CN107463963A - A kind of Fault Classification and device - Google Patents
A kind of Fault Classification and device Download PDFInfo
- Publication number
- CN107463963A CN107463963A CN201710682972.7A CN201710682972A CN107463963A CN 107463963 A CN107463963 A CN 107463963A CN 201710682972 A CN201710682972 A CN 201710682972A CN 107463963 A CN107463963 A CN 107463963A
- Authority
- CN
- China
- Prior art keywords
- fault
- fault category
- category
- sorted
- state
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a kind of Fault Classification and device, including:Obtain fault sample to be sorted;Fault sample to be sorted is inputted into K (K 1)/2 SVMs (SVM) grader respectively, obtains K (K 1)/2 fail result;The fault category of fault sample to be sorted is determined according to K (K 1)/2 fail result.From the embodiment of the present invention, due to there is K (K 1)/2 SVM classifier, after fault sample to be sorted inputs this K (K 1)/2 SVM classifier, K (K 1)/2 fail result will be automatically generated, the fault category of fault sample to be sorted just can be finally determined according to this K (K 1)/2 fail result, and then solved in time according to fault category, so as to ensure that the stability of data center apparatus operation, security and the normal operation of miscellaneous service.
Description
Technical field
The present invention relates to self study field, more particularly to a kind of Fault Classification and device.
Background technology
With the rapid development of Internet, data center carries the functions such as the collection, storage and analysis of various data, one
Denier equipment breaks down, and staff just needs to be gone to analyze failure according to the alarm details of monitored item.
For veteran old employee, perhaps can be quickly find failure cause, then according to failure cause
Solve failure, but for the employee that experience is not very abundant, it is likely that when can not find failure cause or need very long
Between can just find failure cause, therefore can not ensure data center apparatus operation stability, security and miscellaneous service just
Often operation.
The content of the invention
In order to solve the above-mentioned technical problem, the present invention provides a kind of Fault Classification and device, can be automatically according to
Fault sample to be sorted determines fault category, ensure that stability, security and miscellaneous service that data center apparatus is run
Normal operation.
In order to reach the object of the invention, the invention provides a kind of Fault Classification, including:
Obtain fault sample to be sorted;
The fault sample to be sorted is inputted into K (K-1)/2 support vector machines grader respectively, obtain K (K-1)/
2 fail results;Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance, and K is more than 1
Integer;
The fault category of the fault sample to be sorted is determined according to the K (K-1)/2 fail result.
Before the acquisition fault sample to be sorted, in addition to:
Set K kind fault categories;
Trained to obtain the K (K-1)/2 SVM classifier according to the K kinds fault category of setting.
The K kind fault categories according to setting train to obtain K (K-1)/2 SVM classifier, including:
Obtain and represent the primary vector of the state of monitored item corresponding to i-th kind of fault category and expression jth kind fault category
The secondary vector of the state of corresponding monitored item;Wherein, i=1,2...K-1, j=2...K, i < j;
According to the i-th kind of fault category and primary vector, the jth kind fault category and secondary vector instruction
Get and be used for the svm classifier for differentiating i-th kind of fault category and jth kind fault category in the K (K-1)/2 SVM classifier
Device.
Before the acquisition causes the primary vector of i-th kind of fault category and the secondary vector for causing jth kind fault category,
Also include:
The state of monitored item corresponding to every kind of fault category in K kind fault categories is obtained respectively;
It is determined that represent the mark of the state of monitored item;
According to determination mark respectively by every kind of fault category in the K kinds fault category corresponding to monitored item state table
It is shown as vector.
The fault category that fault sample to be sorted is determined according to K (K-1)/2 fail result, including:
Obtain the fault category of each fail result in K (K-1)/2 fail result;
Count the number that the fault category of each fail result occurs;
The maximum fault category of times of acquisition in the number that the fault category of each fail result counted occurs,
As target faults classification;
The fault category for determining the fault sample to be sorted is the target faults classification.
Present invention also offers a kind of failure modes device, including:
Acquisition module, for obtaining fault sample to be sorted;
Processing module, for the fault sample to be sorted to be inputted into K (K-1)/2 support vector machines classification respectively
Device, obtain K (K-1)/2 fail result;Wherein, K (K-1)/2 SVM classifier is obtained according to K kind fault category training in advance
Arrive, K is the integer more than 1;
Determining module, for determining the failure classes of the fault sample to be sorted according to the K (K-1)/2 fail result
Not.
Also include:
Setting module, for setting K kind fault categories;
Training module, for being trained to obtain the K (K-1)/2 svm classifier according to the K kinds fault category of setting
Device.
The training module includes:
First acquisition unit, for obtain represent i-th kind of fault category corresponding to monitored item state primary vector and
Represent the secondary vector of the state of monitored item corresponding to jth kind fault category;Wherein, i=1,2...K-1, j=2...K, i <
j;
Training unit, for according to i-th kind of fault category and the primary vector, the jth kind fault category and
The secondary vector trains to obtain to be used to differentiate i-th kind of fault category and jth kind failure in the K (K-1)/2 SVM classifier
The SVM classifier of classification.
