CN105955219B - Distributed dynamic procedure failure testing method based on mutual information - Google Patents
Distributed dynamic procedure failure testing method based on mutual information Download PDFInfo
- Publication number
- CN105955219B CN105955219B CN201610389000.4A CN201610389000A CN105955219B CN 105955219 B CN105955219 B CN 105955219B CN 201610389000 A CN201610389000 A CN 201610389000A CN 105955219 B CN105955219 B CN 105955219B
- Authority
- CN
- China
- Prior art keywords
- data
- matrix
- variable
- probability
- value
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 80
- 238000012360 testing method Methods 0.000 title claims abstract description 9
- 230000008569 process Effects 0.000 claims abstract description 44
- 238000005259 measurement Methods 0.000 claims abstract description 15
- 238000012544 monitoring process Methods 0.000 claims abstract description 14
- 239000011159 matrix material Substances 0.000 claims description 43
- 238000012549 training Methods 0.000 claims description 12
- 238000004519 manufacturing process Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 230000003416 augmentation Effects 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- 102000002274 Matrix Metalloproteinases Human genes 0.000 claims description 3
- 108010000684 Matrix Metalloproteinases Proteins 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 17
- 238000005070 sampling Methods 0.000 abstract description 10
- 238000007796 conventional method Methods 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 abstract description 2
- 239000000463 material Substances 0.000 description 7
- 239000000498 cooling water Substances 0.000 description 4
- 239000004576 sand Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000007791 liquid phase Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 238000001311 chemical methods and process Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011112 process operation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Complex Calculations (AREA)
- General Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
Abstract
The present invention relates to a kind of distributed dynamic procedure failure testing method based on mutual information, this method are that each measurand of process introduces delay measurements first；Secondly, the correlation metric defined by mutual information distinguishes the autocorrelation and crossing dependency that can be embodied in different sampling instants for each measurand of process；Then, corresponding pivot analysis Fault Model is established respectively to the data set sub-block corresponding to each variable；Finally, when implementing on-line monitoring, the result of different faults detection model is fused into a probabilistic type monitoring index using Bayesian inference, to facilitate final failure decision.Compared with the conventional method, the invention has fully considered the autocorrelation and crossing dependency between the different measurands being embodied in different sampling instants, avoid and lose the useful information that may hide of process data Complex Dynamic, keep corresponding failure detection result relatively reliable with it is accurate.
Description
Technical field
The present invention relates to industrial process fault detection methods, more particularly, to a kind of distributed dynamic mistake based on mutual information
Journey fault detection method.
Background technology
In recent years, the fault detection method of data-driven has been used as because of its simple general-purpose and has ensured production safety and ensure
The important technical of product quality stability and obtained the attention of researchers.For with pivot analysis (PCA) be representative
The research of multivariate statistical process monitoring method received the extensive concern of industrial quarters and academia, basic thought is all
It is that the potential information that can reflect process operation situation is excavated from the data that industrial process acquires.Such methods are avoided that foundation
Accurate process mechanism model, thus it is well suited for the modern large complicated chemical industry process of monitoring.
Generally, the data of production process acquisition make process data inevitably there is dynamic because the sampling interval is short
Property (or autocorrelation).In the conventional method, dynamic PCA (DPCA) methods and variable (DLV) method of dynamically hiding are two kinds common
Solution dynamic process fault detection problem technological means.In view of the extensive property and complexity of modern industry process, mistake
Dynamic of the number of passes between is more complicated, and different measurands can have different autocorrelations, and the mutual shadow between variable
Ringing (i.e. crossing dependency) can also be embodied in different sampling instants.However, method traditional DPCA and DLV etc. is generally all false
If the consistency of autocorrelation and crossing dependency on the sampling time between measurand, the complicated dynamic of process data is had ignored
The useful information that characteristic may be hidden.If during establishing Fault Model, can be embodied from different sampling instants
Go out the autocorrelation and crossing dependency between different measurands, it will obtain more accurate failure detection result, significantly
Promote the reliability of corresponding failure detection model.
Invention content
The purpose of the present invention is to provide a kind of the distributed dynamic procedure failure testing method based on mutual information, the invention
It has fully considered the autocorrelation and crossing dependency between the different measurands being embodied in different sampling instants, has avoided and lose
Lose the useful information that may hide of process data Complex Dynamic, keep corresponding failure detection result relatively reliable with it is accurate.
