampling sample, the threshold value of each memory fault characteristic parameter is adjusted according to the difference between the extracted faulty memory information and the possibility F of memory failure corresponding to different sampling times and 0, so that the threshold value of each memory fault characteristic parameter can be corrected to improve the memory prediction accuracy.

[0036] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 This is a schematic diagram of a process of a method in Example 1 of the present invention;

[0039] Figure 2 A schematic diagram of a process of step S1 in the method of embodiment 1 of the present invention;

[0040] Figure 3 Another schematic flow chart of step S1 in the method of embodiment 1 of the present invention;

[0041] Figure 4 A schematic diagram of the corresponding relationship between the memory information sampling time and the possibility of memory failure in the second prediction model in the method of embodiment 1 of the solution of the present invention;

[0042] Figure 5 This is a flow chart of step S3 in the method of embodiment 1 of the present invention;

[0043] Figure 6 This is a schematic diagram of adding a time series sliding window to the second prediction model in the method of Example 1 in the solution of the present invention;

[0044] Figure 7 This is a schematic diagram of the structure of the device in Example 2 of the present invention;

[0045] Figure 8 This is a schematic diagram of the structure of the device of Example 3 in the solution of the present invention. DETAILED DESCRIPTION

[0046] In order to clearly illustrate the technical features of the present solution, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings. The disclosure below provides many different embodiments or examples for realizing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and settings of specific examples are described below. In addition, the present invention may repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. It should be noted that the components illustrated in the accompanying drawings are not necessarily drawn to scale. The present invention omits the description of known components and processing techniques and processes to avoid unnecessary limitations on the present invention.

[0047] Embodiment 1

[0048] like Figure 1 As shown, the present invention provides a memory fault prediction method, comprising:

[0049] S1, establishing a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of memory failure;

[0050] S2, establishing a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure;

[0051] S3, determining the remaining time until the memory fails based on the time series sliding window and the second memory failure prediction model.

[0052] Among them, Figure 2 As shown, step S1 specifically includes:

[0053] S11, collecting memory information of different models at every interval T, and selecting characteristic parameters that can reflect the fault conditions of the memory from the collected memory information of different models to form a memory fault characteristic parameter set;

[0054] S12, collecting data information of multiple different types of faulty memories, focusing on only one memory fault characteristic parameter, taking the mode of the memory fault characteristic parameter as the threshold value corresponding to the memory fault characteristic parameter;

[0055] S13, the first memory failure prediction model established is specifically:

[0056] Wherein, F is the possibility of memory failure. When the F value is 0, the memory will fail. The smaller the difference between the F value and 0, the higher the possibility of memory failure. N is the number of memory failure characteristic parameters in the memory failure characteristic parameter set. Y i is the weight of the i-th memory fault characteristic parameter when the memory fails; x i is a value of the i-th memory fault characteristic parameter in the memory fault characteristic parameter set; θ i is the threshold at which the i-th memory fault characteristic parameter in the memory fault characteristic parameter set can cause a memory fault.

[0057] Among them, in step S11, every time interval T, different types of memory information are collected once, and characteristic parameters that can reflect the fault conditions of the memory are screened from the collected different types of memory information to form a memory fault characteristic parameter set; wherein the characteristic parameters that affect the memory fault can be:

[0058]

[0059]

[0060] The above 6 characteristic parameters are respectively used: a 1 、a 2 ……a 6 To represent, that is, the memory fault characteristic parameter set A = {a 1 、a 2 ……a 6 The correlation between these characteristic parameters can reflect the current operating status of the memory.

[0061] Among them, in step S12, a large amount of data information of faulty memory is collected, and only one item in the memory fault characteristic parameter set A is focused on, and then the mode of the memory fault characteristic parameter is taken as the threshold value θ. For example, the mode of the parameter: Cell Error in the faulty memory data information is the threshold value θ 1Step S11 and step S12 are mainly to provide a data basis for the establishment and training of the first prediction model in step S13.