The training module also includes:
Second acquisition unit, for obtaining the state of monitored item corresponding to every kind of fault category in K kind fault categories respectively;
First determining unit, the mark of the state for determining to represent monitored item;
Processing unit, for according to the mark of determination respectively by every kind of fault category in the K kinds fault category corresponding to
The state representation of monitored item is vector.
The determining module includes:
3rd acquiring unit, for obtaining the fault category of each fail result in K (K-1)/2 fail result;
Statistic unit, the number that the fault category for counting each fail result occurs;
4th acquiring unit, for being obtained in the number that occurs in the fault category of each fail result counted time
The maximum fault category of number, as target faults classification;
Second determining unit, the fault category for determining the fault sample to be sorted are the target faults classification.
Compared with prior art, the present invention, which comprises at least, obtains fault sample to be sorted;Fault sample to be sorted is distinguished
K (K-1)/2 SVMs (Support Vector Machine, SVM) grader is inputted, obtains K (K-1)/2 failure
As a result;Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance;According to K (K-1)/2 event
Barrier result determines the fault category of fault sample to be sorted.From technical scheme provided by the invention, due to there is K (K-1)/2
Individual SVM classifier, after fault sample to be sorted inputs this K (K-1)/2 SVM classifier, K (K-1)/2 will be automatically generated
Individual fail result, wherein each fail result is one kind of K kind fault category kinds, according to this K (K-1)/2 fail result with regard to energy
The fault category of fault sample to be sorted is finally determined, and then is solved in time according to fault category, so as to ensure that number
According to the normal operation of the stability of central apparatus operation, security and miscellaneous service.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
Obtain it is clear that or being understood by implementing the present invention.The purpose of the present invention and other advantages can be by specification, rights
Specifically noted structure is realized and obtained in claim and accompanying drawing.
Brief description of the drawings
Accompanying drawing is used for providing further understanding technical solution of the present invention, and a part for constitution instruction, with this
The embodiment of application is used to explain technical scheme together, does not form the limitation to technical solution of the present invention.
Fig. 1 is a kind of schematic flow sheet of Fault Classification provided in an embodiment of the present invention;
Fig. 2 is the schematic flow sheet of another Fault Classification provided in an embodiment of the present invention;
Fig. 3 is the schematic flow sheet of another Fault Classification provided in an embodiment of the present invention;
Fig. 4 is a kind of structural representation of failure modes device provided in an embodiment of the present invention;
Fig. 5 is the structural representation of another failure modes device provided in an embodiment of the present invention;
Fig. 6 is the structural representation of another failure modes device provided in an embodiment of the present invention.
Embodiment
For the object, technical solutions and advantages of the present invention are more clearly understood, below in conjunction with accompanying drawing to the present invention
Embodiment be described in detail.It should be noted that in the case where not conflicting, in the embodiment and embodiment in the application
Feature can mutually be combined.
The embodiment of the present invention provides a kind of Fault Classification, as shown in figure 1, this method includes:
Step 101, obtain fault sample to be sorted.
Specifically, fault sample to be sorted is will be according to the information of its determination fault category.
Step 102, fault sample to be sorted is inputted to K (K-1)/2 support vector machines grader respectively, obtain K
(K-1)/2 fail result.
Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance, and K is whole more than 1
Number.
Specifically, it can train to obtain a grader, therefore basis according to every two classes fault category in K kind fault categories
Fault category can train to obtain K (K-1)/2 SVM classifier altogether in K.
Step 103, the fault category for determining according to K (K-1)/2 fail result fault sample to be sorted.
Specifically, each fail result is in K kind fault categories in K (K-1)/2 fail result, therefore according to K (K-
1)/2 a fail result just can determine that the fault category of fault sample to be sorted.
The Fault Classification that the embodiment of the present invention is provided, obtain fault sample to be sorted;By fault sample to be sorted
K (K-1)/2 support vector machines grader is inputted respectively, obtains K (K-1)/2 fail result;Wherein, K (K-1)/2
SVM classifier obtains according to K kind fault category training in advance;Failure to be sorted is determined according to K (K-1)/2 fail result
The fault category of sample.From technical scheme provided by the invention, due to there is K (K-1)/2 SVM classifier, when to be sorted
After fault sample inputs this K (K-1)/2 SVM classifier, K (K-1)/2 fail result will be automatically generated, wherein each event
Barrier result is one kind of K kind fault category kinds, and failure sample to be sorted just can be finally determined according to this K (K-1)/2 fail result
This fault category, and then solved in time according to fault category, so as to ensure that the stabilization of data center apparatus operation
The normal operation of property, security and miscellaneous service.
The embodiment of the present invention provides another Fault Classification, as shown in Fig. 2 this method includes:
Step 201, setting K kind fault categories.
Wherein, K is the integer more than 1.
Step 202, trained according to the K kind fault categories of setting to obtain K (K-1)/2 SVM classifier.
Specifically, step 202 can be realized by step 202a, 202b:
Step 202a, obtain and represent the primary vector of monitored item state corresponding to i-th kind of fault category and expression jth kind event
Hinder the secondary vector of the state of monitored item corresponding to classification.