Technical solution is used by the present invention solves above-mentioned technical problem：A kind of distributed dynamic mistake based on mutual information
Journey fault detection method, includes the following steps：
(1) sampled data under production process normal operating condition is collected, the training dataset X=[x of modeling are formed1,
x2..., xn]T.Wherein, X ∈ Rn×m, n is number of training, and m is process measurement variable number, and R is set of real numbers, Rn×mIndicate n × m
The real number matrix of dimension, upper label T representing matrix transposition.
The measured value that its preceding l moment is introduced for each measurand in data matrix X constitutes augmentation type number as follows
According to matrix Xa∈R(n-l)×(l+1)m：
Then, to matrix XaIt is standardized, it is 0 to obtain mean value, the new data matrix that standard deviation is 1
(3) it is directed to ith measurement variable xi, calculate its withIn each variable xjBetween association relationship Cij=I (xi, xj),
Wherein, i=1,2 ..., m and j=1,2 ..., m (l+1) are respectively each measurand of process and matrixIn each variable subscript
Number, I (xi, xj) indicate to calculate variable xiWith xjBetween mutual information, specific calculation is as follows：
Wherein, marginal probability p (xj) and p (xj) and joint probability p (xj, xj) be by Density Estimator method come really
Make corresponding probability value.
(4) it is directed to process ith measurement variable xi, vectorial C that obtained association relationship is formedI, j∈R1×m(l+1)Into
After the arrangement of row descending, variable and record variable label before selecting corresponding to k maximum value.
(5) according to the variable label of record from data matrixThe middle corresponding row of selection form training dataset sub-block
(6) to data set blockCorresponding Fault Model is established using PCA methods, records corresponding model parameter
ΘiWith spare.
(7) (3)~(6) are repeated the above steps until having its corresponding PCA failures inspection for all m process variables
Survey model.
(8) new process sampled data x ∈ R are collectedm×1, the survey at l moment before first being introduced to wherein each measurand
Magnitude obtains new data xa∈Rm(l+1)×1, after to xaIt is standardized to obtain
(9) using the variable label of record, by data vectorResolve into m different vectorial sub-blocks
(10) i-th of PCA Fault Model parametric configuration is utilized to correspond to vectorial sub-block'sAnd QiStatistic, i.e.,：
Wherein, | | | | it indicates to calculate the 2- norms of vector, and repeats this step until obtaining m group statistics.
(11) by Bayesian inference, m T will be obtained2(or Q) counts magnitude and carries out probability fusion, obtains a unification
Probabilistic type monitoring index(or BIQ), and make about the whether normal decision of current data sample.
Further, the step (6) is specially：It is sub-block using PCA methodsFault Model is established, and is recorded
Model parameterWherein, d is the pivot number retained, Pi∈Rk×dFor the projection of pca model
Transformation matrix, Λi∈Rd×dBe one by d eigenvalue cluster at diagonal matrix, the element on diagonal line is that α is counting statistics
Amount control limitWith QI, limWhen used confidence level.Specific implementation is shown in steps are as follows：
1. calculatingCovariance matrixWherein S ∈ Rk×k。
2. the preceding d maximum eigenvalue λ of setup parameter d, solution matrix S1＞ λ2＞ ... ＞ λdCorresponding feature vector
p1, p2..., pd.So, diagonal matrix Λi=diag { λ1, λ2..., λd, projective transformation matrix Pi=[p1, p2..., pd]。
3. setting confidence alpha, control limitRespectively：
Wherein, FD, n-l-d, αThe F that expression confidence level is α, degree of freedom is respectively d and n-l-d is distributed corresponding value,Table
Show that degree of freedom is h, confidence level is that α is value corresponding to chi square distribution, M and V are respectively estimation mean value and the estimation side of Q statistical magnitude
Difference.
Further, the step (11) is specially：First, using Bayesian inference by m T2(or Q) count magnitude into
Row probability fusion obtains probabilistic type monitoring indexWith BIQ.Then, it will be calculatedWith BIQThe concrete numerical value of index
Limit 1- α are controlled with probability to be compared.If any one index value is more than 1- α, decision new dataFor fault sample；
Conversely, the dataFor normal sample, and then fault detect is continued to next new obtained data that sample.Bayes
The specific implementation of reasoning is shown in steps are as follows：
(A) to m T2Statistic is merged：
1. calculating new data according to the following formulaBelong to the probability of failure：
Wherein, probabilityCalculation it is as follows：
Wherein, N and F indicates normal and fault condition, prior probability respectivelyWithValue α and 1- α respectively,
Conditional probabilityWithCalculation it is as follows：
2. final probabilistic type index is calculated by following formula
(B) m Q statistical magnitude is merged
It is directed to Q statistical magnitude, calculates new data firstThen probability, the conditional probability etc. for belonging to failure obtain final
Probabilistic type index BIQ, T is merged with above-mentioned2The mode of statistic is identical.
