CN116088445A - Distributed industrial process monitoring method and device and electronic equipment - Google Patents
Distributed industrial process monitoring method and device and electronic equipment Download PDFInfo
- Publication number
- CN116088445A CN116088445A CN202211737198.2A CN202211737198A CN116088445A CN 116088445 A CN116088445 A CN 116088445A CN 202211737198 A CN202211737198 A CN 202211737198A CN 116088445 A CN116088445 A CN 116088445A
- Authority
- CN
- China
- Prior art keywords
- variable
- data set
- real
- sub
- fault detection
- 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
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 83
- 238000012544 monitoring process Methods 0.000 title claims abstract description 56
- 238000001514 detection method Methods 0.000 claims abstract description 111
- 239000013598 vector Substances 0.000 claims abstract description 48
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 40
- 238000012806 monitoring device Methods 0.000 claims abstract description 12
- 238000004590 computer program Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 19
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 239000004973 liquid crystal related substance Substances 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 230000009977 dual effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000009776 industrial production Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
- G05B19/41875—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by quality surveillance of production
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/32—Operator till task planning
- G05B2219/32252—Scheduling production, machining, job shop
-
- 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)
- General Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention provides a distributed industrial process monitoring method, a distributed industrial process monitoring device and electronic equipment. The method comprises the following steps: collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets; respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set; collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets; and adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result. The method has better monitoring performance on process variables which do not meet Gaussian assumptions and linear assumptions.
Description
Technical Field
The present invention relates to the field of industrial automation technologies, and in particular, to a method and an apparatus for monitoring a decentralized industrial process, and an electronic device.
Background
With the development of modern industry, the industrial production scale is larger and larger, the working environment is also more and more complex, and the industrial process monitoring technology plays an increasingly important role in ensuring safe production and improving the process productivity.
However, the conventional centralized industrial process monitoring mode is not easy to handle the complex features of high dimension, autocorrelation and the like of the modern industrial production process, so the distributed monitoring (Decentralized Monitoring) mode is widely studied and applied to reduce the complexity of the process and the monitoring redundancy. The distributed monitoring mode is to firstly perform variable decomposition on process variables collected by an industrial process, and then build a process monitoring model for each sub-block after decomposition to monitor faults.
Currently, methods based on multivariate statistics to monitor faults of sub-blocks have been widely studied, however, some conventional multivariate statistical methods, such as: PCA, CCA, PLS, etc. Process variables of industrial processes are required to meet gaussian assumptions and linear assumptions, but this is difficult to meet in practical industrial processes and therefore has poor monitoring performance.
Disclosure of Invention
The invention provides a distributed industrial process monitoring method, a distributed industrial process monitoring device and electronic equipment, which are used for solving the problems that the process variable of an industrial process in the prior art meets Gaussian assumption and linear assumption, and the monitoring performance is poor under the condition of unsatisfied process variable. A distributed industrial process monitoring method is realized, by establishing a fault detection model described based on support vector data, the model does not need a process variable to meet Gaussian assumption and linear assumption, so that the monitoring performance of the process variable which does not meet Gaussian assumption and linear assumption can be improved, and the fault detection model has better robustness to abnormal values in variable data in a monitoring process.
The invention provides a decentralized industrial process monitoring method, which comprises the following steps:
collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets;
respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set;
collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets;
adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result;
the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
According to the distributed industrial process monitoring method provided by the invention, the variable data set of the industrial process in the normal running state is collected, and variable decomposition is carried out on the variable data set to obtain a plurality of variable sub-data sets, and the distributed industrial process monitoring method comprises the following steps:
collecting a variable data set x=r of the industrial process in a normal operating state n×m Wherein n is the number of samples, m is the number of variables, and R is a real number;
converting the variable data set into a graph model containing m nodes based on m variables;
defining a division index, dividing m nodes of the graph model into a plurality of target sub-blocks based on the division index, wherein each target sub-block corresponds to the variable sub-data set.