Wherein, i=1,2...K-1, j=2...K, i < j.
Specifically, primary vector is vector of the mark for representing monitored item state corresponding to i-th kind of fault category, second
Vector is the vector that mark is used to represent monitored item state corresponding to jth kind fault category.Obtain and represent i-th kind of fault category pair
The secondary vector of the state of monitored item refers to corresponding to the primary vector and expression jth kind fault category of the monitored item state answered
It is:Obtain and represent the 1st kind of primary vector of monitored item state corresponding to fault category and represent to supervise corresponding to the 2nd kind of fault category
The secondary vector of the state of item is controlled, obtains and represents the 1st kind of primary vector of monitored item state corresponding to fault category and expression the 3rd
The secondary vector ... of the state of monitored item, which obtains, corresponding to kind fault category represents the 1st kind of monitored item shape corresponding to fault category
The primary vector of state and the secondary vector for representing the state of monitored item corresponding to K kind fault categories;Obtain and represent the 2nd kind of failure
The primary vector of monitored item state corresponding to classification and represent the 3rd kind of state of monitored item corresponding to fault category second to
Amount, obtain and represent the 2nd kind of primary vector of monitored item state corresponding to fault category and represent to supervise corresponding to the 4th kind of fault category
The secondary vector ... for controlling the state of item obtains expression the 2nd kind of primary vector of monitored item state and expression corresponding to fault category
The secondary vector of the state of monitored item corresponding to K kind fault categories;...;Obtain and represent to supervise corresponding to K-2 kinds fault category
Control the primary vector of item state and represent the secondary vector of the state of monitored item corresponding to K-1 kind fault categories, obtain and represent
The primary vector of monitored item state corresponding to K-2 kind fault categories and the shape for representing monitored item corresponding to K kind fault categories
The secondary vector of state;Obtain and represent the primary vector of monitored item state corresponding to K-1 kind fault categories and expression K kind failures
The secondary vector of the state of monitored item corresponding to classification.
Step 202b, train to obtain K according to i-th kind of fault category and primary vector, jth kind fault category and secondary vector
(K-1)/2 it is used for the SVM classifier for differentiating i-th kind of fault category and jth kind fault category in a SVM classifier.
Specifically, train to obtain K according to i-th kind of fault category and primary vector, jth kind fault category and secondary vector
(K-1)/2 it is used to differentiate that the SVM classifier of i-th kind of fault category and jth kind fault category refers in a SVM classifier:Root
Train to obtain K (K-1)/2 SVM classifier according to the 1st kind of fault category and primary vector, the 2nd kind of fault category and secondary vector
In be used for the SVM classifier that differentiates the 1st kind of fault category and the 2nd kind of fault category, according to the 1st kind of fault category and first to
Amount, the 3rd kind of fault category and secondary vector train to obtain to be used to differentiate the 1st kind of fault category in K (K-1)/2 SVM classifier
SVM classifier ... with the 3rd kind of fault category is according to the 1st kind of fault category and primary vector, K kinds fault category and second
Vector training obtains being used for SVM points that differentiate the 1st kind of fault category and K kind fault categories in K (K-1)/2 SVM classifier
Class device;Train to obtain K (K-1)/2 SVM according to the 2nd kind of fault category and primary vector, the 3rd kind of fault category and secondary vector
It is used for the SVM classifier for differentiating the 2nd kind of fault category and the 3rd kind of fault category in grader, according to the 2nd kind of fault category and the
One vector, the 4th kind of fault category and secondary vector train to obtain to be used to differentiate the 2nd kind of failure in K (K-1)/2 SVM classifier
Classification and the SVM classifier ... of the 4th kind of fault category according to the 2nd kind of fault category and primary vector, K kinds fault category and
Secondary vector trains to obtain to be used to differentiate the 2nd kind of fault category and K kind fault categories in K (K-1)/2 SVM classifier
SVM classifier;...;Train to obtain according to K-2 kinds fault category and primary vector, K-1 kinds fault category and secondary vector
It is used for the SVM classifier for differentiating K-2 kinds fault category and K-1 kind fault categories in K (K-1)/2 SVM classifier, according to
K-2 kinds fault category and primary vector, K kinds fault category and secondary vector train to obtain K (K-1)/2 SVM classifier
In be used for the SVM classifier that differentiates K-2 kinds fault category and K kind fault categories;According to K-1 kinds fault category and first
Vector, K kinds fault category and secondary vector train to obtain to be used to differentiate K-1 kind failures in K (K-1)/2 SVM classifier
The SVM classifier of classification and K kind fault categories.
Step 203, obtain fault sample to be sorted.
Step 204, fault sample to be sorted is inputted to K (K-1)/2 support vector machines grader respectively, obtain K
(K-1)/2 fail result.
Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance, therefore will classification event
Fail result obtained by each SVM classifier of barrier sample input is one kind in this K kind fault category.
Step 205, the fault category for determining according to K (K-1)/2 fail result fault sample to be sorted.