Compared with the conventional method, advantages of the present invention and effect are as follows：
1. the present invention chooses measured value associated therewith for each measurand of process in different sampling instants,
Having fully taken into account the complex characteristics between modern industry process data, i.e., different measurands can have different autocorrelations,
And the crossing dependency between variable can be also embodied in different sampling instants.The distributed PCA events established on this basis
Barrier detection model, which can be effectively prevented from, loses the useful information that process data Complex Dynamic may be hidden, and greatly improves
The reliability of corresponding failure detection model.
2. autocorrelation and intersection of the present invention using the mutual information between data come gauge variable in different sampling instants
Correlation, under the premise of needing not rely on any process priori, realize to the dynamic characteristic of process data complexity into
Description is gone.Meanwhile the present invention has also played the advantage of distributed modeling method, by establishing multiple PCA Fault Models
And reduce the difficulty and complexity to process data analysis.Therefore, compared to traditional method, the present invention can effectively improve
Fault detect effect.
Description of the drawings
Fig. 1 is that the present invention is based on the deblocking method schematic diagrams of mutual information.
Fig. 2 is the method for the present invention, DPCA, DLV method to the failure detection result of 16 floor data of TE procedure faults.
Specific implementation mode
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
As shown in Figure 1, the invention discloses a kind of distributed dynamic procedure failure testing method based on mutual information, the party
Method is directed to the fault detection problem of modern industry process, and production process normal operating condition is collected first with data collecting system
Under data set.Secondly, the delay measurements for its preceding l moment being introduced for each measurand of process form augmentation type square
Battle array.Then, matrix sub block corresponding with each measurand is chosen using mutual information, and establishes PCA Fault Models.Most
Afterwards, new sampled data is monitored on-line, that is, builtWith BIQMonitoring index, and just whether the currently monitored data of decision
Often.
The specific implementation step of the present invention is as follows：
Step 1：The sampled data under production process normal operating condition is collected, the training dataset X=of modeling is formed
[x1, x2..., xn]T.Wherein, X ∈ Rn×m, n is number of training, and m is process measurement variable number, and R is set of real numbers, Rm×nIndicate m
The real number matrix of × n dimensions, upper label T representing matrix transposition.
Step 2：The measured value that its preceding l moment is introduced for each measurand in data matrix X constitutes as follows increase
Wide type data matrix Xa∈R(n-l)×(l+1)m：
Then, to matrix XaIt is standardized, it is 0 to obtain mean value, the new data matrix that standard deviation is 1
Step 3：For ith measurement variable xi, calculate its withIn each variable xjBetween association relationship Cij=I (xi,
xj).Wherein, i=1,2 ..., m and j=1,2 ..., m (l+1) are respectively each measurand of process and matrixIn each variable
Upper label, I (xi, xj) it is to calculate variable xiWith xjBetween mutual information, i.e.,：
Wherein, marginal probability p (xj) and p (xj) and joint probability p (xj, xj) be by Density Estimator method come really
Make corresponding probability value.
Step 4：For process ith measurement variable xi, vectorial C that obtained association relationship is formedI, j∈R1×m(l+1)
After carrying out descending arrangement, variable and record variable label before selecting corresponding to k maximum value.
Step 5：According to the variable label of record from data matrixThe middle corresponding row of selection form training data collected works
Block
Step 6：To data set blockCorresponding Fault Model is established using PCA methods, records corresponding model
Parameter ΘiWith spare.
