Summary of the invention
The purpose of the present invention is to provide a kind of disk life-span prediction method, system, device and computer-readable storage mediumsMatter, to solve the problems, such as how accurately to predict the disk service life.
To achieve the above object, the embodiment of the invention provides following technical solutions:
A kind of disk life-span prediction method, comprising:
Determine the performance data of disk to be predicted；
The performance data of the disk to be predicted is predicted using corresponding first regression model in the first service life section,Obtain the first life value；
The first life value corresponding second service life section is determined, corresponding second time using second service life sectionReturn model to predict the performance data of the disk to be predicted, obtains prediction result；Wherein, first service life sectionRange includes second service life section, and first regression model is the History Performance Data using first service life sectionThe model being trained, second regression model are that the History Performance Data in second service life section is trainedThe model arrived.
Wherein, the performance data includes SMART data.
Wherein, it is described using corresponding second regression model in second service life section to the performance of the disk to be predictedData are predicted, prediction result is obtained, comprising:
It is carried out using performance data of corresponding second regression model in second service life section to the disk to be predictedPrediction, obtains the second life value；
It determines the obtained error information of training in advance, is determined using the error information with second life value described pre-Survey result.
Wherein, the range in first service life section includes at least two subintervals, then first life value pair of determinationThe the second service life section answered, comprising:
The subinterval including first life value is determined in described two subintervals using first life value, is madeFor the second service life section.
Wherein, it is described using corresponding second regression model in second service life section to the performance of the disk to be predictedData are predicted, after obtaining prediction result, further includes:
It is determining with the second life value corresponding third service life section using the prediction result as the second life value,The performance data of the disk to be predicted is predicted using the corresponding third regression model in third service life section, is obtainedNew prediction result；Wherein the range in first service life section includes third service life section, third service life sectionRange be less than second service life section range.
Present invention also provides a kind of disk life prediction systems, comprising:
Determining module, for determining the performance data of disk to be predicted；
First prediction module, for utilizing corresponding first regression model in the first service life section to the disk to be predictedPerformance data is predicted, the first life value is obtained；
Second prediction module utilizes second longevity for determining the first life value corresponding second service life sectionCorresponding second regression model in life section predicts the performance data of the disk to be predicted, obtains prediction result；Wherein,The range in first service life section includes second service life section, and first regression model is to utilize first service lifeThe model that the History Performance Data in section is trained, second regression model are the history in second service life sectionThe model that performance data is trained.
Wherein, second prediction module, comprising:
First determination unit, for determining the first life value corresponding second service life section；
Predicting unit, for utilizing corresponding second regression model in second service life section to the disk to be predictedPerformance data is predicted, the second life value is obtained；
Second determination unit utilizes the error information and described for determining the obtained error information of training in advanceTwo life values determine the prediction result.
Wherein, the range in first service life section includes at least two subintervals, then second prediction module, specificallyFor determined in described two subintervals using first life value include first life value subinterval, as theTwo service life sections, using corresponding second regression model in second service life section to the performance data of the disk to be predicted intoRow prediction, obtains prediction result
Present invention also provides a kind of disk life predication apparatus, comprising:
Memory, for storing computer program；
Processor is realized when for executing the computer program such as the step of the disk life-span prediction method.
Present invention also provides a kind of computer readable storage mediums, which is characterized in that the computer-readable storage mediumIt is stored with computer program in matter, realizes when the computer program is executed by processor such as the disk life-span prediction methodStep.
By above scheme it is found that a kind of disk life-span prediction method provided by the invention, comprising: determine disk to be predictedPerformance data；It is carried out using performance data of corresponding first regression model in the first service life section to the disk to be predicted pre-It surveys, obtains the first life value；It determines the first life value corresponding second service life section, utilizes second service life section pairThe second regression model answered predicts the performance data of the disk to be predicted, obtains prediction result；Wherein, described firstThe range in service life section includes second service life section, and first regression model is going through using first service life sectionThe model that history performance data is trained, second regression model are the History Performance Data in second service life sectionThe model being trained.
