CN101477372A - Fault separation technique for chemical production abnormal sub-domain - Google Patents
Fault separation technique for chemical production abnormal sub-domain Download PDFInfo
- Publication number
- CN101477372A CN101477372A CNA2009100100136A CN200910010013A CN101477372A CN 101477372 A CN101477372 A CN 101477372A CN A2009100100136 A CNA2009100100136 A CN A2009100100136A CN 200910010013 A CN200910010013 A CN 200910010013A CN 101477372 A CN101477372 A CN 101477372A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- domain
- subdomain
- abnormal sub
- 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
Landscapes
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention relates to a technical method for separating abnormal subdomain fault in chemical production, which is a novel fault separating method and is realized by using statistical analysis theory. The method comprises the following steps: extracting a principal element from data information acquired by a process; classifying data into information related with the principal element and information in weak relation with the principal element according to the correlation of a variable and the principal element; establishing a statistical process control limit for diagnosis of the fault respectively according to the two classes of data; determining a domain of normal data as a normal subdomain and a domain of abnormal data as an abnormal subdomain; and iterating the processes in the abnormal subdomain until the fault is converged to be separated. The method classifies the process data according to the correlation of the process data and a principal element array, can reflect characteristics and presentation of the fault information better, and improve relative accuracy and reliability of fault separation by statistical analysis. The method has the advantages of advanced technique, theoretical principle existence, and strong actual application and operability.
Description
Technical field
The present invention relates to the supervision and the fault diagnosis of chemical production technical process, the spot sampling data that particularly relate to a kind of control system based on process are carried out the fault detect and the isolation technics method of process.
Background technology
It is the key of fault diagnostic techniques that fault is separated, after detecting the system failure, how to handle and make system recovery arrive normal duty, and be the key factor of fault diagnosis success or not.Separate usually and can point out the position, size and the variation tendency that break down based on the fault of model to the basic reason that is out of order.And can only provide scope of failure, possible variable or the relevant equipment of variable therewith based on the fault isolation technics of statistical study, and can not determine clear and definite failure cause.This brings many problems for fault diagnosis, can not isolate fault after detecting fault, and detection will lose meaning.Therefore, in recent years many achievements in research have been appearred in the fault separation based on statistical method, but real solution awaits further to study based on the problem that statistical model carries out the fault separation.Carry out technology that fault separates at the abnormal sub-domain of this proposition and will have certain theory and using value.
Summary of the invention
The object of the present invention is to provide a kind of supervision and fault diagnosis technology method of chemical production technical process, it is a kind of new fault separating method, improvement, the abnormal sub-domain of using pivot analysis technology, information correlativity analysis, statistics control limit separate procedure fault, be to based on the lifting of statistical study fault diagnosis technology with replenish, to process failure diagnosis, use and have important and practical meanings.
The objective of the invention is to be achieved through the following technical solutions:
The technical method that the chemical production abnormal sub-domain fault is separated, its abnormal sub-domain fault isolation technics require to carry out according to pivot analysis that fault diagnosis realizes; Establish the threshold value of related coefficient, distinguish the related coefficient of each variable of technology and pivot battle array simultaneously, carry out classification of Data with this; Require the pivot related data to comprise more data information, and the new statistics control limit in definite this territory is distinguished normal subdomain and abnormal sub-domain.
The technical method that described chemical production abnormal sub-domain fault is separated, it carries out classification of Data according to correlation of data, and important information and failure message are included in the data battle array relevant with pivot more.
The technical method that described chemical production abnormal sub-domain fault is separated, it adopts normal subdomain to carry out separating of fault with abnormal sub-domain.
Advantage of the present invention and effect are:
1. the present invention classifies by the correlativity of itself and pivot battle array to process data, can reflect the feature and the presentation of failure message better.
2. the present invention sets up new subdomain control limit to carry out fault detect more targeted, and carries out the division of failure domain on the basis of statistical study, has improved statistical study and has carried out relative precision and the reliability that fault is separated.
3. the technology of the present invention advanced person has theoretical basis, practical application and workable.