It is sub-block using PCA methodsEstablish Fault Model, and record cast parameterWherein, d is the pivot number retained, Pi∈Rk×dFor the projective transformation square of pca model
Battle array, Λi∈Rd×dBe one by d eigenvalue cluster at diagonal matrix, the element on diagonal line is that α is Counting statistics amount control
LimitWith QI, limWhen used confidence level.Specific implementation is shown in steps are as follows：
1. calculatingCovariance matrixWherein S ∈ Rk×k。
2. the preceding d maximum eigenvalue λ of setup parameter d, solution matrix S1＞ λ2＞ ... ＞ λdCorresponding feature vector
p1, p2..., pd.So, diagonal matrix Λi=diag (λ1, λ2..., λd, projective transformation matrix Pi=[p1, p2..., pd]。
3. setting confidence alpha, control limitWith QI, limRespectively：
Wherein, FD, n-l-d, αThe F that expression confidence level is α, degree of freedom is respectively d and n-l-d is distributed corresponding value,Table
Show that degree of freedom is h, confidence level is that α is value corresponding to chi square distribution, M and V are respectively estimation mean value and the estimation side of Q statistical magnitude
Difference.
Step 7：(3)~(6) repeat the above steps until having its corresponding PCA event for all m process variables
Hinder detection model.
Step 8：Collect new process sampled data x ∈ Rm×1, the l moment before first being introduced to wherein each measurand
Measured value, obtain new data xa∈Rm(l+1)×1, after to xaIt is standardized to obtain
Step 9：Using the variable label of record, by data vectorResolve into m different vectorial sub-blocks
Step 10：Correspond to vectorial sub-block using i-th of PCA Fault Model parametric configuration'sAnd QiStatistic,
And this step is repeated until obtaining m group statistics.ConstructionAnd QiThe concrete mode of statistic is as follows：
Step 11：By Bayesian inference, m T will be obtained2(or Q) counts magnitude and carries out probability fusion, obtains a system
One probabilistic type monitoring index(or BIQ), and make about the whether normal decision of current data sample.
First, using Bayesian inference by m T2(or Q) statistics magnitude carries out probability fusion and obtains probabilistic type monitoring indexWith BIQ.Then, it will be calculatedWith BIQThe concrete numerical value of index controls limit 1- α with probability and is compared.If appointing
What index value is more than 1- α, then decision new dataFor fault sample；Conversely, the dataFor normal sample, in turn
Fault detect is continued to next new obtained data that sample.The specific implementation of Bayesian inference is shown in steps are as follows：
(A) to m T2Statistic is merged
1. calculating new data according to the following formulaBelong to the probability of failure：
Wherein, probabilityCalculation it is as follows：
Wherein, N and F indicates normal and fault condition, prior probability respectivelyWithValue α and 1- α respectively,
Conditional probabilityWithCalculation it is as follows：
2. final probabilistic type index is calculated by following formula
(B) m Q statistical magnitude is merged
It is directed to Q statistical magnitude, calculates new data firstThen probability, the conditional probability etc. for belonging to failure obtain final
Probabilistic type index BIQ, T is merged with above-mentioned2The mode of statistic is identical.
With reference to the example of a specific industrial process come illustrate the present invention relative to existing method superiority with
Reliability.The process data comes from the experiment of the U.S. Tennessee-Yi Siman (TE) chemical process, and prototype is Yi Siman Chemical Manufactures
One actual process flow in workshop.Currently, complexity of the TE processes because of its flow, has been used as a standard test platform wide
It is general to be studied for fault detect.Entire TE processes include that 22 measurands, 12 performance variables and 19 composition measurements become
Amount.The data acquired are divided into 22 groups, including the data set and 21 groups of fault datas under 1 group of nominal situation.And at these
In fault data, 16 are known fault types, such as the changing of cooling water inlet temperature or feed constituents, valve viscous, anti-
Dynamics drift etc. is answered, also 5 fault types are unknown.In order to be monitored to the process, as shown in Table 1 33 are chosen
Next a process variable is explained in detail specific implementation step of the present invention in conjunction with the TE processes.
1. acquiring the process data under nominal situation, while fault data different in 21 is acquired, and chooses 960 normally
Data form matrix X ∈ R960×33。
2. being directed to training data X, distributed PCA Fault Models are established.
(1) it is that each measurand introduces the measured value composition augmentation type data at its preceding l=2 moment in data matrix X
Matrix Xa∈R958×99, and it is standardized to obtain mean value to be 0, the new data matrix that standard deviation is 1
(2) it is directed to ith measurement variable xi, calculate its withIn each variable xjBetween association relationship Cij=I (xi, xj),
Wherein, upper label i=1,2 ..., 33, upper label j=1,2 ..., 99.