According to the distributed industrial process monitoring method provided by the invention, a plurality of fault detection models based on support vector data description are respectively built by adopting each variable sub-data set, and the distributed industrial process monitoring method comprises the following steps:
for each variable sub-data set, projecting the variable sub-data set to a feature space through a kernel function;
screening out support vectors from all sample vectors in the variable subset data set to obtain a support vector sample set;
and obtaining the sphere center and the radius of the hypersphere with the smallest volume in the characteristic space based on the support vector sample set, and taking the hypersphere as a fault detection model.
According to the method for monitoring the distributed industrial process provided by the invention, each fault detection model is adopted to respectively detect each corresponding real-time variable sub-data set so as to obtain a fault detection result, and the method comprises the following steps:
calculating the target distance from the real-time variable sub-data set to the sphere center of the corresponding fault detection model according to each real-time variable sub-data set;
and when the target distance is larger than the radius of the fault detection model, indicating that the industrial system has a fault.
The distributed industrial process monitoring method provided by the invention further comprises the following steps:
and respectively calculating the contribution rate index of each real-time variable in the real-time variable sub-data set for the real-time variable sub-data set with faults, and determining the position with faults in the industrial system based on the contribution rate index.
The distributed industrial process monitoring method provided by the invention further comprises the following steps:
probability fusion is carried out on fault detection results of all the real-time variable sub-data sets based on Bayesian reasoning so as to obtain a global detection index BIC;
when the global detection index BIC is larger than the prior probability of the fault state of the industrial system, the industrial system is in the fault state;
and checking fault positioning results of the real-time variable sub-data sets, and processing faults of the industrial system.
The invention also provides a decentralized industrial process monitoring device comprising:
the first acquisition and decomposition module is used for acquiring a variable data set in a normal operation state of the industrial process and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets;
the model building module is used for respectively building a plurality of fault detection models based on the support vector data description by adopting each variable sub-data set;
the second acquisition and decomposition module is used for acquiring a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set so as to obtain a plurality of real-time variable sub-data sets;
the detection module is used for respectively detecting each corresponding real-time variable sub-data set by adopting each fault detection model so as to obtain an industrial system fault detection result;
the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the decentralized industrial process monitoring method according to any one of the above when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a decentralized industrial process monitoring method as described in any one of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a decentralized industrial process monitoring method according to any one of the above.
According to the distributed industrial process monitoring method, device and electronic equipment provided by the invention, the fault detection model based on the support vector data description is established by aiming at each sub-part of variable division of the process variable, the model does not need the process variable to meet Gaussian assumption and linear assumption, the monitoring performance on the process variable which does not meet Gaussian assumption and linear assumption is better, and the fault detection model has better robustness on abnormal values in variable data in the monitoring process.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow diagram of a distributed industrial process monitoring method provided by the present invention;
FIG. 2 is a flow chart of a method for variable decomposition of a variable dataset provided by the present invention;
FIG. 3 is a flow chart of a method for establishing a fault detection model provided by the invention;
FIG. 4 is a schematic diagram of a variable sub-dataset provided by the present invention;
FIG. 5 is a flow chart of a method for obtaining fault detection results according to the present invention;
FIG. 6 is a second flow chart of a distributed industrial process monitoring method according to the present invention;
FIG. 7 is a schematic diagram of a distributed industrial process monitoring device provided by the present invention;
fig. 8 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The decentralized industrial process monitoring method of the present invention is described below in connection with fig. 1-6, as shown in fig. 1, and comprises:
s101: and collecting variable data sets in the normal running state of the industrial process, and performing variable decomposition on the variable data sets to obtain a plurality of variable sub-data sets.
Specifically, a variable data set of the industrial process in a normal running state is collected, wherein the variable data set comprises samples at a plurality of moments, and each sample comprises a plurality of variables. The variable decomposition is to decompose the variable types in the variable dataset, and the variable decomposition method approximately comprises two types: knowledge driving and data driving, wherein the knowledge driving method needs to obtain mechanism knowledge in advance, and the mechanism knowledge can be assumed to be obtained in advance; data-driven methods typically decompose based on correlations between variables, assuming that the mechanism knowledge is not known in advance or is partially known.
For example, the method may use a data driven method to decompose a variable dataset to obtain multiple variable sub-datasets.
S102: and respectively establishing a plurality of fault detection models based on the support vector data description by adopting each variable sub-data set.
