CN108052974A - A kind of method for diagnosing faults, system, equipment and storage medium - Google Patents
A kind of method for diagnosing faults, system, equipment and storage medium Download PDFInfo
- Publication number
- CN108052974A CN108052974A CN201711320019.4A CN201711320019A CN108052974A CN 108052974 A CN108052974 A CN 108052974A CN 201711320019 A CN201711320019 A CN 201711320019A CN 108052974 A CN108052974 A CN 108052974A
- Authority
- CN
- China
- Prior art keywords
- data set
- characteristic
- data
- distances
- fault
- 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.)
- Granted
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Abstract
This application discloses a kind of method for diagnosing faults, system, equipment and storage medium, this method includes:With initial characteristics collection corresponding characteristic data set of the target device in normal operation and failure operation is obtained respectively, obtains the training data for including normal characteristics data set and fault signature data set;Corresponding KL distances between each characteristic characteristic corresponding with fault signature data set are calculated in normal characteristics data set respectively, obtain KL distance sets;Cross validation is carried out using support vector cassification on the training data;According to the size of KL distances in verification result and KL distance sets, from initial characteristics concentration determine with the relevant feature of failure operation, obtain optimal characteristics collection;When the diagnostic data for getting target device, then characteristic corresponding with optimal characteristics collection is determined from diagnostic data, corresponding fault diagnosis then is carried out to target device using this feature data.The application effectively improves the accuracy of consequent malfunction diagnostic result.
Description
Technical field
The present invention relates to fault diagnosis technology field, more particularly to a kind of method for diagnosing faults, system, equipment and storage are situated between
Matter.
Background technology
Currently, people carry out when judging whether an equipment breaks down typically by way of artificial judgment
It determines, so on the one hand due to the influence of artificial subjective factor, the reliability for causing fault diagnosis result is unstable, on the other hand
It then needs to consume substantial amounts of cost of labor, and diagnosis efficiency is relatively low.In order to reduce cost of labor, diagnosis efficiency is improved, now
People gradually start that fault diagnosis is unfolded to equipment using support vector machines, but the accuracy of diagnostic result need into one
The promotion of step.
To sum up, how to promote the accuracy of fault diagnosis result is current urgent problem to be solved.
The content of the invention
In view of this, it is an object of the invention to provide a kind of method for diagnosing faults, system, equipment and storage medium, energy
Enough promote the accuracy of fault diagnosis result.Its concrete scheme is as follows:
In a first aspect, the invention discloses a kind of method for diagnosing faults, including:
With initial characteristics collection corresponding characteristic data set of the target device in normal operation and failure operation is obtained respectively,
Obtain the training data for including normal characteristics data set and fault signature data set;Wherein, the initial characteristics collection includes a variety of
Feature;
It is corresponding with the fault signature data set that each characteristic in the normal characteristics data set is calculated respectively
Corresponding KL distances, obtain KL distance sets between characteristic;
Cross validation is carried out on the training data using support vector cassification, is verified result;
According to the size of KL distances in the verification result and the KL distance sets, concentrate and determine from the initial characteristics
Go out with the relevant feature of failure operation, obtain optimal characteristics collection;
When the diagnostic data for getting the target device, then determined from the diagnostic data with it is described optimal
Then the corresponding characteristic of feature set carries out corresponding fault diagnosis using this feature data to the target device.
Optionally, it is described to calculate each characteristic and the fault signature data in the normal characteristics data set respectively
Concentrate between corresponding characteristic corresponding KL apart from the step of, including:
It determines respectively corresponding to each characteristic in the normal characteristics data set and the fault signature data set
Gaussian Profile;
Each corresponding Gaussian Profile of characteristic and the fault signature in the normal characteristics data set are calculated respectively
KL distances in data set between the corresponding Gaussian Profile of corresponding characteristic.
Optionally, each feature determined respectively in the normal characteristics data set and the fault signature data set
The step of Gaussian Profile corresponding to data, including:
Standard is carried out to each characteristic in the normal characteristics data set and the fault signature data set respectively
Change is handled, after then determining each standardization in the normal characteristics data set and the fault signature data set respectively
Characteristic corresponding to Gaussian Profile.