(3) the vectorial C that obtained association relationship is formedI, j∈R1×99After carrying out descending arrangement, k maximum before selection
The corresponding variable of value and record variable label.
(4) k=-10 is set, according to the variable label of record from data matrixThe middle corresponding row composition training of selection
Data set sub-block
(5) utilize PCA methods to training dataset sub-blockCorresponding Fault Model is established, corresponding model is recorded
ParameterWith spare
(6) multiple above-mentioned steps 2~5 for all 33 process variables until have its corresponding PCA fault detects mould
Type.
3. obtaining new sampled data, and the measurement for introducing the preceding l=2 moment to it is worth to new data vector xa∈R99 ×1, then it is standardized, is finally calculatedWith BIQProbabilistic type monitoring index.
(1) using the variable label of record, by data vectorResolve into m different vectorial sub-blocks
(2) i-th of PCA Fault Model parametric configuration is utilized to correspond to vectorial sub-block'sAnd QiStatistic is laid equal stress on
This multiple step is until obtain 33 groups of statistics.
(3) by Bayesian inference, 33 T will be obtained2(or Q) counts magnitude and carries out probability fusion, obtains a unification
Probabilistic type monitoring index(or BIQ)。
Table 1：TE process monitoring variables.
| Serial number | Variable description | Serial number | Variable description | Serial number | Variable description |
| 1 | Material A flow | 12 | Separator liquid level | 23 | D material inlet valves position |
| 2 | Material D flows | 13 | Separator pressure | 24 | E material inlet valves position |
| 3 | Material E flows | 14 | Separator bottom of tower flow | 25 | A material inlet valves position |
| 4 | Combined feed flow | 15 | Stripper grade | 26 | A and C material inlet valves position |
| 5 | Circular flow | 16 | Pressure of stripping tower | 27 | Compressor cycle valve location |
| 6 | Reactor feed | 17 | Stripper bottom rate | 28 | Empty valve location |
| 7 | Reactor pressure | 18 | Stripper temperature | 29 | Separator liquid phase valve location |
| 8 | Reactor grade | 19 | Stripper upper steam | 30 | Stripper liquid phase valve location |
| 9 | Temperature of reactor | 20 | Compressor horsepower | 31 | Stripper steam valve position |
| 10 | Rate of evacuation | 21 | Reactor cooling water outlet temperature | 32 | Reactor condensate flow |
| 11 | Separator temperature | 22 | Separator cooling water outlet temperature | 33 | Condenser cooling water flow |
4. online fault detect
According to what is be currently calculatedAnd BIQThe occurrence of index is compared with α=0.01 control limit 1-, is judged
Whether current data is normal.The failure detection result such as Fig. 2 institutes of the method for the present invention, DPCA and DLV to TE procedure faults 16
Show.It can be seen that the method for the present invention is obviously superior to DPCA and DLV methods to the detection result of the failure.
Above-described embodiment is only used for explaining the present invention, rather than limits the invention, in the spirit and power of the present invention
In the protection domain that profit requires, to any modifications and changes that the present invention makes, both fall in protection scope of the present invention.
Claims (3)
1. a kind of distributed dynamic procedure failure testing method based on mutual information, which is characterized in that include the following steps：
(1) sampled data under production process normal operating condition is collected, the training dataset X=[x of modeling are formed1,
x2..., xn]T, wherein X ∈ Rn×m, n is number of training, and m is process measurement variable number, and R is set of real numbers, Rn×mIndicate n × m
The real number matrix of dimension, upper label T representing matrix transposition；
(2) it is that each measurand introduces the measured value composition augmentation type data as follows at its preceding l moment in data matrix X
Matrix Xa∈R(n-l)×(l+1)m：
Then, to matrix XaIt is standardized, it is 0 to obtain mean value, the new data matrix that standard deviation is 1
(3) it is directed to ith measurement variable xi, calculate its withIn each variable xjBetween association relationship Cij=I (xi, xj),
In, i=1,2 ..., m and j=1,2 ..., m (l+1) are respectively each measurand of process and matrixIn each variable subscript
Number, I (xi, xj) indicate to calculate variable xiWith xjBetween mutual information, specific calculation is as follows：
Wherein, marginal probability p (xj) and p (xj) and joint probability p (xj, xj) it is to be determined by Density Estimator method
Corresponding probability value；
(4) it is directed to process ith measurement variable xi, vectorial C that obtained association relationship is formedI, j∈R1×m(l+1)It is dropped
After sequence arrangement, the variable before selecting corresponding to k maximum value, and record variable label；
(5) according to the variable label of record from data matrixThe middle corresponding row of selection form training dataset sub-block
(6) to data set blockCorresponding Fault Model is established using PCA methods, records corresponding model parameter ΘiWith
It is spare；
(7) (3)~(6) are repeated the above steps until having its corresponding PCA fault detects mould for all m process variables
Type；
(8) new process sampled data x ∈ R are collectedm×1, the measurement at l moment before first being introduced to wherein each measurand
Value, obtains new data xa∈Rm(l+1)×1, after to xaSame standard is carried out to handle to obtain
(9) using the variable label of record, by data vectorResolve into m different vectorial sub-blocks
(10) i-th of PCA Fault Model parametric configuration is utilized to correspond to vectorial sub-blockTi 2And QiStatistic, i.e.,：
Wherein, | | | | it indicates to calculate the length of vector, and repeats this step until obtaining m group statistics；
(11) by Bayesian inference, the m T that will be obtained2Statistic and m Q statistical magnitude carry out probability fusion respectively, to deserved
To probabilistic type monitoring indexWith BIQ, and make about the whether normal decision of current data sample.
2. the distributed dynamic procedure failure testing method according to claim 1 based on mutual information, which is characterized in that institute
Stating step (6) is specially：It is sub-block using PCA methodsEstablish Fault Model, and record cast parameterWherein, d is the pivot number retained, Pi∈Rk×dFor the projective transformation square of pca model
Battle array, Λi∈Rd×dBe one by d eigenvalue cluster at diagonal matrix, α be Counting statistics amount control limitWith QI, limWhen institute
The confidence level of use, specific implementation is shown in steps are as follows：
1. calculatingCovariance matrixWherein S ∈ Rk×k；
2. the preceding d maximum eigenvalue λ of setup parameter d, solution matrix S1＞ λ2＞ ... ＞ λdCorresponding feature vector p1,
p2..., pd, then, diagonal matrix Λi=diag { λ1, λ2..., λd, projective transformation matrix Pi=[p1, p2..., pd]；
3. setting confidence alpha, control limitWith QI, limRespectively：
Wherein, FD, n-l-d, αThe F that expression confidence level is α, degree of freedom is respectively d and n-l-d is distributed corresponding value,It indicates certainly
Be α it is value corresponding to chi square distribution by spending for h, confidence level, M and V are respectively the estimation mean value and estimate variance of Q statistical magnitude.
3. the distributed dynamic procedure failure testing method according to claim 1 based on mutual information, which is characterized in that institute
Stating step (11) is specially：First, using Bayesian inference by m T2Statistic and m Q statistical magnitude carry out probability fusion respectively
Correspondence obtains probabilistic type monitoring indexWith BIQ；Then, it will be calculatedWith BIQThe concrete numerical value and probability of index
Control limit 1- α are compared, if any one index value is more than 1- α, decision new dataFor fault sample；Conversely, should
DataFor normal sample, and then fault detect is continued to next new obtained data that sample；The tool of Bayesian inference
Shown in body realizes that steps are as follows：
(A) to m T2Statistic is merged：
1. calculating new data according to the following formulaBelong to the probability of failure：
Wherein, probabilityCalculation it is as follows：
Wherein, N and F indicates normal and fault condition, prior probability respectivelyWithValue α and 1- α respectively, condition
ProbabilityWithCalculation it is as follows：
2. final probabilistic type index is calculated by following formula
(B) m Q statistical magnitude is merged
Be directed to Q statistical magnitude, according to step (A) in merge T2The same way of statistic is calculated final probabilistic type and refers to
Mark BIQ。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610389000.4A CN105955219B (en) | 2016-05-30 | 2016-05-30 | Distributed dynamic procedure failure testing method based on mutual information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610389000.4A CN105955219B (en) | 2016-05-30 | 2016-05-30 | Distributed dynamic procedure failure testing method based on mutual information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105955219A CN105955219A (en) | 2016-09-21 |
| CN105955219B true CN105955219B (en) | 2018-08-17 |
Family
ID=56907626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610389000.4A Expired - Fee Related CN105955219B (en) | 2016-05-30 | 2016-05-30 | Distributed dynamic procedure failure testing method based on mutual information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105955219B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106444666B (en) * | 2016-09-22 | 2018-11-20 | 宁波大学 | Dynamic process monitoring method based on the dynamic distributed pca model of weighting type |
| CN106647650B (en) * | 2016-09-22 | 2018-11-20 | 宁波大学 | Distributing Industrial Process Monitoring method based on variable weighting pca model |
| CN106444706B (en) * | 2016-09-22 | 2018-08-17 | 宁波大学 | The industrial process fault detection method kept based on data neighborhood feature |
| CN106354125A (en) * | 2016-11-09 | 2017-01-25 | 上海电机学院 | Method for utilizing block PCA (Principal Component Analysis) to detect fault of chemical process |
| CN107153409B (en) * | 2017-06-02 | 2019-08-16 | 宁波大学 | A kind of nongausian process monitoring method based on missing variable modeling thinking |
| CN108388234B (en) * | 2018-03-19 | 2020-06-16 | 宁波大学 | Fault monitoring method based on relevance division multi-variable block PCA model |
| CN108762228B (en) * | 2018-05-25 | 2020-08-04 | 江南大学 | Distributed PCA-based multi-working-condition fault monitoring method |
| CN108762242B (en) * | 2018-06-11 | 2020-10-27 | 宁波大学 | Distributed fault detection method based on multiple typical correlation analysis models |
| CN109146004B (en) * | 2018-10-09 | 2021-07-23 | 宁波大学 | Dynamic process monitoring method based on iteration missing data estimation strategy |
| CN109407652B (en) * | 2018-12-10 | 2020-03-06 | 中国石油大学(华东) | Multivariable industrial process fault detection method based on main and auxiliary PCA models |
| CN109634240B (en) * | 2018-12-13 | 2021-03-09 | 宁波大学 | Dynamic process monitoring method based on novel dynamic principal component analysis |
| CN110501986B (en) * | 2019-09-03 | 2021-06-18 | 山东科技大学 | Quality related process monitoring method based on weighted partial mutual information |
| CN110794797A (en) * | 2019-11-27 | 2020-02-14 | 江南大学 | PCA fault monitoring method based on mutual information and multi-block information extraction |
| CN111160776A (en) * | 2019-12-30 | 2020-05-15 | 华东理工大学 | Method for detecting abnormal working condition in sewage treatment process by utilizing block principal component analysis |
| CN112947649B (en) * | 2021-03-19 | 2021-11-23 | 安阳师范学院 | Multivariate process monitoring method based on mutual information matrix projection |
| CN112947290B (en) * | 2021-05-16 | 2021-08-20 | 北京赛博联物科技有限公司 | Edge cloud cooperation-based equipment state monitoring method and system and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1870538A (en) * | 2006-05-08 | 2006-11-29 | 国家数字交换系统工程技术研究中心 | Method and system for fault management |
| CN102299829A (en) * | 2011-09-01 | 2011-12-28 | 北京市天元网络技术股份有限公司 | Network failure probing and positioning method |
| EP2533549A1 (en) * | 2011-06-07 | 2012-12-12 | Alcatel Lucent | Fault detector for optical network communication system |
| US9002678B1 (en) * | 2014-01-10 | 2015-04-07 | King Fahd University Of Petroleum And Minerals | Unified approach to detection and isolation of parametric faults using a kalman filter residual-based approach |
| CN104753178A (en) * | 2015-04-16 | 2015-07-01 | 河南行知专利服务有限公司 | Power grid fault handling system |
| CN105530122A (en) * | 2015-12-03 | 2016-04-27 | 国网江西省电力公司信息通信分公司 | Network failure diagnosis method based on selective hidden Naive Bayesian classifier |
-
2016
- 2016-05-30 CN CN201610389000.4A patent/CN105955219B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1870538A (en) * | 2006-05-08 | 2006-11-29 | 国家数字交换系统工程技术研究中心 | Method and system for fault management |
| EP2533549A1 (en) * | 2011-06-07 | 2012-12-12 | Alcatel Lucent | Fault detector for optical network communication system |
| CN102299829A (en) * | 2011-09-01 | 2011-12-28 | 北京市天元网络技术股份有限公司 | Network failure probing and positioning method |
| US9002678B1 (en) * | 2014-01-10 | 2015-04-07 | King Fahd University Of Petroleum And Minerals | Unified approach to detection and isolation of parametric faults using a kalman filter residual-based approach |
| CN104753178A (en) * | 2015-04-16 | 2015-07-01 | 河南行知专利服务有限公司 | Power grid fault handling system |
| CN105530122A (en) * | 2015-12-03 | 2016-04-27 | 国网江西省电力公司信息通信分公司 | Network failure diagnosis method based on selective hidden Naive Bayesian classifier |
Non-Patent Citations (1)
| Title |
|---|
| 基于互信息的PCA 方法及其在过程监测中的应用;童楚东，等;《化工学报》;20151031;第66卷(第10期);第4101-4106页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105955219A (en) | 2016-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105955219B (en) | Distributed dynamic procedure failure testing method based on mutual information | |
| CN106092625B (en) | The industrial process fault detection method merged based on amendment type independent component analysis and Bayesian probability | |