Specifically, the fault detection model based on the support vector data description (Support Vector Data Description, SVDD) does not require that variables in the variable sub-data set satisfy gaussian assumption and linear assumption, and has better monitoring performance for variable data which do not satisfy gaussian assumption and linear assumption.
S103: and acquiring a real-time variable data set under the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets.
Specifically, a real-time variable dataset of the running state of the industrial process is acquired, namely, the variable dataset of the current moment is acquired when the industrial process is monitored, so that only one group of samples exists in the real-time variable dataset. Based on a variable decomposition principle of the variable data set, decomposing the real-time variable data set to obtain a plurality of real-time variable sub-data sets. Exemplary, e.g., variable m in a variable dataset, n samples 1 、m 2 Dividing into a group, decomposing the real-time variable data set to obtain the variable m 1 、m 2 Divided into a group.
S104: adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result; the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
Specifically, as can be seen from the above steps, the decomposition principle of the real-time variable dataset and the variable dataset is consistent, so that each variable sub-dataset has a certain corresponding relationship with each real-time variable sub-dataset, that is, the variable sub-datasets with the same variable group have a corresponding relationship with the real-time variable sub-dataset. It can be understood that the fault detection model established by using the variable sub-data set also has a corresponding relationship with the real-time variable sub-data set corresponding to the variable sub-data set, and is used for detecting the corresponding real-time variable sub-data set so as to obtain the fault detection result of the industrial system.
Illustratively, an industrial system fails when at least one of the plurality of real-time variable sub-data sets is detected as likely to fail. After the industrial system fails, the equipment corresponding to each variable in the real-time variable subset data set with the detected failure can be subjected to failure detection, and the failure can be eliminated as soon as possible.
According to the distributed industrial process monitoring method, the fault detection model based on the support vector data description is established by aiming at each sub-part of variable division of the process variable, the model does not need to meet Gaussian assumption and linear assumption of the process variable, the monitoring performance of the process variable which does not meet Gaussian assumption and linear assumption is good, and the fault detection model has good robustness on abnormal values in variable data in the monitoring process.
In one embodiment, as shown in fig. 2, the collecting a variable dataset of a normal operation state of an industrial process, and performing variable decomposition on the variable dataset to obtain a plurality of variable sub-datasets, includes:
s201: collecting a variable data set x=r of the industrial process in a normal operating state n×m Wherein n is the same asThe number, m, is the number of variables and R is a real number.
S202: the variable dataset is converted into a graph model containing m nodes based on m variables.
S203: defining a division index, dividing m nodes of the graph model into a plurality of target sub-blocks based on the division index, wherein each target sub-block corresponds to the variable sub-data set.
Specifically, variable data sets under normal operation state of complex industrial process are collected
Wherein n is the number of samples, which indicates that variable data acquisition is performed at n times, m is the number of variables, and R is a real number.
Converting a variable dataset into a graph model g= (V, E) containing m nodes, where v= { x 1 ,…x r ,…x m And represents a collection of nodes (i.e., variables) in the graph model,
representing a set of edges (i.e., weights between variables) in the graph model.
When dividing the graph model g= (V, E), the division index Q is defined first:
wherein e rl Representing node x r And node x l Weights in between; p (x) r ) And p (x) l ) Respectively represent node x r And node x l Edge probability of (2); p (x) r ,x l ) Representing node x r And node x l Is a joint probability of (2); a, a r ＝∑ l e rl 。
The graph model G= (V, E) is segmented by an MI-FN (Mutual Information-Fast Newman, mutually trusted Fast Newman) method, and the specific segmentation method is as follows:
each node in the graph model is defined as a separate initial block.
And calculating the dividing index value after combining any two initial blocks, and finding out the two initial blocks with the maximum dividing index value Q value increase for combining. And (3) for the multiple blocks after combination, carrying out any two-by-two combination, respectively calculating the Q value after combination, finding out two blocks with the largest increase of the Q value for combination, and repeating the steps in a circulating way until the Q value is unchanged and does not increase any more after the two-by-two combination, and indicating that the combination is completed and taking the previous combination result with the unchanged Q value as a division result.
After the merging is completed, the initial blocks which are merged together form a target sub-block, and each target sub-block corresponds to the corresponding sub-data set.