It can be seen that a kind of disk life-span prediction method provided by the present application, utilizes the first service life section corresponding firstRegression model first carries out more rough prediction to the performance data of disk to be predicted, the first life value is obtained, according to the first service lifeIt is worth higher second regression model of accuracy for determining that the performance data is applicable, thus using the second regression model to performance dataCarrying out prediction can be obtained by the prediction result in more accurate disk service life.Present invention also provides a kind of disk life prediction systemsSystem, device and computer readable storage medium, equally may be implemented above-mentioned technical effect.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, completeSite preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based onEmbodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every otherEmbodiment shall fall within the protection scope of the present invention.
The embodiment of the invention discloses a kind of disk life-span prediction method, system, device and computer readable storage medium,To solve the problems, such as how accurately to predict the disk service life.
Referring to Fig. 1, a kind of disk life-span prediction method provided in an embodiment of the present invention is specifically included:
S101 determines the performance data of disk to be predicted.
The performance data of disk to be predicted is determined first.It should be noted that in the present solution, performance data is to influence magneticThe performance data of disk service life.
In a preferred embodiment, the performance data includes SMART data.
The full name of SMART is " Self-Monitoring Analysis and Reporting Technology ", i.e.," self-monitoring, analysis and reporting techniques " are a kind of automatic disk state detection and early warning system and specification, SMART dataThe data in magnetic disk analyzed using the technical monitoring.Major part hard disk is equipped with this technology now, utilizes SMART dataThe performance that can more accurately reflect disk, since present most of disk is equipped with this technology, SMART dataAcquisition is also more convenient, so that disk life prediction can make prediction process more convenient this time using SMART data, makesPrediction result is more accurate.
S102 is carried out using performance data of corresponding first regression model in the first service life section to the disk to be predictedPrediction, obtains the first life value.
In the present solution, including the corresponding training pattern of two granularity level, i.e. the first training pattern and the second training mouldType.First granularity level is the corresponding coarse grain level in the first service life section, and second granularity level is the second service life sectionCorresponding fine granularity rank.Coarseness service life section includes fine granularity section, for example, the first lifetime region as coarseness sectionBetween be [1,90], as fine granularity section the second service life section be [1,30], in addition, fine granularity section can also include [31,60], there is a training pattern in [61,90], each corresponding section.
First training pattern is what the performance data of the disk using the service life in the first service life section was trainedRegression model, the second training pattern are the recurrence mould being trained using performance data of the service life in the second service life sectionType.
It is predicted first with performance data of first regression model to disk to be predicted, obtains the first life value.?That is predicting first with the biggish training pattern of service life interval range performance data to be predicted, one is obtainedPreliminary life value, as the first life value.
Since input data collected when training the first regression model includes the different life values in one a wide range ofThe performance data of disk, therefore the result predicted is accurate not enough.
S103 determines the first life value corresponding second service life section, corresponding using second service life sectionSecond regression model predicts the performance data of the disk to be predicted, obtains prediction result；Wherein, first service lifeThe range in section includes second service life section, and first regression model is the history using first service life sectionThe model that energy data are trained, second regression model are that the History Performance Data in second service life section carries outThe model that training obtains.
Specifically, after obtaining the first life value, that is, a life value substantially of disk to be predicted is determined, thusOne can be found according to this life value can be to the regression model that this disk is more accurately predicted.
Due to including the training pattern of two granularity level, the training pattern of coarse grain level, such as the first instruction in this programmePractice the training pattern of model and fine granularity rank, such as the second training pattern.Therefore, a performance data to be predicted is being obtainedThe first rough life value after, so that it may determine that prediction accuracy is higher according to this life value, the instruction of fine granularity rankPractice model again to predict the performance data of disk to be predicted, to obtain the higher life value of accuracy.
That is, in the present solution, first to determine that the performance data of disk to be predicted should according to the first regression modelIt selects which more accurate regression model as the second regression model, prediction knot is then accurately determined using the second regression modelFruit.
It should be noted that since the second regression model is trained using the History Performance Data in the second service life sectionIt obtains, and in second service life section namely the first service life section, the smaller section of range, in this smaller section,The gap in disk service life also can be relatively small, therefore the model trained carries out in advance the performance data in this service life sectionSurveying also will be more accurate.
In one preferred embodiment, during model training, trained model can be tested, is obtainedThe error amount of training result and actual result, so as to improve accuracy when actual prediction using the error amount.
Specifically, using corresponding second regression model in second service life section to the performance number of the disk to be predictedAccording to being predicted, the second life value is obtained；It determines the obtained error information of training in advance, utilizes the error information and described theTwo life values determine the prediction result.
It can be seen that a kind of disk life-span prediction method provided by the embodiments of the present application, corresponding using the first service life sectionThe first regression model more rough prediction is first carried out to the performance data of disk to be predicted, the first life value is obtained, according to theOne life value determines applicable higher second regression model of accuracy of the performance data, thus using the second regression model to propertyEnergy data, which carry out prediction, can be obtained by the prediction result in more accurate disk service life.
A kind of specific disk life-span prediction method provided by the embodiments of the present application is introduced below, it is described belowA kind of specific disk life-span prediction method can be cross-referenced with above-described embodiment.
Referring to fig. 2, a kind of specific disk life-span prediction method provided by the embodiments of the present application, specifically includes:
S201 determines the performance data of disk to be predicted.
S202 is carried out using performance data of corresponding first regression model in the first service life section to the disk to be predictedPrediction, obtains the first life value.
S203 determines the first life value corresponding second service life section, corresponding using second service life sectionSecond regression model predicts the performance data of the disk to be predicted, obtains prediction result；Wherein, first service lifeThe range in section includes second service life section, and first regression model is the history using first service life sectionThe model that energy data are trained, second regression model are that the History Performance Data in second service life section carries outThe model that training obtains.
In a specific embodiment, the range in first service life section includes at least two subintervals, then describedDetermine the first life value corresponding second service life section, comprising:
The subinterval including first life value is determined in described two subintervals using first life value, is madeFor the second service life section.
For example, the first service life section is [1,90] comprising three subintervals [1,30], [31,60], [61,90] are correspondingThere is a training pattern in each section.
After obtaining the first life value, i.e., area belonging to the first life value is found in subinterval using the first life valueBetween, to be trained to obtain training result using the training pattern in the section as the second training pattern.
For example, inputting the 40th day SMART value, the prediction result of output, that is, the first service life in the first training patternValue is 30 days, then the corresponding training pattern in [1,30] this section is selected to be predicted to obtain prediction result to above-mentioned SMART value.
S204 determines the third service life corresponding with second life value using the prediction result as the second life valueSection is carried out pre- using performance data of the corresponding third regression model in third service life section to the disk to be predictedIt surveys, obtains new prediction result；Wherein the range in second service life section includes third service life section.
In the present solution, prediction model is not only divided into two layers in order to further improve the accuracy of prediction, obtain it is pre-After surveying result, the higher third regression model of precision can also be determined using prediction result, be treated using third regression model pre-The performance data for surveying disk is predicted to obtain the higher prediction result of accuracy.
For example, the first service life section is [1,90] comprising three subintervals [1,30], [31,60], and [61,90], firstService life section further includes range subinterval [1,22] more smaller than the second service life section, [23,45], [46,68], [69,90],The corresponding training pattern in selection [1,30] this section is predicted to obtain after the second life value is 23 to above-mentioned SMART value, afterIt is continuous that the corresponding third regression model in [23,45] this section is utilized to predict the performance data of above-mentioned disk to be predicted, mostObtaining new prediction result eventually is 28.
It should be noted that in order to improve precision of prediction, can further to divide the smaller service life section of range, fromAnd the corresponding training pattern in the smaller service life section of result range of choice is predicted according to the training pattern in upper one layer of service life sectionContinue to train, so that precision is more accurate.
A kind of disk life prediction system provided by the embodiments of the present application is introduced below, a kind of magnetic described belowDisk life prediction system can be cross-referenced with above-mentioned any embodiment.
Referring to Fig. 3, a kind of disk life prediction system provided by the embodiments of the present application is specifically included:
Determining module 301, for determining the performance data of disk to be predicted.
First prediction module 302, for utilizing corresponding first regression model in the first service life section to the magnetic to be predictedThe performance data of disk is predicted, the first life value is obtained.
Second prediction module 303 utilizes described second for determining the first life value corresponding second service life sectionCorresponding second regression model in service life section predicts the performance data of the disk to be predicted, obtains prediction result；ItsIn, the range in first service life section includes second service life section, and first regression model is to utilize described firstThe model that the History Performance Data in service life section is trained, second regression model are second service life sectionThe model that History Performance Data is trained.
In one preferred embodiment, second prediction module 303 specifically includes:
First determination unit, for determining the first life value corresponding second service life section；
Predicting unit, for utilizing corresponding second regression model in second service life section to the disk to be predictedPerformance data is predicted, the second life value is obtained；
Second determination unit utilizes the error information and described for determining the obtained error information of training in advanceTwo life values determine the prediction result.
In one preferred embodiment, the range in first service life section includes at least two subintervals, then instituteThe second prediction module 303 is stated, being specifically used for determining in described two subintervals using first life value includes described theThe subinterval of one life value, as the second service life section, using corresponding second regression model in second service life section to instituteThe performance data for stating disk to be predicted is predicted, prediction result is obtained.
The disk life prediction system of the present embodiment is for realizing disk life-span prediction method above-mentioned, therefore the disk service lifeThe embodiment part of the visible disk life-span prediction method hereinbefore of specific embodiment in forecasting system, for example, determining mouldBlock 301, the first prediction module 302, the second prediction module 303 are respectively used to realize step in above-mentioned disk life-span prediction methodS101, S102, S104, so, specific embodiment is referred to the description of corresponding various pieces embodiment, herein no longerIt repeats.
A kind of disk life predication apparatus provided by the embodiments of the present application is introduced below, a kind of magnetic described belowDisk life predication apparatus can be cross-referenced with any of the above-described embodiment.
Referring to fig. 4, a kind of disk life predication apparatus provided by the embodiments of the present application, specifically includes:
Memory 100, for storing computer program；
Processor 200 realizes that the disk service life as described in above-mentioned any embodiment is pre- when for executing the computer programThe step of survey method.
Specifically, memory 100 includes non-volatile memory medium, built-in storage.Non-volatile memory medium storageThere are operating system and computer-readable instruction, which is that the operating system and computer in non-volatile memory medium canThe operation of reading instruction provides environment.Processor 200 provides calculating and control ability for disk life predication apparatus, may be implementedState step provided by any disk life-span prediction method embodiment.
On the basis of the above embodiments, preferably, the disk life predication apparatus further include:
Input interface 300 is controlled through processor and is saved for obtaining computer program, parameter and the instruction of external importingInto memory.The input interface 300 can be connected with input unit, receive parameter or instruction that user is manually entered.This is defeatedEntering device can be the touch layer covered on display screen, be also possible to the key being arranged in terminal enclosure, trace ball or Trackpad,It is also possible to keyboard, Trackpad or mouse etc..Specifically, in the present embodiment, user can be inputted by input interface 300 toPredict the performance data of disk.
Display unit 400, the data sent for video-stream processor.The display unit 40 can be the display in PC machineScreen, liquid crystal display or electric ink display screen etc..Specifically, in this example it is shown that unit 400 can be shown treatsPredict the information such as the prediction result of disk.
The network port 500, for being communicatively coupled with external each terminal device.Skill is communicated used by the communication connectionArt can be cable communicating technology or wireless communication technique, as mobile high definition chained technology (MHL), universal serial bus (USB),High-definition media interface (HDMI), Bluetooth Communication Technology, the low-power consumption bluetooth communication technology, is based on adopting wireless fidelity technology (WiFi)The communication technology etc. of IEEE802.11s.Specifically in the present embodiment, the network port 500 and other hosts be can useIt interacts, communicating predicted result, the performance data for obtaining disk to be predicted etc..
Present invention also provides a kind of computer readable storage mediums, are stored thereon with computer program, the computerStep provided by above-described embodiment may be implemented when program is executed by processor.The storage medium may include: USB flash disk, movementHard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory,RAM), the various media that can store program code such as magnetic or disk.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with otherThe difference of embodiment, the same or similar parts in each embodiment may refer to each other.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined hereinGeneral Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the inventionIt is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase oneThe widest scope of cause.