Optionally, it is described using support vector cassification on the training data carry out cross validation the step of, including:
Ten folding cross validations are carried out on the training data using support vector cassification.
Optionally, the size according to KL distances in the verification result and the KL distance sets, from described initial
It is determined in feature set with the relevant feature of failure operation, the step of obtaining optimal characteristics collection, including:
Screening meets the KL distances of preset condition from the KL distance sets, obtains target KL distances;
It is concentrated from the initial characteristics and screens the target KL apart from corresponding feature, obtain target signature;
According to the verification result from the target signature determine with the relevant feature of failure operation, obtain optimal characteristics
Collection.
Optionally, the screening from the KL distance sets meet the KL of preset condition apart from the step of, including:
Descending sort, distance set after being sorted are carried out to the KL distance sets;
The KL distances for the default quantity for coming front are filtered out from distance set after the sequence.
Optionally, the screening from the KL distance sets meet the KL of preset condition apart from the step of, including:
The KL distances that KL distances are more than predetermined threshold value are filtered out from the KL distance sets.
Second aspect, the invention discloses a kind of fault diagnosis system, including:
Characteristic acquisition module, for obtain respectively target device in normal operation and failure operation with initial spy
Corresponding characteristic data set is collected, obtains the training data for including normal characteristics data set and fault signature data set;Wherein, institute
Stating initial characteristics collection includes various features;
KL distance calculation modules, for calculating each characteristic and the failure in the normal characteristics data set respectively
Characteristic concentrates corresponding KL distances between corresponding characteristic, obtains KL distance sets;
Cross validation module for carrying out cross validation on the training data using support vector cassification, obtains
Verification result;
Characteristic determination module, for the size according to KL distances in the verification result and the KL distance sets, from institute
State initial characteristics concentration determine with the relevant feature of failure operation, obtain optimal characteristics collection;
Fault diagnosis module, for working as the diagnostic data for getting the target device, then from the diagnostic data
In determine characteristic corresponding with the optimal characteristics collection, then using this feature data to the target device carry out phase
The fault diagnosis answered.
The third aspect, the invention discloses a kind of failure diagnosis apparatus, including processor and memory;Wherein, the place
Foregoing disclosed method for diagnosing faults is realized during the fault diagnostic program that reason device execution is stored in the memory.
Fourth aspect, the invention discloses a kind of computer readable storage medium, for storing fault diagnostic program;Its
In, the fault diagnostic program realizes foregoing disclosed method for diagnosing faults when being executed by processor.
As it can be seen that it is of the invention after the training data comprising normal characteristics data set and fault signature data set is got,
Calculate in normal characteristics data set in each characteristic and fault signature data set corresponding KL between corresponding characteristic
Distance, and cross validation is carried out on the training data using support vector cassification, subsequently according to above-mentioned verification result and
The above-mentioned KL distances calculated, from initial characteristics concentration determine with the relevant feature of failure operation, wait to diagnose when getting
After data, can be determined from diagnostic data with above-mentioned characteristic corresponding with the relevant feature of failure operation,
Since whether these characteristics can reflect equipment in failure operation state, so can subsequently utilize these features
Data carry out fault diagnosis to equipment.It can be seen that the present invention to equipment before fault diagnosis is carried out, it is first true based on KL distances
Make with the relevant feature of failure operation, since these features can reflect failure operation feature, thus effectively improve
The accuracy of consequent malfunction diagnostic result.
Description of the drawings
It in order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of invention, for those of ordinary skill in the art, without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of method for diagnosing faults flow chart disclosed by the embodiments of the present invention;
Fig. 2 a are the monitored results schematic diagram using " failure 21 " during SVM;
Fig. 2 b are the monitored results schematic diagram using " failure 21 " during KL-FS-SVM;
Fig. 3 is a kind of fault diagnosis system structure diagram disclosed by the embodiments of the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without making creative work
Embodiment belongs to the scope of protection of the invention.
Shown in Figure 1 the embodiment of the invention discloses a kind of method for diagnosing faults, this method includes:
Step S11:With initial characteristics collection corresponding spy of the target device in normal operation and failure operation is obtained respectively
Data set is levied, obtains the training data for including normal characteristics data set and fault signature data set;Wherein, the initial characteristics collection
Including various features.