By the method, m nodes in the graph model are divided into q subblocks, namely variable data sets
Is divided into x= { X 1 ,...,X c ,...,X q Q variable sub-data sets in total.
The method for converting the variable data set into the graph model and then dividing the variable data set is a pure data-driven variable decomposition method, and the method can improve the dividing efficiency and practicability of the variable data set (namely, the dividing result is more in accordance with practical mechanism knowledge). Meanwhile, variable data sets are respectively monitored after variable decomposition, and the condition that monitoring performance is influenced due to overlarge scale of the variable is improved by the distributed monitoring method.
In one embodiment, as shown in fig. 3, the employing each of the variable sub-data sets to respectively build a plurality of fault detection models described based on support vector data includes:
s301: for each of the variable sub-data sets, projecting the variable sub-data set into a feature space through a kernel function.
S302: and screening out support vectors from all sample vectors in the variable subset data set to obtain a support vector sample set.
S303: and obtaining the sphere center and the radius of the hypersphere with the smallest volume in the characteristic space based on the support vector sample set, and taking the hypersphere as a fault detection model.
Specifically, for each variable sub-data set
The corresponding fault detection model is obtained by the following method:
the sample vectors in the variable subset are passed through a kernel function K (x i ,x j )＝(Φ(x i )×Φ(x j ) The method comprises the steps of) projecting the super sphere to a feature space, and obtaining a fault detection model by searching the super sphere with the minimum volume in the feature space, wherein the super sphere is used as the fault detection model.
In order to obtain the supersphere with the smallest volume in the feature space, the problem of searching the supersphere is converted into the following optimization problem:
wherein R is c Is the radius of the hypersphere; a, a c Is the sphere center of the super sphere; c is a penalty factor for weighing the hypersphere; zeta type toy i Is a relaxation factor.
Solving the dual problem of the optimization problem by adopting a Lagrangian multiplier method:
wherein alpha is i For sample vector X i Corresponding Lagrangian coefficient, alpha j For sample vector X j Corresponding lagrangian coefficients.
By solving the dual problem, all sample vectors X can be obtained i The Lagrange coefficient is satisfied with 0 < alpha i The sample vector < C is defined as a support vector, resulting in a support vector set SV.
Based on the support vector set SV, the radius R of the hypersphere with the minimum volume is obtained c And sphere center a c ：
Wherein X is i And E, SV, the hypersphere is a fault detection model.
Exemplary, as shown in FIG. 4, if the variable data set is subjected to variable decomposition, and one variable sub-data set is shown in the dashed box, then the sample vector X 1 Is the vector in the solid line box in fig. 4.
In one embodiment, as shown in fig. 5, the detecting each real-time variable sub-data set by using each fault detection model to obtain a fault detection result includes:
s501: and calculating the target distance from the real-time variable sub-data set to the sphere center of the corresponding fault detection model according to each real-time variable sub-data set.
S502: and when the target distance is larger than the radius of the fault detection model, indicating that the industrial system has a fault.
Specifically, for each real-time variable sub-dataset
Calculating the sphere center a of the hypersphere which is the failure detection model corresponding to the hypersphere c Target distance d of (2) c When the target distance is larger than the radius of the corresponding fault detection model, namely the hypersphere, the device corresponding to the real-time variable sub-data set is indicated to have faults, namely the industrial system has faults.
Exemplary, the real-time variable sub-dataset x is calculated new And the corresponding fault detection model is the sphere center a of the hypersphere c Target distance d of (2) c ：
Wherein, when:
in one embodiment, further comprising:
and respectively calculating the contribution rate index of each real-time variable in the real-time variable sub-data set for the real-time variable sub-data set with faults, and determining the position with faults in the industrial system based on the contribution rate index.
Specifically, in order to further improve the speed of fault detection in the industrial system, for the real-time variable sub-data set with faults, the contribution rate index of each real-time variable is calculated, and the greater the contribution rate index is, the more likely the equipment corresponding to the variable is to be faulty. Therefore, the position of the fault in the industrial system can be rapidly positioned, and the speed of fault investigation in the industrial system is improved.