It should be noted that if the fault category currently divided can not meet the needs of client, new event is reset
Hinder classification, re -training distinguishes the SVM classifier of new fault category to differentiate new fault category.
The Fault Classification that the embodiment of the present invention is provided, obtain fault sample to be sorted;By fault sample to be sorted
K (K-1)/2 support vector machines grader is inputted respectively, obtains K (K-1)/2 fail result;Wherein, K (K-1)/2
SVM classifier obtains according to K kind fault category training in advance;Determine to be sorted according to K (K-1)/2 fail result therefore drop
The fault category of sample.From technical scheme provided by the invention, due to there is K (K-1)/2 SVM classifier, when to be sorted
After fault sample inputs this K (K-1)/2 SVM classifier, K (K-1)/2 fail result will be automatically generated, wherein each event
Barrier result is one kind of K kind fault category kinds, and failure sample to be sorted just can be finally determined according to this K (K-1)/2 fail result
This fault category, and then solved in time according to fault category, so as to ensure that the stabilization of data center apparatus operation
The normal operation of property, security and miscellaneous service.
The embodiment of the present invention provides another Fault Classification, as shown in figure 3, this method includes:
Step 301, setting K kind fault categories.
Wherein, K is the integer more than 1.
Step 302, the state for obtaining monitored item corresponding to every kind of fault category in K kind fault categories respectively.
Specifically, the state of monitored item corresponding to fault category refers to the monitored item that certain fault category can result in
State, the state of monitored item corresponding to every kind of fault category in K kind fault categories is obtained respectively and is referred to:Obtain K kind failure classes
Supervised in not in the 1st kind of state of monitored item corresponding to fault category, acquisition K kind fault categories corresponding to the 2nd kind of fault category
The state ... for controlling item obtains the state of monitored item corresponding to K kind fault categories in K kind fault categories.Assuming that monitored item includes
Fan sensor, voltage sensor, temperature sensor and hard disk sensor, to obtain the 1st kind of fault category in K kind fault categories
Illustrated exemplified by the state of corresponding monitored item, the 1st kind of state of monitored item corresponding to fault category refers in K kind fault categories
Be the 1st kind of state of fan sensor corresponding to fault category in K kind fault categories, the state of voltage sensor, temperature pass
The state of sensor and the state of hard disk sensor, it can result in the fan sensor of the 1st kind of fault category in K kind fault categories
State, the state of voltage sensor, the state of the state of temperature sensor and hard disk sensor are probably a kind of, it is also possible to more
Kind.
Step 303, determine to represent the mark of the state of monitored item.
Specifically, monitored item state can include normal condition, malfunction and other states, it is assumed that the state of monitored item
Including normal condition and malfunction, it is determined that representing that the mark of the state of monitored item refers to determining to represent monitored item normal condition
With the mark of malfunction, it is assumed that with mark, " 0,1 " represents the state of monitored item, and mark " 1 " represents that monitored item is in normal shape
State, mark " 0 " represent that monitored item is in malfunction, when some monitored item represents to identify " 1 ", illustrate the shape of the monitored item
State is in normal condition, when some monitored item represents to identify " 0 ", illustrates that the state of the monitored item is in malfunction.
Step 304, according to determination mark respectively by every kind of fault category in K kind fault categories corresponding to monitored item shape
State is expressed as vector.
Specifically, according to determination mark respectively by every kind of fault category in K kind fault categories corresponding to monitored item shape
State is expressed as vector and referred to:The shape of monitored item according to corresponding to the mark of determination by the 1st kind of fault category in K kind fault categories
State is expressed as vector, the state representation of monitored item according to corresponding to the mark of determination by the 2nd kind of fault category in K kind fault categories
For vector ... according to determination mark respectively by K kind fault categories in K kind fault categories corresponding to monitored item state representation
For vector.Assuming that monitored item includes fan sensor, voltage sensor, temperature sensor and hard disk sensor, it is assumed that with mark
" 0,1 " represents the state of monitored item, and mark " 1 " represents that monitored item is in normal condition, and mark " 0 " represents that monitored item is in failure
State, it is assumed that K is equal to 3, and fault category includes fan sensor fault, voltage sensor failure and temperature sensor fault, false
If causing the monitored item state of voltage sensor failure to include fan sensor fault, voltage sensor failure, temperature sensor
Normal and hard disk sensor is normal, and fan sensor is normal, voltage sensor failure, temperature sensor are normal and hard disk sensor
Normally;So by fan sensor fault, voltage sensor failure, temperature sensor be normal and hard disk sensor it is normal this
Group state can be expressed as vectorial (0 01 1), by fan sensor is normal, voltage sensor failure, temperature sensor are normal
It can be expressed as vectorial (1 01 1) with normal this group of state of hard disk sensor.
It should be noted that judging for monitored item state is based on all subclasses corresponding to a monitored item species.