| CN101446831B (en) | Decentralized process monitoring method | |
| CN110738274A (en) | nuclear power device fault diagnosis method based on data driving | |
| CN104699077B (en) | A kind of failure variable partition method based on nested iterations Fei Sheer discriminant analyses | |
| CN108508865B (en) | A kind of fault detection method based on distributing OSC-PLS regression model | |
| CN106940808A (en) | A kind of fault detection method based on modified Principal Component Analysis Model | |
| CN108345284A (en) | A kind of quality dependent failure detection method becoming gauge block based on two | |
| CN105629958B (en) | A kind of batch process method for diagnosing faults based on sub-period MPCA SVM | |
| CN108388234A (en) | A kind of fault monitoring method dividing changeable gauge block pca model based on correlation | |
| CN108445867A (en) | A kind of nongausian process monitoring method based on distributing ICR models | |
| CN104731083B (en) | A kind of industrial method for diagnosing faults and application based on self-adaptive feature extraction | |
| CN105404280A (en) | Industrial process fault detection method based on autoregression dynamic hidden variable model | |
| CN105607631B (en) | The weak fault model control limit method for building up of batch process and weak fault monitoring method | |
| CN108375965A (en) | A kind of nongausian process monitoring method rejected based on changeable gauge block crossing dependency | |
| CN108345294B (en) | A kind of fault detection method based on distributing principal component regression model | |
| CN108919755A (en) | A kind of distributed fault detection method based on muti-piece Nonlinear and crossing relational model | |
| CN107657274A (en) | A kind of y-bend SVM tree unbalanced data industry Fault Classifications based on k means | |
| CN114757269A (en) | Complex process refined fault detection method based on local subspace-neighborhood preserving embedding | |
| CN108830006A (en) | Linear-nonlinear industrial processes fault detection method based on the linear evaluation factor | |
| CN108762242A (en) | A kind of distributed fault detection method based on polylith canonical correlation analysis model | |
| CN104731970B (en) | The quality testing and control method of the multi-source heterogeneous data of highway | |
| CN106354125A (en) | Method for utilizing block PCA (Principal Component Analysis) to detect fault of chemical process | |
| CN104317778A (en) | Massive monitoring data based substation equipment fault diagnosis method | |
| CN113486950B (en) | Intelligent pipe network water leakage detection method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221123 Address after: Room 2202, 22 / F, Wantong building, No. 3002, Sungang East Road, Sungang street, Luohu District, Shenzhen City, Guangdong Province Patentee after: Shenzhen dragon totem technology achievement transformation Co.,Ltd. Address before: 315211, Fenghua Road, Jiangbei District, Zhejiang, Ningbo 818 Patentee before: Ningbo University Effective date of registration: 20221123 Address after: Room 302, Building 3, No. 16 Ganyangshujia, Chaoyang District, Beijing 100101 Patentee after: Liu Renda Patentee after: Mu Dan Patentee after: Dai Wenbo Patentee after: Zhou Haipeng Address before: Room 2202, 22 / F, Wantong building, No. 3002, Sungang East Road, Sungang street, Luohu District, Shenzhen City, Guangdong Province Patentee before: Shenzhen dragon totem technology achievement transformation Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230110 Address after: Room 10, Floor 10, Floor 101, Building 46, Building 46, No. 69, Yanfu Road, Fangshan District, Beijing 102400 Patentee after: Beijing Ansheng Huaxin Technology Co.,Ltd. Address before: Room 302, Building 3, No. 16 Ganyangshujia, Chaoyang District, Beijing 100101 Patentee before: Liu Renda Patentee before: Mu Dan Patentee before: Dai Wenbo Patentee before: Zhou Haipeng |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180817 |