It is it is pointed out that corresponding with initial characteristics collection when obtaining target device in failure operation in the present embodiment
Can be corresponding with initial characteristics collection when acquisition target device is run under a type of failure when characteristic data set
Characteristic data set, so finally obtained a kind of corresponding fault signature data set;Certainly, the present embodiment can also obtain mesh
Characteristic data set corresponding with initial characteristics collection when marking device is run under the failure of K types, has so finally obtained phase
The K kind fault signature data sets answered.Wherein, above-mentioned K values are greater than 1 integer.
Step S12:Each characteristic and the fault signature data set in the normal characteristics data set are calculated respectively
In corresponding KL distances (i.e. Kullback-Leibler divergence) between corresponding characteristic, obtain KL distance sets.
It is above-mentioned to calculate each characteristic and the fault signature in the normal characteristics data set respectively in the present embodiment
In data set between corresponding characteristic corresponding KL apart from the step of, can specifically include:
It determines respectively corresponding to each characteristic in the normal characteristics data set and the fault signature data set
Gaussian Profile;It is special that each corresponding Gaussian Profile of characteristic and the failure in the normal characteristics data set are calculated respectively
Levy the KL distances between the corresponding Gaussian Profile of corresponding characteristic in data set.
There may be different dimension and the order of magnitude between the different characteristics that consideration initial acquisition arrives, for this purpose, this
Embodiment is in the corresponding Gaussian Profile of definite characteristic, it is necessary to be first standardized to characteristic.It is that is, above-mentioned
The Gauss point corresponding to each characteristic in the normal characteristics data set and the fault signature data set is determined respectively
The step of cloth, can specifically include:
Standard is carried out to each characteristic in the normal characteristics data set and the fault signature data set respectively
Change is handled, after then determining each standardization in the normal characteristics data set and the fault signature data set respectively
Characteristic corresponding to Gaussian Profile.
It is understood that for the failure of each type, the present embodiment can be directed to corresponding fault signature data
The process flow disclosed in above-mentioned steps S12 and below step S13, S14 and S15 is unfolded in collection.
Step S13:Cross validation is carried out on the training data using support vector cassification, is verified result.
Wherein, it is above-mentioned using support vector cassification on the training data carry out cross validation the step of, specifically may be used
To include:Ten folding cross validations are carried out on the training data using support vector cassification.
Step S14:According to the size of KL distances in the verification result and the KL distance sets, from the initial characteristics
Concentration determine with the relevant feature of failure operation, obtain optimal characteristics collection.
In the present embodiment, the above-mentioned size according to KL distances in the verification result and the KL distance sets, from described
Initial characteristics concentration is determined, with the relevant feature of failure operation, the step of obtaining optimal characteristics collection, can specifically include:
Screening meets the KL distances of preset condition from the KL distance sets, obtains target KL distances;From the initial spy
The target KL is screened in collection apart from corresponding feature, obtains target signature;It is special from the target according to the verification result
Sign determine with the relevant feature of failure operation, obtain optimal characteristics collection.
In a kind of specific embodiment, the above-mentioned screening from the KL distance sets meets the step of the KL distances of preset condition
Suddenly, can specifically include:
Descending sort, distance set after being sorted are carried out to the KL distance sets;From being screened after the sequence in distance set
Go out to come the KL distances of the default quantity of front.
It is understood that above-mentioned default quantity can specifically be set according to actual needs, it is not carried out herein
It limits.
In another embodiment specific implementation mode, the above-mentioned screening from the KL distance sets meets the KL distances of preset condition
Step can specifically include:
The KL distances that KL distances are more than predetermined threshold value are filtered out from the KL distance sets.
It is understood that above-mentioned predetermined threshold value can specifically be set according to actual needs, it is not carried out herein
It limits.
Step S15:When the diagnostic data for getting the target device, then determined from the diagnostic data with
Then the corresponding characteristic of the optimal characteristics collection carries out corresponding failure to the target device using this feature data and examines
It is disconnected.
In the present embodiment, determined from the diagnostic data characteristic corresponding with the optimal characteristics collection it
Afterwards, this feature data, and combination supporting vector machine (SVM, i.e. Support Vector Machine) can be utilized, target is set
It is standby to carry out corresponding fault diagnosis.