Embodiment
The present invention at first carries out information analysis with all process variable, can make the correlativity between more abundant information, information stronger, the sign that goes out for same fault detect is more obvious, also makes particularly relevant with the pivot failure message of relevant information more concentrated simultaneously.Therefore the correlation analysis that carries out process variable and pivot can be classified variable, and pivot correlated variables (principal relatedvariable-PRV) is followed successively by PRV
0, PRV
1..., PRV
n
Carry out sorted variable information and comprise the information field of fault for all, information is not only relevant but also overlapping between them, therefore the entire variable information field must be decomposed.The present invention is divided into abnormal sub-domain with it, and other are normal subdomain, and fault can be separated within the specific limits.
The foundation of abnormal sub-domain is to realize on the basis of pivot analysis, by the mapping of hiding to process variable, can divide pivot subspace (Principal Component sub-Space-PCS) and residual error subspace (Residual sub-Space-RS), the pivot subspace comprises the main information of process, and the full detail that the residual error subspace comprises process and main information is poor.Correlationship according to procedural information and pivot subspace, can procedural information be projected to two spaces by correlativity, the one, the variable subspace relevant with pivot is PRV (Principal Related Variable), the 2nd, and with variable subspace OV (Other Variable) relevant a little less than the pivot.Decompose by can in the subspace, carry out pivot to the division of the variable space, and set up the statistics control limit δ of pivot subspace and residual error subspace
(α)The detection of fault is carried out in the projection of SPE among PRV and OV statistics controlling index in pivot subspace and residual error subspace, testing result makes a part of procedural information be in normal processes subdomain NOR, and a part is in improper process subdomain ANSR.
Process is constantly carried out loop iteration, can obtain continuous convergent abnormal sub-domain ANSR
1, ANSR
2..., ANSR
nNormal subdomain OV with continuous expansion
1, OV
2..., OV
nThis iteration result is mapped to failure message in the abnormal sub-domain, realizes the separation of fault.
As following table:
The hypothetical process variable matrix is X=[x
1, x
2..., x
m], then the pivot of X is decomposed:
T is that pivot gets sub matrix.If m variable arranged, n sampled value, m variable can be decomposed into two parts according to the correlativity with T.
ρ (x
i, T) be the related coefficient of variable and pivot battle array.If after threshold value was determined, it was PRV that the raw data variable can be decomposed into the variable subspace relevant with pivot, other and the weak relevant variable subspace OV of pivot.Respectively PRV and OV are projected to the residual error space and constitute two new statistical indicator SPEPVR and SPEOV.
Based on new statistical indicator SPEPVR and SPEOV, the PCA of application enhancements can carry out fault diagnosis.Two new statistical indicators obtain in PVR and OV two subspaces respectively, if statistical indicator is at the statistics territory of SPEPVR and SPEOV δ
(α)In, think that then the variable subspace that PRV or OV are in is normal subdomain (NOR); Otherwise be abnormal sub-domain (ANSR).
Abnormal sub-domain is ANSR, and PRV among the ANSR and OV can carry out fault detect in ANSR separately.
With
It is the control limit of PRV and OV.If the projection of SPE difference then has two kinds of situations to produce in model separately among PRV and the OV.The one, the SPE among the PRV is in the control limit, and the SPE among the OV shows that the fault variable is in OV outside the control limit; Another kind be SPE among the PRV outside control limit, and the SPE among the OV shows that the fault variable is in PRV in limit.
Be respectively ANSR if establish the normal subdomain and the abnormal sub-domain of process variable projection first
0And OV
0And the secondary subdomain of projection then is ANSR successively
1And OV
1, ANSR
2And OV
2..., ANSR
nAnd OV
n
Claims (3)
1. the technical method that separates of chemical production abnormal sub-domain fault is characterized in that abnormal sub-domain fault isolation technics requires to carry out according to pivot analysis that fault diagnosis realizes; Establish the threshold value of related coefficient, distinguish the related coefficient of each variable of technology and pivot battle array simultaneously, carry out classification of Data with this; Require the pivot related data to comprise more data information, and the new statistics control limit in definite this territory is distinguished normal subdomain and abnormal sub-domain.
2. the technical method that chemical production abnormal sub-domain fault according to claim 1 is separated is characterized in that carrying out classification of Data according to correlation of data, and important information and failure message are included in the data battle array relevant with pivot more.
3. the technical method that chemical production abnormal sub-domain fault according to claim 1 is separated is characterized in that adopting normal subdomain to carry out separating of fault with abnormal sub-domain.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2009100100136A CN101477372A (en) | 2009-01-07 | 2009-01-07 | Fault separation technique for chemical production abnormal sub-domain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2009100100136A CN101477372A (en) | 2009-01-07 | 2009-01-07 | Fault separation technique for chemical production abnormal sub-domain |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101477372A true CN101477372A (en) | 2009-07-08 |
Family
ID=40838099
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNA2009100100136A Pending CN101477372A (en) | 2009-01-07 | 2009-01-07 | Fault separation technique for chemical production abnormal sub-domain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101477372A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103616889A (en) * | 2013-11-29 | 2014-03-05 | 渤海大学 | Chemical process fault classifying method of reconstitution sample center |
CN104699077A (en) * | 2015-02-12 | 2015-06-10 | 浙江大学 | Nested iterative fisher discriminant analysis-based fault diagnosis isolation method |
CN109240276A (en) * | 2018-11-09 | 2019-01-18 | 江南大学 | Muti-piece PCA fault monitoring method based on Fault-Sensitive Principal variables selection |
-
2009
- 2009-01-07 CN CNA2009100100136A patent/CN101477372A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103616889A (en) * | 2013-11-29 | 2014-03-05 | 渤海大学 | Chemical process fault classifying method of reconstitution sample center |
CN103616889B (en) * | 2013-11-29 | 2015-12-09 | 渤海大学 | A kind of chemical process Fault Classification of reconstructed sample center |
CN104699077A (en) * | 2015-02-12 | 2015-06-10 | 浙江大学 | Nested iterative fisher discriminant analysis-based fault diagnosis isolation method |
CN109240276A (en) * | 2018-11-09 | 2019-01-18 | 江南大学 | Muti-piece PCA fault monitoring method based on Fault-Sensitive Principal variables selection |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108986869B (en) | Disk fault detection method using multi-model prediction | |
Li et al. | Battery fault diagnosis for electric vehicles based on voltage abnormality by combining the long short-term memory neural network and the equivalent circuit model | |
CN105629958B (en) | A kind of batch process method for diagnosing faults based on sub-period MPCA SVM | |
CN103245861B (en) | A kind of transformer fault diagnosis method based on Bayesian network | |
CN101726688A (en) | Method for diagnosing multi-data-source information fusion-based power system fault | |
CN101614786B (en) | Online intelligent fault diagnosis method of power electronic circuit based on FRFT and IFSVC | |
CN103198175B (en) | Based on the Diagnosis Method of Transformer Faults of fuzzy clustering | |
CN102818948B (en) | Synthetic diagnosing method based on fuzzy fault diagnosis and relevancy model diagnosis | |
CN107368854A (en) | A kind of circuit breaker failure diagnostic method based on improvement evidence theory | |
CN109660206B (en) | Wasserstein GAN-based photovoltaic array fault diagnosis method | |
CN112611936A (en) | Distribution network transformer fault dynamic detection and classification system based on edge calculation | |
CN105974265A (en) | SVM (support vector machine) classification technology-based power grid fault cause diagnosis method | |
CN102944416A (en) | Multi-sensor signal fusion technology-based fault diagnosis method for wind turbine blades | |
CN104267668A (en) | Bayes-method-based spaceflight valve part fault diagnosis method in machining process | |
CN103323228A (en) | Mining drill fault intelligent identification method | |
CN112327189B (en) | Comprehensive judging method for health state of energy storage battery based on KNN algorithm | |
CN103839080A (en) | Video streaming anomalous event detecting method based on measure query entropy | |
Gu et al. | An improved sensor fault diagnosis scheme based on TA-LSSVM and ECOC-SVM | |
CN104880217A (en) | Fault sensor information reconstruction method based on measured value association degree | |
CN101477372A (en) | Fault separation technique for chemical production abnormal sub-domain | |
CN110596490A (en) | Intelligent detection method for railway turnout fault | |
CN105334823A (en) | Supervision-based industrial process fault detection method of linear dynamic system model | |
CN106959397A (en) | A kind of method for designing of small fault diagnostic system for high ferro inverter | |
CN116466241B (en) | Thermal runaway positioning method for single battery | |
CN104102726A (en) | Modified K-means clustering algorithm based on hierarchical clustering |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090708 |