For example, based on the hypersphere disclosed in the above embodiment, the method for determining the contribution rate index of each real-time variable may be:
obtaining the sphere center a of the super sphere through a corresponding fault detection model c Coordinates of (c)
For real-time subsampled dataset +.>
Respectively adopt->
Instead of a 1 ，/>
Instead of a k ，……/>
Replace->
In this way a plurality of new sample vectors can be constructed. For example->
Instead of a k The new sample vector constructed is +.>
For the sample vector, the kth variable +.>
The contribution ratio of (2) is as follows:
wherein, the liquid crystal display device comprises a liquid crystal display device,
in one embodiment, as shown in fig. 6, further comprising:
s601: and carrying out probability fusion on fault detection results of each real-time variable sub-dataset based on Bayesian reasoning so as to obtain a global detection index BIC.
S602: when the global detection index BIC is larger than the prior probability of the fault state of the industrial system, the industrial system is in the fault state.
S603: and checking fault positioning results of the real-time variable sub-data sets, and processing faults of the industrial system.
Specifically, probability fusion is carried out on the fault detection result of each sub-block by using Bayesian reasoning, so as to obtain a global monitoring index BIC:
wherein, the liquid crystal display device comprises a liquid crystal display device,
P(d c )＝P(d c /N)P(N)+P(d c /F)P(F)；/>
n represents a normal state, and F represents a fault state; p (N) and the prior probability representing the normal state are defined as 1-beta, P (F) represents the prior probability of the fault state and is defined as beta, the prior probability is artificially set, and for example, the prior probability beta is more than or equal to 80 percent.
Judging whether the BIC index meets the condition: BIC is less than or equal to beta, if the BIC is satisfied, the industrial system is in a normal state at the current moment; if not, the industrial system is in a fault state at the current moment, and fault detection and fault positioning results of variable sub-data sets are checked, so that fault processing is performed quickly and timely.
The global detection index can be used for remotely monitoring and fault positioning of the industrial process, and is convenient for remote management.
The decentralized industrial process monitoring device provided by the invention is described below, and the decentralized industrial process monitoring device described below and the decentralized industrial process monitoring method described above can be referred to correspondingly.
The decentralized industrial process monitoring device, as shown in fig. 7, comprises:
the first acquisition and decomposition module 701 is configured to acquire a variable data set in a normal operation state of an industrial process, and perform variable decomposition on the variable data set to obtain a plurality of variable sub-data sets;
a model building module 702, configured to build a plurality of fault detection models based on the support vector data description by using each of the variable sub-data sets;
the second collection and decomposition module 703 is configured to collect a real-time variable dataset in the operating state of the industrial process, and decompose the real-time variable dataset based on a variable decomposition principle of the variable dataset to obtain a plurality of real-time variable sub-datasets;
the detection module 704 is configured to detect each corresponding real-time variable sub-data set by using each fault detection model, so as to obtain an industrial system fault detection result;
the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
According to the distributed industrial process monitoring device, the fault detection model based on the support vector data description is established by aiming at each sub-part of variable division of the process variable, the model does not need to meet Gaussian assumption and linear assumption of the process variable, the distributed industrial process monitoring device has good monitoring performance on the process variable which does not meet Gaussian assumption and linear assumption, and the fault detection model has good robustness on abnormal values in variable data in a monitoring process.
In one embodiment, the first acquisition decomposition module 701 is specifically configured to:
collecting a variable data set x=r of the industrial process in a normal operating state n×m Where n is the number of samples, m is the number of variables, and R is a real number.
The variable dataset is converted into a graph model containing m nodes based on m variables.
Defining a division index, dividing m nodes of the graph model based on the division index, dividing the graph model into a plurality of target sub-blocks, wherein each target sub-block corresponds to the variable sub-data set.
In one embodiment, the model building module 702 is specifically configured to:
for each of the variable sub-data sets, projecting the variable sub-data set into a feature space through a kernel function.
And screening out support vectors from all sample vectors in the variable subset data set to obtain a support vector sample set.
And obtaining the sphere center and the radius of the hypersphere with the smallest volume in the characteristic space based on the support vector sample set, and taking the hypersphere as a fault detection model.