As it can be seen that the embodiment of the present invention is getting the training data comprising normal characteristics data set and fault signature data set
Afterwards, calculate corresponding between corresponding characteristic in each characteristic and fault signature data set in normal characteristics data set
KL distances, and cross validation is carried out on the training data using support vector cassification, subsequently according to above-mentioned verification result
And the above-mentioned KL distances calculated, from initial characteristics concentration determine with the relevant feature of failure operation, treated when getting
After diagnostic data, it can be determined from diagnostic data and above-mentioned characteristic corresponding with the relevant feature of failure operation
According to since whether these characteristics can reflect equipment in failure operation state, so can subsequently utilize these
Characteristic carries out fault diagnosis to equipment.It can be seen that the embodiment of the present invention to equipment carry out fault diagnosis before, first base
In KL distances determine with the relevant feature of failure operation, since these features can reflect failure operation feature, thus have
Improve to effect the accuracy of consequent malfunction diagnostic result.
Based on previous embodiment, the embodiment of the present invention is in Tennessee Yi Siman processes (Tennessee-Eastman
Process, TEP) corresponding test has been carried out on data set.The corresponding characteristic data set of normal operation is included in TEP data sets
Characteristic data set corresponding with 21 kinds of different faults.For each failure, training set has 480 fault signature data, test
Collection includes 960 observation data, and each data of observing include 52 variables, and the data of test set are started with normal data, to the
161 samplings are broken down, and every sampling in 3 minutes once, all data are generated all data by TEP simulation softwares.This
Embodiment takes 480 training datas of 500 training datas and a kind of failure in normal characteristics data set as training set
Input carries out fault detect to the test set of each failure.Specific implementation step is as follows:
1) to the normal characteristics data set in the industrial process that has been collected intoAnd failure
Characteristic data setN0And NkIt is normal characteristics data set and kth kind fault signature data respectively
The sample number of collection, m are Characteristic Numbers, and n is fault category number, here N1=500, Nk=480, m=52, n=21, then it is right respectively
Characteristic is standardized pretreatment.Standardizing formula is:
Wherein,For the average of j-th of characteristic of normal characteristics data set,For j-th of spy of normal characteristics data set
Levy the standard deviation of data.
2) assume that each characteristic obeys different Gaussian Profiles, respectively to the individual features number of different classes of data
According toEstimate its generalized Gaussian distribution, the distribution of each characteristic can be near by its Gaussian density function
Seemingly obtain:
Wherein,Different parametersWithRepresent different Gaussian density functions, i.e., its
Gaussian Profile is also different.So the present invention proposes the Gaussian Profile of j-th of characteristic of kth class by parameterWith
It determines,WithIt is then to be obtained by the maximum likelihood estimate to Gaussian density function.Maximum likelihood estimate calculates such as
Under:
Wherein Ψ (z)=Φ ' (z)/Φ (z) first passes through above-mentioned formula and acquiresWith
It is initialized first with Moment method estimatorsNewton iterative method is used againIt acquiresF (β) calculates as follows:
It obtainsAfterwards,Calculating it is as follows:
It acquiresWithIt can obtain the Gaussian Profile of j-th of characteristic of kth class.
3) KL divergences measure the distance of two distribution P and Q, calculation formula:So by feature j's
Gaussian density function is substituted into KL distances, obtains feature j respectively under normal and kth class fault condition between individual features data
KL distances:
4) for same fault type k, the corresponding KL distances of different characteristic j are expressed asIt is rightIt is ranked up, accordingly sorts from big to small
To feature set be then denoted as R.
5) ten folding cross validations are being carried out using support vector cassification on the training data, classifying quality is taken from R most
Good character subset F is optimal characteristics collection.
After above-mentioned optimal characteristics collection is obtained, the present embodiment can be tested based on test set.Specifically:
1) test data of real-time collecting industrial process(m is characterized number) herein, has
960 test samples, characteristic m=52, and test data is standardized, the corresponding formula that standardizes is:
2) feature that test data is chosen according to foregoing obtained optimal characteristics collection F forms input data, then with support to
Amount machine is classified, and is exported as a result, judging whether test sample is faulty, if belong to such failure.