Specifically, it is assumed that monitored item includes fan sensor, voltage sensor, temperature sensor and hard disk sensor, fan sensing
Device has 3, is expressed as fan sensor 1, fan sensor 2 and fan sensor 3;Voltage sensor has 2, respectively table
It is shown as voltage sensor 1 and voltage sensor 2;Temperature sensor has 3, is expressed as temperature sensor 1, temperature sensor
2 and temperature sensor 3;Hard disk sensor has 3, is expressed as hard disk sensor 1, hard disk sensor 2 and hard disk sensor
3;It is determined that each monitored item state when, can be judged according to the number of the specific monitored item included in each monitored item,
Illustrate by taking the state for fanning sensor as an example, be with this 3 sensors of fan sensor 1, fan sensor 2 and fan sensor 3
State determine the state of fan sensor, i.e., to this 3 sensings of fan sensor 1, fan sensor 2 and fan sensor 3
The state of device does "AND" processing, only in the presence of 3 fan sensors state all in it is normal when, the state of fan sensor is
Normal condition can be determined to be in, once the state that the state of this 3 fan sensors has a fan sensor is in event
Barrier state, the state of fan sensor will be determined to be in malfunction.
Step 305, obtain the primary vector of the state of monitored item corresponding to i-th kind of fault category of expression and represent jth kind
The secondary vector of the state of monitored item corresponding to fault category.
Wherein, i=1,2...K-1, j=2...K, i < j.
Specifically, the number of primary vector is exactly the state of monitored item corresponding to i-th kind of fault category in K kind fault categories
The represented number into vector, the number of secondary vector is exactly monitored item corresponding to jth kind fault category in K kind fault categories
The represented number into vector of state, the number of primary vector and the number of secondary vector are probably multiple.
Step 306, according to i-th kind of fault category and primary vector, jth kind fault category and secondary vector train to obtain K
(K-1)/2 it is used for the SVM classifier for differentiating i-th kind of fault category and jth kind fault category in a SVM classifier.
Specifically, SVM is proposed from the optimal classification line in the case of linear separability, so-called optimal classification line is exactly to require
Classifying line not only can be faultless separated by two class samples, and to make the distance between two classes maximum.
If linear separability sample set is (xi, yi), i=1,2...n, x ∈ Rd, y ∈ {+1, -1 } category label.D dimension spaces
The general type of middle linear discriminant function is:G (x)=ωTX+b, classification line equation are:
ωTX+b=0 (1)
If classification line is correctly classified to all samples, meet
yi(ωTxi+ b) -1 >=0 i=1,2...n (2)
Class interval maximum is equivalent to | | ω | |2Minimum, therefore optimal classification problem can be converted into constrained optimization problem, i.e.,
Under the constraint of formula (2), (1/2) ω of solved function formulaTω minimum values.It can be changed into antithesis from method of Lagrange multipliers to ask
Topic:
Wherein,Qij=yiyjK(xi, xj), K (xi, xj)=<xi, xj>
It is linear it is inseparable in the case of, it is necessary in conditional (2) one loose item ξ of increase, then constrained optimization problem
It is changed into:
Step 307, fault sample to be sorted is inputted to K (K-1)/2 support vector machines grader respectively, obtain K
(K-1)/2 fail result.
Step 308, the fault category for obtaining each fail result in K (K-1)/2 fail result.
Specifically, each fail result is in K kind fault categories in K (K-1)/2 fail result, therefore K (K-1)/2
The fault category of each fail result is one kind in this K kind fault category in individual fail result.
The number that step 309, the fault category of each fail result of statistics occur.
Specifically, due in K (K-1)/2 fail result each fail result be one kind in N kind fault categories, because
The number that this fault category that each fail result can be counted by the way of " ballot " is occurred, that is, count K (K-1)/2 event
How many fail result belongs to the 1st kind of fault category in N kind fault categories in barrier result, counts in K (K-1)/2 fail result
How many how many fail result belongs in N kind fault categories in the 2nd kind of fault category ... statistics K (K-1)/2 fail result
Fail result belongs to N kind fault categories in N kind fault categories.
Step 310, the times of acquisition maximum in the number of the fault category appearance of each fail result counted
Fault category, as target faults classification.
Specifically, assuming there are 3 kinds of fault categories, 6 fail results, the 1st failure are obtained according to fault sample to be sorted
As a result fault category is the 1st kind of fault category, and the fault category of the 2nd fail result is the 2nd kind of fault category, and the 3rd former
The fault category of barrier result is the 1st kind of fault category, and the fault category of the 4th fail result is the 3rd kind of fault category, the 5th
The fault category of fail result is the 1st kind of fault category, and the fault category of the 6th fail result is the 2nd kind of fault category, thus
It can be seen that the number that the 1st kind of fault category occurs is 3 times, the number that the 2nd kind of fault category occurs is 2 times, the 3rd kind of fault category
The number of appearance is 1 time, therefore the 1st kind of fault category is exactly the maximum fault category of number, i.e. target faults classification.
Step 311, the fault category for determining fault sample to be sorted are target faults classification.