In one embodiment, the detection module 704 is specifically configured to:
and calculating the target distance from the real-time variable sub-data set to the sphere center of the corresponding fault detection model according to each real-time variable sub-data set.
And when the target distance is larger than the radius of the fault detection model, indicating that the industrial system has a fault.
In one embodiment, the decentralized industrial process monitoring device further comprises:
and the positioning module is used for respectively calculating contribution rate indexes of all real-time variables in the real-time variable sub-data set for the real-time variable sub-data set with faults, and determining the position with faults in the industrial system based on the contribution rate indexes.
In one embodiment, the decentralized industrial process monitoring device further comprises:
the global index obtaining module is used for carrying out probability fusion on fault detection results of each real-time variable sub-dataset based on Bayesian reasoning so as to obtain a global detection index BIC.
And the judging module is used for judging that the industrial system is in a fault state when the global detection index BIC is larger than the prior probability of the fault state of the industrial system.
And the checking module is used for checking fault positioning results of the real-time variable sub-data sets and processing faults of the industrial system.
Fig. 8 illustrates a physical structure diagram of an electronic device, as shown in fig. 8, which may include: processor 810, communication interface (Communications Interface) 820, memory 830, and communication bus 840, wherein processor 810, communication interface 820, memory 830 accomplish communication with each other through communication bus 840. The processor 810 may invoke logic instructions in the memory 830 to perform a decentralized industrial process monitoring method comprising: collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets; respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set; collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets; adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result; the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
Further, the logic instructions in the memory 830 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, is capable of performing the method of decentralized industrial process monitoring provided by the methods described above, the method comprising: collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets; respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set; collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets; adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result; the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method of decentralized industrial process monitoring provided by the methods described above, the method comprising: collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets; respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set; collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets; adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result; the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A method of decentralized industrial process monitoring, comprising:
collecting a variable data set in a normal operation state of an industrial process, and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets;
respectively establishing a plurality of fault detection models based on support vector data description by adopting each variable sub-data set;
collecting a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set to obtain a plurality of real-time variable sub-data sets;
adopting each fault detection model to respectively detect each corresponding real-time variable sub-data set so as to obtain an industrial system fault detection result;
the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
2. The method of claim 1, wherein collecting a variable dataset for a normal operating condition of an industrial process and performing variable decomposition on the variable dataset to obtain a plurality of variable sub-datasets comprises:
collecting a variable data set x=r of the industrial process in a normal operating state n×m Wherein n is the number of samples, m is the number of variables, and R is a real number;
converting the variable data set into a graph model containing m nodes based on m variables;
defining a division index, dividing m nodes of the graph model based on the division index, dividing the graph model into a plurality of target sub-blocks, wherein each target sub-block corresponds to the variable sub-data set.
3. The method of claim 2, wherein said using each of said variable sub-data sets to separately build a plurality of fault detection models based on support vector data descriptions comprises:
for each variable sub-data set, projecting the variable sub-data set to a feature space through a kernel function;
screening out support vectors from all sample vectors in the variable subset data set to obtain a support vector sample set;
and obtaining the sphere center and the radius of the hypersphere with the smallest volume in the characteristic space based on the support vector sample set, and taking the hypersphere as a fault detection model.
4. A distributed industrial process monitoring method according to claim 3, wherein said employing each of said fault detection models to separately detect each corresponding real-time variable sub-data set to obtain a fault detection result comprises:
calculating the target distance from the real-time variable sub-data set to the sphere center of the corresponding fault detection model according to each real-time variable sub-data set;
and when the target distance is larger than the radius of the fault detection model, indicating that the industrial system has a fault.
5. The decentralized industrial process monitoring method according to any one of claims 1 to 4, further comprising:
and respectively calculating the contribution rate index of each real-time variable in the real-time variable sub-data set for the real-time variable sub-data set with faults, and determining the position with faults in the industrial system based on the contribution rate index.
6. The distributed industrial process monitoring method of claim 5, further comprising:
probability fusion is carried out on fault detection results of all the real-time variable sub-data sets based on Bayesian reasoning so as to obtain a global detection index BIC;
when the global detection index BIC is larger than the prior probability of the fault state of the industrial system, the industrial system is in the fault state;
and checking fault positioning results of the real-time variable sub-data sets, and processing faults of the industrial system.