The feature selection approach based on KL distances proposed by the embodiment of the present invention, with the normal characteristics data set of TEP
With fault signature data set as training data, the fault test data set of TEP is tested, it is found that the embodiment of the present invention can be to every
Class failure effectively completes feature selecting, finds out key feature, rejects useless feature, so as to improve the performance of grader.Experiment
Show feature selection approach based on KL distances that the embodiment of the present invention proposes and support vector machines combine (i.e. KL-FS-SVM) can
To improve the fault diagnosis result of traditional support vector machines (SVM), simultaneously for complicated process data, based on KL distances
The performance of feature selection approach will be far superior to some traditional feature selection approach (such as Fscore and Relief).Different faults
Fault diagnosis result as shown in Table 1, the present embodiment find Fscore-SVM and Relief-SVM " failure 20 " verification and measurement ratio
There is no the height of SVM, show to cause that the aspect ratio of the failure is more and difference is little, the feature selecting side based on KL distances of proposition
Method selects the SVM that the verification and measurement ratio outline that preceding 47 features obtain is higher than no feature selecting, this illustrates that the embodiment of the present invention proposes
Feature selection approach validity.And in the fault diagnosis model of KL-FS-SVM, the detection of " failure 11 " and " failure 21 "
Rate is all greatly improved.
Table one
Fault type | SVM | Fscore‐SVM | Relief‐SVM | KL‐FS‐SVM |
Failure 11 | 83.50% | 76.88% | 84.75% | 87.88% |
Failure 20 | 80.63% | 79.00% | 78.88% | 80.75% |
Failure 21 | 12.88% | 13.38% | 42.13% | 100% |
Table two
Fault type | Fscore | Relief | KL‐FS |
Failure 11 | 43 | 15 | 2 |
Failure 20 | 27 | 29 | 47 |
Failure 21 | 33 | 35 | 1 |
In addition to fault detect rate, referring to shown in table two, tri- kinds of feature selectings of Fscore, Relief and KL-FS are compared
The Characteristic Number for the optimal characteristics collection that method obtains, the Characteristic Number of KL-FS selections is minimum, and largely improves
The diagnosis performance of the diagnostic result of SVM, particularly " failure 21 " is the most apparent, as shown in Figure 2 a and 2 b.Wherein, Fig. 2 a are to make
With the monitored results schematic diagram of " failure 21 " during SVM, Fig. 2 b are to be illustrated using the monitored results of " failure 21 " during KL-FS-SVM
Figure.
Correspondingly, the embodiment of the invention also discloses a kind of fault diagnosis systems, shown in Figure 3, which includes:
Characteristic acquisition module 11, for obtain respectively target device in normal operation and failure operation with it is initial
The corresponding characteristic data set of feature set obtains the training data for including normal characteristics data set and fault signature data set;Wherein,
The initial characteristics collection includes various features;
KL distance calculation modules 12, for calculating each characteristic and the event in the normal characteristics data set respectively
Hinder characteristic and concentrate corresponding KL distances between corresponding characteristic, obtain KL distance sets;
Cross validation module 13 for carrying out cross validation on the training data using support vector cassification, obtains
To verification result;
Characteristic determination module 14, for the size according to KL distances in the verification result and the KL distance sets, from
Initial characteristics concentration determine with the relevant feature of failure operation, obtain optimal characteristics collection;
Fault diagnosis module 15, for working as the diagnostic data for getting the target device, then from the number to be diagnosed
Characteristic corresponding with the optimal characteristics collection is determined in, then the target device is carried out using this feature data
Corresponding fault diagnosis.
The corresponding contents disclosed in previous embodiment are may be referred on the more specifical course of work of above-mentioned modules,
It is no longer repeated herein.
Further, the invention also discloses a kind of failure diagnosis apparatus, including processor and memory;Wherein, it is described
Foregoing disclosed method for diagnosing faults is realized during the fault diagnostic program that processor execution is stored in the memory.On this
The more specific step of method may be referred to the corresponding contents disclosed in previous embodiment, no longer be repeated herein.