The Fault Classification that the embodiment of the present invention is provided, obtain fault sample to be sorted;By fault sample to be sorted
K (K-1)/2 support vector machines grader is inputted respectively, obtains K (K-1)/2 fail result;Wherein, K (K-1)/2
SVM classifier obtains according to K kind fault category training in advance;Failure to be sorted is determined according to K (K-1)/2 fail result
The fault category of sample.From technical scheme provided by the invention, due to there is K (K-1)/2 SVM classifier, when to be sorted
After fault sample inputs this K (K-1)/2 SVM classifier, K (K-1)/2 fail result will be automatically generated, wherein each event
Barrier result is one kind of K kind fault category kinds, and failure sample to be sorted just can be finally determined according to this K (K-1)/2 fail result
This fault category, and then solved in time according to fault category, so as to ensure that the stabilization of data center apparatus operation
The normal operation of property, security and miscellaneous service.
The embodiment of the present invention provides a kind of failure modes device, as shown in figure 4, the sorter 4 includes:
Acquisition module 41, for obtaining fault sample to be sorted.
Processing module 42, for fault sample to be sorted to be inputted into K (K-1)/2 support vector machines grader respectively,
Obtain K (K-1)/2 fail result;Wherein, K (K-1)/2 SVM classifier is obtained according to K kind fault category training in advance
, K is the integer more than 1.
Determining module 43, for determining the fault category of fault sample to be sorted according to K (K-1)/2 fail result.
Further, on the basis of embodiment corresponding to Fig. 4, the embodiment of the present invention provides another failure modes device,
As shown in figure 5, the sorter 4 also includes:
Setting module 44, for setting K kind fault categories.
Training module 45, for being trained to obtain K (K-1)/2 SVM classifier according to the K kind fault categories of setting.
Training module 45 includes:
First acquisition unit 451, the primary vector of the state of monitored item corresponding to i-th kind of fault category is represented for obtaining
With the secondary vector for representing the state of monitored item corresponding to jth kind fault category;Wherein, i=1,2...K-1, j=2...K, i
< j.
Training unit 452, for being instructed according to i-th kind of fault category and primary vector, jth kind fault category and secondary vector
Get and be used for the SVM classifier for differentiating i-th kind of fault category and jth kind fault category in K (K-1)/2 SVM classifier.
Further, training module 45 also includes:
Second acquisition unit 453, for obtaining the shape of monitored item corresponding to every kind of fault category in K kind fault categories respectively
State.
First determining unit 454, the mark of the state for determining to represent monitored item.
Processing unit 455, for according to the mark of determination respectively by every kind of fault category in K kind fault categories corresponding to prison
The state representation for controlling item is vector.
Further, on the basis of embodiment corresponding to Fig. 5, the embodiment of the present invention provides another failure modes device,
As shown in fig. 6, determining module 43 includes:
3rd acquiring unit 431, for obtaining the fault category of each fail result in K (K-1)/2 fail result.
Statistic unit 432, the number that the fault category for counting each fail result occurs.
4th acquiring unit 433, for being obtained in the number that occurs in the fault category of each fail result counted
The maximum fault category of number is taken, as target faults classification.
Second determining unit 434, the fault category for determining fault sample to be sorted are target faults classification.
The failure modes device that the embodiment of the present invention is provided, obtains fault sample to be sorted;By fault sample to be sorted
K (K-1)/2 support vector machines grader is inputted respectively, obtains K (K-1)/2 fail result;Wherein, K (K-1)/2
SVM classifier obtains according to K kind fault category training in advance;Failure to be sorted is determined according to K (K-1)/2 fail result
The fault category of sample.From technical scheme provided by the invention, due to there is K (K-1)/2 SVM classifier, when to be sorted
After fault sample inputs this K (K-1)/2 SVM classifier, K (K-1)/2 fail result will be automatically generated, wherein each event
Barrier result is one kind of K kind fault category kinds, and failure sample to be sorted just can be finally determined according to this K (K-1)/2 fail result
This fault category, and then solved in time according to fault category, so as to ensure that the stabilization of data center apparatus operation
The normal operation of property, security and miscellaneous service.
In actual applications, the acquisition module 41, processing module 42, determining module 43, the 3rd acquiring unit 431, system
Count unit 432, the 4th acquiring unit 433, the second determining unit 434, setting module 44, training module 45, first acquisition unit
451st, training unit 452, second acquisition unit 453, the first determining unit 454, processing unit 455 can be by positioned at failure modes
Central processing unit (Central Processing Unit, CPU), microprocessor (Micro Processor in device
Unit, MPU), digital signal processor (Digital Signal Processor, DSP) or field programmable gate array
(Field Programmable Gate Array, FPGA) etc. is realized.
Although disclosed herein embodiment as above, described content be only readily appreciate the present invention and use
Embodiment, it is not limited to the present invention.Technical staff in any art of the present invention, taken off not departing from the present invention
On the premise of the spirit and scope of dew, any modification and change, but the present invention can be carried out in the form and details of implementation
Scope of patent protection, still should be subject to the scope of the claims as defined in the appended claims.