7. A decentralized industrial process monitoring device, comprising:
the first acquisition and decomposition module is used for acquiring a variable data set in a normal operation state of the industrial process and performing variable decomposition on the variable data set to obtain a plurality of variable sub-data sets;
the model building module is used for respectively building a plurality of fault detection models based on the support vector data description by adopting each variable sub-data set;
the second acquisition and decomposition module is used for acquiring a real-time variable data set in the running state of the industrial process, and decomposing the real-time variable data set based on a variable decomposition principle of the variable data set so as to obtain a plurality of real-time variable sub-data sets;
the detection module is used for respectively detecting each corresponding real-time variable sub-data set by adopting each fault detection model so as to obtain an industrial system fault detection result;
the real-time variable sub-data set corresponding to the fault detection model is the real-time variable sub-data set with the same variable category as the variable sub-data set in which the fault part detection model is built.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the decentralized industrial process monitoring method according to any one of claims 1 to 6 when the program is executed by the processor.
9. A non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the decentralized industrial process monitoring method according to any one of claims 1 to 6.
10. A computer program product comprising a computer program which, when executed by a processor, implements the decentralized industrial process monitoring method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211737198.2A CN116088445A (en) | 2022-12-30 | 2022-12-30 | Distributed industrial process monitoring method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211737198.2A CN116088445A (en) | 2022-12-30 | 2022-12-30 | Distributed industrial process monitoring method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116088445A true CN116088445A (en) | 2023-05-09 |
Family
ID=86186372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211737198.2A Pending CN116088445A (en) | 2022-12-30 | 2022-12-30 | Distributed industrial process monitoring method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116088445A (en) |
-
2022
- 2022-12-30 CN CN202211737198.2A patent/CN116088445A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9542255B2 (en) | Troubleshooting based on log similarity | |
| Caceres et al. | A probabilistic Bayesian recurrent neural network for remaining useful life prognostics considering epistemic and aleatory uncertainties | |
| Su et al. | Detecting outlier machine instances through gaussian mixture variational autoencoder with one dimensional cnn | |
| CN114465874B (en) | Fault prediction method, device, electronic equipment and storage medium | |
| CN110570544A (en) | method, device, equipment and storage medium for identifying faults of aircraft fuel system | |
| CN112187554A (en) | Operation and maintenance system fault positioning method and system based on Monte Carlo tree search | |
| CN117041029A (en) | Network equipment fault processing method and device, electronic equipment and storage medium | |
| Wang et al. | Incremental causal graph learning for online root cause analysis | |
| CN114637649B (en) | Alarm root cause analysis method and device based on OLTP database system | |
| CN116167370A (en) | Log space-time characteristic analysis-based distributed system anomaly detection method | |
| CN116743555A (en) | Robust multi-mode network operation and maintenance fault detection method, system and product | |
| CN113676343B (en) | Fault source positioning method and device for power communication network | |
| CN110244690B (en) | Multivariable industrial process fault identification method and system | |
| CN112396166A (en) | Graph convolution neural network training method and device based on mixed granularity aggregator | |
| CN114297453B (en) | Alarm prediction method and device, electronic equipment and storage medium | |
| CN116088445A (en) | Distributed industrial process monitoring method and device and electronic equipment | |
| CN113824583B (en) | Root element positioning method and device, computer equipment and storage medium | |
| CN117640346B (en) | Communication equipment fault diagnosis method, storage medium and computer equipment | |
| CN115174421B (en) | Network fault prediction method and device based on self-supervision unwrapping hypergraph attention | |
| CN118033429A (en) | Method, device, equipment and storage medium for predicting voltage faults of storage battery | |
| CN115001997B (en) | Extreme value theory-based smart city network equipment performance abnormal threshold evaluation method | |
| CN117596598B (en) | Unmanned aerial vehicle communication protocol data anomaly detection method | |
| CN116125922B (en) | Complex industrial process monitoring method and system based on parallel dictionary learning | |
| CN116362628B (en) | Method and device for analyzing and evaluating running state of power system | |
| US20240061740A1 (en) | Disentangled graph learning for incremental causal discovery and root cause analysis |
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 |