Further, the invention also discloses a kind of computer readable storage medium, for storing fault diagnostic program;Its
In, the fault diagnostic program realizes foregoing disclosed method for diagnosing faults when being executed by processor.More have on this method
The step of body, may be referred to the corresponding contents disclosed in previous embodiment, no longer be repeated herein.
Each embodiment is described by the way of progressive in this specification, the highlights of each of the examples are with it is other
The difference of embodiment, just to refer each other for same or similar part between each embodiment.For dress disclosed in embodiment
For putting, since it is corresponded to the methods disclosed in the examples, so description is fairly simple, related part is referring to method part
Explanation.
Professional further appreciates that, with reference to each exemplary unit of the embodiments described herein description
And algorithm steps, can be realized with the combination of electronic hardware, computer software or the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is performed actually with hardware or software mode, specific application and design constraint depending on technical solution.Specialty
Technical staff can realize described function to each specific application using distinct methods, but this realization should not
Think beyond the scope of this invention.
It can directly be held with reference to the step of method or algorithm that the embodiments described herein describes with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, term " comprising ", "comprising" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or equipment including a series of elements not only include that
A little elements, but also including other elements that are not explicitly listed or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except also there are other identical elements in the process, method, article or apparatus that includes the element.
Detailed Jie has been carried out to a kind of method for diagnosing faults provided by the present invention, system, equipment and storage medium above
It continues, specific case used herein is set forth the principle of the present invention and embodiment, and the explanation of above example is only
It is the method and its core concept for being used to help understand the present invention;Meanwhile for those of ordinary skill in the art, according to this hair
Bright thought, there will be changes in specific embodiments and applications, in conclusion this specification content should not manage
It solves as limitation of the present invention.
Claims (10)
1. a kind of method for diagnosing faults, which is characterized in that including:
With initial characteristics collection corresponding characteristic data set of the target device in normal operation and failure operation is obtained respectively, is obtained
Training data including normal characteristics data set and fault signature data set;Wherein, the initial characteristics collection includes various features;
Each characteristic and corresponding feature in the fault signature data set in the normal characteristics data set are calculated respectively
Corresponding KL distances, obtain KL distance sets between data;
Cross validation is carried out on the training data using support vector cassification, is verified result;
According to the size of KL distances in the verification result and the KL distance sets, from initial characteristics concentration determine with
The relevant feature of failure operation, obtains optimal characteristics collection;
When the diagnostic data for getting the target device, then determined from the diagnostic data and the optimal characteristics
Collect corresponding characteristic, corresponding fault diagnosis then is carried out to the target device using this feature data.
2. method for diagnosing faults according to claim 1, which is characterized in that described to calculate the normal characteristics data respectively
Concentrate between each characteristic and corresponding characteristic in the fault signature data set corresponding KL apart from the step of, bag
It includes:
The height corresponding to each characteristic in the normal characteristics data set and the fault signature data set is determined respectively
This distribution;
Each corresponding Gaussian Profile of characteristic and the fault signature data in the normal characteristics data set are calculated respectively
Concentrate the KL distances between the corresponding Gaussian Profile of corresponding characteristic.
3. method for diagnosing faults according to claim 2, which is characterized in that described to determine the normal characteristics data respectively
Collection and the fault signature data set in each characteristic corresponding to Gaussian Profile the step of, including:
Place is standardized to each characteristic in the normal characteristics data set and the fault signature data set respectively
Then reason determines the spy after each standardization in the normal characteristics data set and the fault signature data set respectively
Levy the Gaussian Profile corresponding to data.
4. method for diagnosing faults according to claim 1, which is characterized in that described to use support vector cassification described
The step of cross validation is carried out on training data, including:
Ten folding cross validations are carried out on the training data using support vector cassification.
5. method for diagnosing faults according to any one of claims 1 to 4, which is characterized in that described to be tied according to the verification
The size of KL distances in fruit and the KL distance sets is determined and the relevant spy of failure operation from initial characteristics concentration
The step of levying, obtaining optimal characteristics collection, including:
Screening meets the KL distances of preset condition from the KL distance sets, obtains target KL distances;
It is concentrated from the initial characteristics and screens the target KL apart from corresponding feature, obtain target signature;
According to the verification result from the target signature determine with the relevant feature of failure operation, obtain optimal characteristics collection.