Claims (10)
- A kind of 1. Fault Classification, it is characterised in that including:Obtain fault sample to be sorted;The fault sample to be sorted is inputted into K (K-1)/2 support vector machines grader respectively, obtains K (K-1)/2 Fail result;Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance, and K is whole more than 1 Number;The fault category of the fault sample to be sorted is determined according to the K (K-1)/2 fail result.
- 2. sorting technique according to claim 1, it is characterised in that before the acquisition fault sample to be sorted, also wrap Include:Set K kind fault categories;Trained to obtain the K (K-1)/2 SVM classifier according to the K kinds fault category of setting.
- 3. sorting technique according to claim 2, it is characterised in that the K kind fault categories according to setting are trained To K (K-1)/2 SVM classifier, including:Obtain and represent that the primary vector of the state of monitored item corresponding to i-th kind of fault category is corresponding with jth kind fault category is represented Monitored item state secondary vector;Wherein, i=1,2...K-1, j=2...K, i < j;Trained according to the i-th kind of fault category and primary vector, the jth kind fault category and secondary vector It is used for the SVM classifier for differentiating i-th kind of fault category and jth kind fault category into the K (K-1)/2 SVM classifier.
- 4. sorting technique according to claim 3, it is characterised in that the acquisition causes the first of i-th kind of fault category Before vector sum causes the secondary vector of jth kind fault category, in addition to:The state of monitored item corresponding to every kind of fault category in K kind fault categories is obtained respectively;It is determined that represent the mark of the state of monitored item;According to the mark of determination respectively by every kind of fault category in the K kinds fault category corresponding to the state representation of monitored item be Vector.
- 5. according to the sorting technique described in claim any one of 1-4, it is characterised in that described according to K (K-1)/2 failure knot Fruit determines the fault category of fault sample to be sorted, including:Obtain the fault category of each fail result in K (K-1)/2 fail result;Count the number that the fault category of each fail result occurs;The maximum fault category of times of acquisition in the number that the fault category of each fail result counted occurs, as Target faults classification;The fault category for determining the fault sample to be sorted is the target faults classification.
- A kind of 6. failure modes device, it is characterised in that including:Acquisition module, for obtaining fault sample to be sorted;Processing module, for the fault sample to be sorted to be inputted into K (K-1)/2 support vector machines grader respectively, obtain To K (K-1)/2 fail result;Wherein, K (K-1)/2 SVM classifier obtains according to K kind fault category training in advance, K is the integer more than 1;Determining module, for determining the fault category of the fault sample to be sorted according to the K (K-1)/2 fail result.
- 7. sorter according to claim 6, it is characterised in that also include:Setting module, for setting K kind fault categories;Training module, for being trained to obtain the K (K-1)/2 SVM classifier according to the K kinds fault category of setting.
- 8. sorter according to claim 7, it is characterised in that the training module includes:First acquisition unit, the primary vector of the state of monitored item and expression corresponding to i-th kind of fault category are represented for obtaining The secondary vector of the state of monitored item corresponding to jth kind fault category;Wherein, i=1,2...K-1, j=2...K, i < j;Training unit, for according to i-th kind of fault category and the primary vector, the jth kind fault category and described Secondary vector trains to obtain to be used to differentiate i-th kind of fault category and jth kind fault category in the K (K-1)/2 SVM classifier SVM classifier.
- 9. sorter according to claim 8, it is characterised in that the training module also includes:Second acquisition unit, for obtaining the state of monitored item corresponding to every kind of fault category in K kind fault categories respectively;First determining unit, the mark of the state for determining to represent monitored item;Processing unit, for according to the mark of determination respectively by every kind of fault category in the K kinds fault category corresponding to monitoring The state representation of item is vector.
- 10. according to the sorter described in claim any one of 6-9, it is characterised in that the determining module includes:3rd acquiring unit, for obtaining the fault category of each fail result in K (K-1)/2 fail result;Statistic unit, the number that the fault category for counting each fail result occurs;4th acquiring unit, for times of acquisition in the number that occurs in the fault category of each fail result counted most Big fault category, as target faults classification;Second determining unit, the fault category for determining the fault sample to be sorted are the target faults classification.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710682972.7A CN107463963A (en) | 2017-08-10 | 2017-08-10 | A kind of Fault Classification and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710682972.7A CN107463963A (en) | 2017-08-10 | 2017-08-10 | A kind of Fault Classification and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107463963A true CN107463963A (en) | 2017-12-12 |
Family
ID=60548704
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710682972.7A Pending CN107463963A (en) | 2017-08-10 | 2017-08-10 | A kind of Fault Classification and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107463963A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108153618A (en) * | 2017-12-22 | 2018-06-12 | 国网浙江杭州市萧山区供电有限公司 | Hard disk data recovery, device and hard disc data restorer |
CN109086187A (en) * | 2018-07-26 | 2018-12-25 | 郑州云海信息技术有限公司 | A kind of performance alarm method and apparatus |
CN109840183A (en) * | 2018-12-05 | 2019-06-04 | 平安科技(深圳)有限公司 | Data center's grading forewarning system method, apparatus and storage medium |
CN115031357A (en) * | 2022-05-10 | 2022-09-09 | 南京信息工程大学 | Novel voting strategy-based fault diagnosis method suitable for different types of fault characteristics |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103810374A (en) * | 2013-12-09 | 2014-05-21 | 中国矿业大学 | Machine fault prediction method based on MFCC feature extraction |
CN104502103A (en) * | 2014-12-07 | 2015-04-08 | 北京工业大学 | Bearing fault diagnosis method based on fuzzy support vector machine |
CN105046279A (en) * | 2015-08-07 | 2015-11-11 | 合肥工业大学 | Analog circuit fault mode classification method |
CN105046210A (en) * | 2015-07-01 | 2015-11-11 | 西安理工大学 | Multi-label diagnostic method for train air conditioning faults |
CN105184316A (en) * | 2015-08-28 | 2015-12-23 | 国网智能电网研究院 | Support vector machine power grid business classification method based on feature weight learning |
CN106443259A (en) * | 2016-09-29 | 2017-02-22 | 国网山东省电力公司电力科学研究院 | Transformer fault diagnosis new method based on Euclidean clustering and SPO-SVM |
-
2017
- 2017-08-10 CN CN201710682972.7A patent/CN107463963A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103810374A (en) * | 2013-12-09 | 2014-05-21 | 中国矿业大学 | Machine fault prediction method based on MFCC feature extraction |
CN104502103A (en) * | 2014-12-07 | 2015-04-08 | 北京工业大学 | Bearing fault diagnosis method based on fuzzy support vector machine |
CN105046210A (en) * | 2015-07-01 | 2015-11-11 | 西安理工大学 | Multi-label diagnostic method for train air conditioning faults |
CN105046279A (en) * | 2015-08-07 | 2015-11-11 | 合肥工业大学 | Analog circuit fault mode classification method |
CN105184316A (en) * | 2015-08-28 | 2015-12-23 | 国网智能电网研究院 | Support vector machine power grid business classification method based on feature weight learning |
CN106443259A (en) * | 2016-09-29 | 2017-02-22 | 国网山东省电力公司电力科学研究院 | Transformer fault diagnosis new method based on Euclidean clustering and SPO-SVM |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108153618A (en) * | 2017-12-22 | 2018-06-12 | 国网浙江杭州市萧山区供电有限公司 | Hard disk data recovery, device and hard disc data restorer |
CN109086187A (en) * | 2018-07-26 | 2018-12-25 | 郑州云海信息技术有限公司 | A kind of performance alarm method and apparatus |
CN109840183A (en) * | 2018-12-05 | 2019-06-04 | 平安科技(深圳)有限公司 | Data center's grading forewarning system method, apparatus and storage medium |
CN115031357A (en) * | 2022-05-10 | 2022-09-09 | 南京信息工程大学 | Novel voting strategy-based fault diagnosis method suitable for different types of fault characteristics |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106502234B (en) | Industrial control system method for detecting abnormality based on double skeleton patterns | |
CN108322347B (en) | Data detection method, device, detection server and storage medium | |
CN107463963A (en) | A kind of Fault Classification and device | |
CN104598367B (en) | Data center's event of failure management specialty and method | |
CN113597634B (en) | Process mapping and monitoring using artificial intelligence | |
CN106844161B (en) | Abnormity monitoring and predicting method and system in calculation system with state flow | |
JP5098821B2 (en) | Monitoring device and monitoring method for detecting a sign of failure of monitored system | |
CN113645232B (en) | Intelligent flow monitoring method, system and storage medium for industrial Internet | |
CN107291911A (en) | A kind of method for detecting abnormality and device | |
CN106506556A (en) | A kind of network flow abnormal detecting method and device | |
CN110366031A (en) | A kind of monitoring abnormal state and method for diagnosing faults digitizing workshop MES system view-based access control model | |
CN108710555A (en) | A kind of server error diagnosis method based on supervised learning | |
AU2019275633B2 (en) | System and method of automated fault correction in a network environment | |
CN104050361A (en) | Intelligent analysis early warning method for dangerousness tendency of prison persons serving sentences | |
CN110297207A (en) | Method for diagnosing faults, system and the electronic device of intelligent electric meter | |
CN106845526A (en) | A kind of relevant parameter Fault Classification based on the analysis of big data Fusion of Clustering | |
CN112163572A (en) | Method and device for identifying object | |
CN110502677A (en) | A kind of device identification method, device and equipment, storage medium | |
KR101281460B1 (en) | Method for anomaly detection using statistical process control | |
CN114168374A (en) | Abnormal data positioning method based on integration of abnormal detector and root cause analysis | |
CN114912678A (en) | Online automatic detection and early warning method and system for abnormal operation of power grid regulation and control | |
Rafique et al. | TSDN-enabled network assurance: A cognitive fault detection architecture | |
CN106933097A (en) | A kind of Fault Diagnosis for Chemical Process method based on multi-level optimization PCC SDG | |
CN116756659A (en) | Intelligent operation and maintenance management method, device, equipment and storage medium | |
CN116302809A (en) | Edge end data analysis and calculation device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171212 |
|
RJ01 | Rejection of invention patent application after publication |