[0062] In step S13, the first memory failure prediction model established is specifically:

[0063] Wherein, F is the possibility of memory failure. When the F value is 0, the memory will fail. The smaller the difference between the F value and 0, the higher the possibility of memory failure. N is the number of memory failure characteristic parameters in the memory failure characteristic parameter set. Y i is the weight of the i-th memory fault characteristic parameter when the memory fails; x i is a value of the i-th memory fault characteristic parameter in the memory fault characteristic parameter set; θ i is the threshold at which the i-th memory fault characteristic parameter in the memory fault characteristic parameter set can cause a memory fault.

[0064] The specific calculation method of the weight of the i-th memory fault characteristic parameter when the memory fails is: i =100*y i , where yi is the probability of the influence of the i-th memory fault characteristic parameter on the memory fault, and the method for determining the probability of the influence of the memory fault characteristic parameter on the memory fault is:

[0065] P(a i ) is the i-th memory fault characteristic parameter a i The prior probability of the impact on memory failure, P(x i |a i ) is the i-th memory fault characteristic parameter a i The value of x i The conditional probability of memory failure.

[0066] Because the failure of a memory may be the result of one or more characteristic parameters in the memory failure characteristic parameter set A. Generally speaking, if a memory failure is caused by one of the characteristic parameters, it is relatively easy to determine whether the memory is about to fail; if multiple characteristic parameters act together to cause a memory failure, it is difficult to determine when the memory is about to fail. Therefore, the characteristic parameter combinations that affect the memory are: So many combinations greatly increase the difficulty of determining and predicting memory failures.

[0067] Therefore, in the embodiment of the present invention, it is assumed that the conditions between the various memory characteristic parameters are independent of each other, and then the formula for calculating the influence of a memory fault characteristic parameter in the memory fault characteristic parameter set A on the memory fault is:

[0068]

[0069] The probability y(x) at this time can represent the memory characteristic parameter a i The probability of impact on memory failure. i is the characteristic parameter a in the memory fault characteristic parameter set A i One of the values ​​(0 <x i <θ i ,θ i is the characteristic parameter a i threshold that can cause memory failure).

[0070] Specifically, the characteristic parameter a can be estimated by the memory fault characteristic parameter sample training set M. i The prior probability of causing hard disk failure is recorded as P(a i ), and estimate the conditional probability P(x i |a i ): The expression at this time is interpreted as the memory failure is caused by the characteristic parameter a i caused by, then a i The value of x i The probability of P(x i |a i ). Let M a Represents the memory fault feature parameter a in the training set M i For a set of samples, if there are sufficient independent and identically distributed samples, the prior probability can be easily estimated:

[0071]

[0072] make Indicates M a The value of the i-th characteristic parameter is x i The conditional probability P(x i |a i ) can be estimated as:

[0073]

[0074] From this, we can deduce the probability of each characteristic parameter in A causing memory failure, which are recorded as: y 1 ,y 2 ……y 6 , then the weight of each characteristic parameter when the memory fails can be calculated as: Y i =100*y i .

[0075] Furthermore, if Figure 3 As shown, step S1 also includes:

[0076] S14, in the memory information sampling samples, according to the difference between the probability F of memory failure corresponding to different sampling times and 0, the threshold value of each memory failure characteristic parameter is adjusted.

[0077] In step S14, the threshold of the memory fault characteristic parameter used is the mode. The prediction of the memory fault diagnosis here is a little rough, so this step is to correct the threshold to improve the accuracy of memory prediction. Track and observe W normal memories, and collect memory information once every interval T. Use the first prediction model to calculate the probability F of memory failure after each collection. As time goes by, the value of F will gradually approach 0 or greater than 0 from F<0. In this process, there may be several situations:

[0078] ①When F<0, the memory fails;

[0079] ②When F=0, the memory fails;

[0080] ③When F>0, the memory fails;

[0081] As for ① and ③, there is a deviation in the prediction of memory failure. At this time, it is necessary to adjust the threshold θ of the feature parameter based on the data collected when the memory failure occurs. i , making the value of F close to 0.

[0082] Threshold correction process: In the above-collected memory information samples, the information of the fault memory is extracted and recorded as sample G. The first prediction model is used to gradually calculate the value in sample G according to the time sequence of memory information collection. In order to make the value of function F change from F<0 at the beginning to F gradually approaching (close to) 0, the θ of each feature parameter is gradually adjusted. i (Threshold) has been satisfied by the process.

[0083] In step S2, a relatively accurate formula for predicting memory failure can be obtained according to step S1. A batch of memory information is sampled and according to the first prediction model, it is theoretically possible to determine which memory is about to fail, but it is currently impossible to determine how long it will take for the memory to fail.

[0084] Therefore, in step S2, a batch of memory information is continuously collected at each interval T, and the time point is recorded each time data is collected. After a period of time, the memory information that has caused the fault is extracted. Then a two-dimensional relationship table between the sampling time t and the probability F of predicting memory failure can be obtained:

[0085]

[0086]

[0087] Through the above two-dimensional table of the probability of memory failure F and sampling time t, the functional relationship between the two can be derived: F = f(t). A curve of sampling time (t) and memory failure function (F) can be obtained, such as Figure 4 shown.

[0088] Further, among them, Figure 5 As shown, step S3 specifically comprises:

[0089] S31, setting a time series sliding window, and obtaining the current moving position of the time series sliding window;

[0090] S32, determining the cumulative average value of the memory fault characteristic parameter information within the current time series sliding window range and the cumulative average value of the probability of memory fault occurrence corresponding to the memory fault characteristic parameter information according to the current moving position of the time series sliding window;

[0091] S33, determining a sampling time corresponding to the cumulative average value of the possibility of memory failure in the current time series sliding window according to the cumulative average value of the possibility of memory failure corresponding to the memory failure characteristic parameter information and the second prediction model;

[0092] S34, determining the remaining time until the memory fails according to the sampling time corresponding to the cumulative average value of the probability of the memory failing in the current time series sliding window and the time when the memory fails.

[0093] Among them, in step S31, then consider through a simple point (t i ,F i ) to estimate how long it will take for the memory to fail. This method still has a large error. Because when the memory is running in the server, its memory characteristic parameter information does not increase evenly over time, so in step S3 of the embodiment of the present invention, a time series sliding window is used, and the length of this window is H (hours, that is, from t s to e ). Add a sliding window and move it to a certain position, such as Figure 6 As shown, the characteristic parameter information of the time series sliding window time range is accumulated and averaged. Assuming that the time series sliding window is within the range of H (hours), the sampling is M times, then the average value of a certain characteristic parameter is:

[0094] Then the first prediction model can be evolved as follows:

[0095]

[0096] It can be calculated Then according to the corresponding relationship of t / F in the second prediction model, the corresponding sampling time point can be found The sampling time point and prediction function within the sliding window are obtained as follows: Sampling time point At the point (t j ,0) (this point is the time point when the memory failure occurs) The time required in the process is the remaining time before the memory fails.

[0097] That is, the probability of the memory failure corresponding to the memory information collected at any sampling time point t is F. To predict how long it will take for the memory to fail, the point (t, F) needs to be put into a sliding window, where the time range of the sliding window is tH / 2 to t+H / 2. Then, the value of the memory information collected within this time range is calculated using formula (1.2): according to exist Figure 4 You can get Then the remaining time before the memory fails is:

[0098] Specifically, the remaining time until the memory fails is the difference between the time when the memory fails and the sampling time corresponding to the cumulative average of the possibility of the memory failure in the current time series sliding window. The time when the memory fails is the time point when the value of the first prediction model of the memory failure is zero.

[0099] It should be noted that, in the memory information sampling process, the technical solution of the present invention can select information of different types of memory for sampling to reduce the impact of a single type of memory on the memory fault prediction result, and further improve the reliability of the memory fault prediction result.

[0100] The present invention establishes a first memory failure prediction model for predicting the possibility of memory failure based on a memory failure characteristic parameter set, establishes a second memory failure prediction model for establishing a corresponding relationship between memory information sampling time and the possibility of memory failure based on the first memory failure prediction model and the memory information sampling time, and determines the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The remaining time until the memory fails can be accurately obtained, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

[0101] The first memory failure prediction model in the technical solution of the present invention includes the weight of the i-th memory failure characteristic parameter when the memory fails. It not only takes into account the situation where multiple influencing factors work together to cause memory failure, but also reduces the difficulty of predicting memory failure caused by the joint action of multiple influencing factors.

[0102] In the technical solution of the present invention, in the memory information sampling sample, the threshold value of each memory fault characteristic parameter is adjusted according to the difference between the probability F of memory failure corresponding to different sampling times and 0 after the fault memory information is extracted. The threshold value of each memory fault characteristic parameter can be corrected to improve the memory prediction accuracy.

[0103] Embodiment 2

[0104] like Figure 7 As shown, the technical solution of the present invention also provides a memory failure prediction device, comprising:

[0105] A first establishing module 101, establishing a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of memory failure;

[0106] A second establishing module 102, establishing a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure;

[0107] The determination module 103 determines the remaining time until the memory fails based on the time series sliding window and the second memory failure prediction model.

[0108] The present invention establishes a first memory failure prediction model for predicting the possibility of memory failure based on a memory failure characteristic parameter set, establishes a second memory failure prediction model for establishing a corresponding relationship between memory information sampling time and the possibility of memory failure based on the first memory failure prediction model and the memory information sampling time, and determines the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The remaining time until the memory fails can be accurately obtained, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

[0109] The first memory failure prediction model in the technical solution of the present invention includes the weight of the i-th memory failure characteristic parameter when the memory fails. It not only takes into account the situation where multiple influencing factors work together to cause memory failure, but also reduces the difficulty of predicting memory failure caused by the joint action of multiple influencing factors.

[0110] In the technical solution of the present invention, in the memory information sampling sample, the threshold value of each memory fault characteristic parameter is adjusted according to the difference between the probability F of memory failure corresponding to different sampling times and 0 after the fault memory information is extracted. The threshold value of each memory fault characteristic parameter can be corrected to improve the memory prediction accuracy.

[0111] Embodiment 3

[0112] like Figure 8 As shown, the technical solution of the present invention also provides an electronic device, including: a memory 201, used to store a computer program; a processor 202, used to implement the steps of a memory failure prediction method as in Example 1 when executing the computer program.

[0113] The memory 201 in the embodiment of the present application is used to store various types of data to support the operation of the electronic device. Examples of these data include: any computer program for operating on the electronic device. It can be understood that the memory 201 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a read-only optical disc (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. Volatile memory may be Random Access Memory (RAM), which is used as external cache.By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), direct memory bus random access memory (DRRAM). The memory 201 described in the embodiments of the present application is intended to include but is not limited to these and any other suitable types of memory.

[0114] The method disclosed in the above embodiment of the present application can be applied to the processor 202, or implemented by the processor 202. The processor 202 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 202 or the instruction in the form of software. The above processor 202 may be a general-purpose processor, a DSP (Digital Signal Processing, that is, a chip capable of implementing digital signal processing technology), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 202 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiment of the present application, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in the memory 201, and the processor 202 reads the program in the memory 201 and completes the steps of the above method in combination with its hardware. When the processor 202 executes the program, the corresponding processes in the various methods of the embodiments of the present application are implemented, which will not be described here for the sake of brevity.

[0115] The present invention establishes a first memory failure prediction model for predicting the possibility of memory failure based on a memory failure characteristic parameter set, establishes a second memory failure prediction model for establishing a corresponding relationship between memory information sampling time and the possibility of memory failure based on the first memory failure prediction model and the memory information sampling time, and determines the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The remaining time until the memory fails can be accurately obtained, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

[0116] The first memory failure prediction model in the technical solution of the present invention includes the weight of the i-th memory failure characteristic parameter when the memory fails. It not only takes into account the situation where multiple influencing factors work together to cause memory failure, but also reduces the difficulty of predicting memory failure caused by the joint action of multiple influencing factors.

[0117] In the technical solution of the present invention, in the memory information sampling sample, the threshold value of each memory fault characteristic parameter is adjusted according to the difference between the probability F of memory failure corresponding to different sampling times and 0 after the fault memory information is extracted. The threshold value of each memory fault characteristic parameter can be corrected to improve the memory prediction accuracy.

[0118] Embodiment 4

[0119] The technical solution of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of a memory failure prediction method as described in Example 1 are implemented.

[0120] For example, it includes a memory 201 storing a computer program, and the computer program can be executed by a processor 202 to complete the steps of the aforementioned method. The computer readable storage medium can be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, FlashMemory, magnetic surface storage, optical disk, or CD-ROM.

[0121] It can be understood by those skilled in the art that all or part of the steps of the above-mentioned method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiment are executed; and the aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, disks or optical disks. Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application can be essentially or partly embodied in the form of a software product, which is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. And the aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, disks or optical disks.

[0122] The present invention establishes a first memory failure prediction model for predicting the possibility of memory failure based on a memory failure characteristic parameter set, establishes a second memory failure prediction model for establishing a corresponding relationship between memory information sampling time and the possibility of memory failure based on the first memory failure prediction model and the memory information sampling time, and determines the remaining time until the memory fails based on a time series sliding window and the second memory failure prediction model. The remaining time until the memory fails can be accurately obtained, effectively solving the problem of low reliability of memory failure prediction caused by the prior art, and effectively improving the reliability of memory failure prediction.

[0123] The first memory failure prediction model in the technical solution of the present invention includes the weight of the i-th memory failure characteristic parameter when the memory fails. It not only takes into account the situation where multiple influencing factors work together to cause memory failure, but also reduces the difficulty of predicting memory failure caused by the joint action of multiple influencing factors.

[0124] In the technical solution of the present invention, in the memory information sampling sample, the threshold value of each memory fault characteristic parameter is adjusted according to the difference between the probability F of memory failure corresponding to different sampling times and 0 after the fault memory information is extracted. The threshold value of each memory fault characteristic parameter can be corrected to improve the memory prediction accuracy.

[0125] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A memory failure prediction method, Its characteristics are: include: Establishing a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of a memory failure; wherein establishing the first memory failure prediction model based on a memory failure characteristic parameter set specifically includes: At each time interval T, memory information of different models is collected, and characteristic parameters that can reflect the fault conditions of the memory are selected from the collected memory information of different models to form a memory fault characteristic parameter set; Collect data information of multiple different types of faulty memory, focus on a certain memory fault characteristic parameter, and take the mode of the memory fault characteristic parameter as the threshold value corresponding to the memory fault characteristic parameter; The first memory failure prediction model established is specifically: , where F is the possibility of memory failure. When the F value is 0, the memory will fail. The smaller the difference between the F value and 0, the higher the possibility of memory failure. N is the number of memory failure characteristic parameters in the memory failure characteristic parameter set. is the weight of the i-th memory fault characteristic parameter when the memory fails; is a value of the i-th memory fault characteristic parameter in the memory fault characteristic parameter set; is the threshold at which the i-th memory fault characteristic parameter in the memory fault characteristic parameter set can cause a memory fault; When a memory fault occurs, the weight of the i-th memory fault characteristic parameter is calculated as follows: Wherein, y is the probability of memory fault characteristic parameters affecting memory fault, and the probability of memory fault characteristic parameters affecting memory fault is determined as follows: , P( ) is the i-th memory fault characteristic parameter The prior probability of the impact of memory failure, is the characteristic parameter of the i-th memory fault The value of The conditional probability of the impact on memory failure; Establishing a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure; The remaining time until the memory distance to the fault is determined based on the time series sliding window and the second memory fault prediction model; wherein the remaining time until the memory distance to the fault is determined based on the time series sliding window and the second memory fault prediction model is specifically: Set a time series sliding window and get the current moving position of the time series sliding window; Determine, according to the current moving position of the time series sliding window, the cumulative average value of the memory fault characteristic parameter information within the current time series sliding window range and the cumulative average value of the probability of memory fault corresponding to the memory fault characteristic parameter information; Determine a sampling time corresponding to the cumulative average value of the possibility of memory failure in the current time series sliding window according to the cumulative average value of the possibility of memory failure corresponding to the memory failure characteristic parameter information and the second prediction model; The remaining time until the memory fails is determined according to the sampling time corresponding to the cumulative average value of the probability of memory failure in the current time series sliding window and the time when the memory fails.

2. A memory failure prediction method according to claim 1, Its characteristic is that include: In the memory information sampling samples, the thresholds of various memory fault characteristic parameters are adjusted according to the difference between the probability F of memory fault occurrence corresponding to different sampling times and 0 after extracting the faulty memory information.

3. A memory failure prediction method according to claim 1, Its characteristics are: The remaining time until the memory fails is the difference between the time when the memory fails and the sampling time corresponding to the cumulative average of the probability of memory failure in the current time series sliding window.

4. A memory failure prediction method according to claim 1 or 3, Its characteristics are: The time when the memory failure occurs is the time point when the value of the first memory failure prediction model is zero.

5. A memory failure prediction device, Its characteristics are: include: The first establishment module establishes a first memory failure prediction model based on a memory failure characteristic parameter set, wherein the first prediction model is used to predict the possibility of a memory failure; wherein establishing the first memory failure prediction model based on the memory failure characteristic parameter set specifically includes: At each time interval T, memory information of different models is collected, and characteristic parameters that can reflect the fault conditions of the memory are selected from the collected memory information of different models to form a memory fault characteristic parameter set; Collect data information of multiple different types of faulty memory, focus on a certain memory fault characteristic parameter, and take the mode of the memory fault characteristic parameter as the threshold value corresponding to the memory fault characteristic parameter; The first memory failure prediction model established is specifically: , where F is the possibility of memory failure. When the F value is 0, the memory will fail. The smaller the difference between the F value and 0, the higher the possibility of memory failure. N is the number of memory failure characteristic parameters in the memory failure characteristic parameter set. is the weight of the i-th memory fault characteristic parameter when the memory fails; is a value of the i-th memory fault characteristic parameter in the memory fault characteristic parameter set; is the threshold at which the i-th memory fault characteristic parameter in the memory fault characteristic parameter set can cause a memory fault; When a memory fault occurs, the weight of the i-th memory fault characteristic parameter is calculated as follows: Wherein, y is the probability of memory fault characteristic parameters affecting memory fault, and the probability of memory fault characteristic parameters affecting memory fault is determined as follows: , P( ) is the i-th memory fault characteristic parameter The prior probability of the impact of memory failure, is the characteristic parameter of the i-th memory fault The value of The conditional probability of the impact on memory failure; A second establishing module is used to establish a second memory failure prediction model based on the first memory failure prediction model and the memory information sampling time, wherein the second prediction model is used to establish a corresponding relationship between the memory information sampling time and the possibility of memory failure; The determination module determines the remaining time until the memory fails based on the time series sliding window and the second memory failure prediction model; wherein the remaining time until the memory fails based on the time series sliding window and the second memory failure prediction model is specifically: Set a time series sliding window and get the current moving position of the time series sliding window; Determine, according to the current moving position of the time series sliding window, the cumulative average value of the memory fault characteristic parameter information within the current time series sliding window range and the cumulative average value of the probability of memory fault corresponding to the memory fault characteristic parameter information; Determine a sampling time corresponding to the cumulative average value of the possibility of memory failure in the current time series sliding window according to the cumulative average value of the possibility of memory failure corresponding to the memory failure characteristic parameter information and the second prediction model; The remaining time until the memory fails is determined according to the sampling time corresponding to the cumulative average value of the probability of memory failure in the current time series sliding window and the time when the memory fails.

6. An electronic device, Its characteristics are: include: Memory for storing computer programs; A processor, configured to implement the steps of a memory failure prediction method as claimed in any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, Its characteristics are: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a memory failure prediction method according to any one of claims 1 to 4 are implemented.

Images