6. method for diagnosing faults according to claim 5, which is characterized in that described screened from the KL distance sets meets
The KL of preset condition apart from the step of, including:
Descending sort, distance set after being sorted are carried out to the KL distance sets;
The KL distances for the default quantity for coming front are filtered out from distance set after the sequence.
7. method for diagnosing faults according to claim 5, which is characterized in that described screened from the KL distance sets meets
The KL of preset condition apart from the step of, including:
The KL distances that KL distances are more than predetermined threshold value are filtered out from the KL distance sets.
8. a kind of fault diagnosis system, which is characterized in that including:
Characteristic acquisition module, for obtain respectively target device in normal operation and failure operation with initial characteristics collection
Corresponding characteristic data set obtains the training data for including normal characteristics data set and fault signature data set;Wherein, it is described first
Beginning feature set includes various features;
KL distance calculation modules, for calculating each characteristic and the fault signature in the normal characteristics data set respectively
Corresponding KL distances between corresponding characteristic, obtain KL distance sets in data set;
Cross validation module for carrying out cross validation on the training data using support vector cassification, is verified
As a result;
Characteristic determination module, for the size according to KL distances in the verification result and the KL distance sets, at the beginning of described
Determined in beginning feature set with the relevant feature of failure operation, obtain optimal characteristics collection;
Fault diagnosis module, for working as the diagnostic data for getting the target device, then from the diagnostic data really
Characteristic corresponding with the optimal characteristics collection is made, then the target device is carried out using this feature data corresponding
Fault diagnosis.
9. a kind of failure diagnosis apparatus, which is characterized in that including processor and memory;Wherein, the processor performs preservation
Method for diagnosing faults as described in any one of claim 1 to 7 is realized during fault diagnostic program in the memory.
10. a kind of computer readable storage medium, which is characterized in that for storing fault diagnostic program;Wherein, the failure is examined
Disconnected program realizes method for diagnosing faults as described in any one of claim 1 to 7 when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711320019.4A CN108052974B (en) | 2017-12-12 | 2017-12-12 | Fault diagnosis method, system, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711320019.4A CN108052974B (en) | 2017-12-12 | 2017-12-12 | Fault diagnosis method, system, equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108052974A true CN108052974A (en) | 2018-05-18 |
CN108052974B CN108052974B (en) | 2022-05-17 |
Family
ID=62124331
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711320019.4A Active CN108052974B (en) | 2017-12-12 | 2017-12-12 | Fault diagnosis method, system, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108052974B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902739A (en) * | 2019-02-27 | 2019-06-18 | 厦门理工学院 | A kind of mechanical equipment fault degree recognition methods, terminal device and storage medium |
CN109976316A (en) * | 2019-04-25 | 2019-07-05 | 山东科技大学 | A kind of relevant Variable Selection of failure |
CN110212519A (en) * | 2019-05-23 | 2019-09-06 | 山西金晖隆开关有限公司 | Data processing method for low and medium voltage distribution network |
CN111078456A (en) * | 2019-12-26 | 2020-04-28 | 新奥数能科技有限公司 | Equipment fault diagnosis method and device, computer readable storage medium and electronic equipment |
CN111275135A (en) * | 2020-02-27 | 2020-06-12 | 苏州大学 | Fault diagnosis method, device, equipment and medium |
CN111707355A (en) * | 2020-06-19 | 2020-09-25 | 浙江讯飞智能科技有限公司 | Equipment operation state detection method, device, equipment and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080203601A1 (en) * | 2007-02-28 | 2008-08-28 | Airbus Espana, S.L. | Jig and method of manufacturing aircraft frames in a composite material |
CN105389326A (en) * | 2015-09-16 | 2016-03-09 | 中国科学院计算技术研究所 | Image annotation method based on weak matching probability canonical correlation model |
CN107067023A (en) * | 2017-01-10 | 2017-08-18 | 苏州大学 | A kind of method for diagnosing faults and device |
-
2017
- 2017-12-12 CN CN201711320019.4A patent/CN108052974B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080203601A1 (en) * | 2007-02-28 | 2008-08-28 | Airbus Espana, S.L. | Jig and method of manufacturing aircraft frames in a composite material |
CN105389326A (en) * | 2015-09-16 | 2016-03-09 | 中国科学院计算技术研究所 | Image annotation method based on weak matching probability canonical correlation model |
CN107067023A (en) * | 2017-01-10 | 2017-08-18 | 苏州大学 | A kind of method for diagnosing faults and device |
Non-Patent Citations (1)
Title |
---|
蔡晋辉等: "基于概率分布测度的高炉故障检测研究", 《仪器仪表学报》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109902739A (en) * | 2019-02-27 | 2019-06-18 | 厦门理工学院 | A kind of mechanical equipment fault degree recognition methods, terminal device and storage medium |
CN109976316A (en) * | 2019-04-25 | 2019-07-05 | 山东科技大学 | A kind of relevant Variable Selection of failure |
CN109976316B (en) * | 2019-04-25 | 2021-06-18 | 山东科技大学 | Fault-related variable selection method |
CN110212519A (en) * | 2019-05-23 | 2019-09-06 | 山西金晖隆开关有限公司 | Data processing method for low and medium voltage distribution network |
CN110212519B (en) * | 2019-05-23 | 2023-04-07 | 山西金晖隆开关股份有限公司 | Data processing method for medium and low voltage distribution network |
CN111078456A (en) * | 2019-12-26 | 2020-04-28 | 新奥数能科技有限公司 | Equipment fault diagnosis method and device, computer readable storage medium and electronic equipment |
CN111078456B (en) * | 2019-12-26 | 2023-05-16 | 新奥数能科技有限公司 | Device fault diagnosis method, device, computer readable storage medium and electronic device |
CN111275135A (en) * | 2020-02-27 | 2020-06-12 | 苏州大学 | Fault diagnosis method, device, equipment and medium |
CN111275135B (en) * | 2020-02-27 | 2023-09-12 | 苏州大学 | Fault diagnosis method, device, equipment and medium |
CN111707355A (en) * | 2020-06-19 | 2020-09-25 | 浙江讯飞智能科技有限公司 | Equipment operation state detection method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108052974B (en) | 2022-05-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108052974A (en) | A kind of method for diagnosing faults, system, equipment and storage medium | |
CN106021771A (en) | Method and device for diagnosing faults | |
CN107478455B (en) | Timing truncation reliability test method suitable for Weibull distribution type product | |
CN103698687B (en) | In integrated circuit hardware Trojan horse detection signal processing method and system | |
CN106201871A (en) | Based on the Software Defects Predict Methods that cost-sensitive is semi-supervised | |
CN110031226A (en) | A kind of diagnostic method and device of bearing fault | |
US20090196186A1 (en) | Root cause problem detection in network traffic information | |
CN106843941B (en) | Information processing method, device and computer equipment | |
CN109992916A (en) | A kind of engineering outfield aircraft engine failure prediction method based on flight data, terminal and readable storage medium storing program for executing | |
CN110399302A (en) | Risk Identification Method, device, electronic equipment and the medium of Software Testing Project | |
CN106203637A (en) | A kind of mobile system health evaluating method based on mahalanobis distance and health index | |
CN107133343A (en) | Big data abnormal state detection method and device based on time series approximate match | |
CN105955890B (en) | A kind of generation method and device of functional test case | |
CN106612216A (en) | Method and apparatus of detecting website access exception | |
CN107678908A (en) | Log recording method, device, computer equipment and storage medium | |
CN105701013A (en) | Software defect data feature selection method based on mutual information | |
CN104809104A (en) | Method and system for identifying micro-blog textual emotion | |
CN104715027B (en) | A kind of distributed data unusual fluctuation judges localization method and system | |
CN114111352B (en) | Carbide furnace vacuumizing system abnormity detection method and system | |
CN111752481A (en) | Memory monitoring and service life prediction method and system based on SPD | |
CN110301892A (en) | A kind of detection method and Related product based on hand vein recognition | |
CN108242411A (en) | The method and system of defect on management and control line | |
CN109409091B (en) | Method, device and equipment for detecting Web page and computer storage medium | |
CN116414717A (en) | Automatic testing method, device, equipment, medium and product based on flow playback | |
CN116136950A (en) | Chip verification method, device, system, electronic equipment and storage medium |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |