CN106951707B - Method for constructing Bayesian network of reconfigurable system - Google Patents
Method for constructing Bayesian network of reconfigurable system Download PDFInfo
- Publication number
- CN106951707B CN106951707B CN201710168449.2A CN201710168449A CN106951707B CN 106951707 B CN106951707 B CN 106951707B CN 201710168449 A CN201710168449 A CN 201710168449A CN 106951707 B CN106951707 B CN 106951707B
- Authority
- CN
- China
- Prior art keywords
- fault
- test
- switch
- node
- bayesian network
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention provides a Bayesian network construction method of a reconfigurable system. The method comprises the following steps: (1) establishing a model for the correlation of reconfigurable system faults and tests; (2) representing the correlation between the reconfigurable system fault and the test; (3) determining a Bayesian network structure of a reconfigurable system; (4) and determining Bayesian network parameters of a reconstruction system and representing by adopting a tree structure. The invention has the beneficial effects that: a method for constructing a programmed reconfigurable system Bayesian network is provided, the Bayesian network parameter storage and retrieval are optimized, a new solution is provided for solving the problem of constructing the reconfigurable system Bayesian network, and the application of the Bayesian network in the field of fault diagnosis can be promoted.
Description
Technical Field
The invention relates to a Bayesian network construction method for a reconfigurable system, which is suitable for Bayesian network construction of the reconfigurable system and further can support fault diagnosis and reasoning of equipment.
Background
Fault diagnosis is critical to the proper safe operation of the equipment. The Bayesian network has strong capability in the fields of uncertain knowledge expression and reasoning as a credibility network, and has been successfully applied to fault diagnosis. However, the construction of the bayesian network has been a very difficult task, requires a great deal of expert knowledge, is a very time-consuming and labor-consuming task, and limits the application thereof in large-scale systems. Currently, the construction of the bayesian network is mainly manual, mainly aims at a single-mode small-scale system, and lacks a systematic automatic method. Particularly, for a reconfigurable system, because the system structure changes in the application process, the expression of the bayesian network is difficult, and a bayesian network construction method applicable to engineering is lacking at present.
Disclosure of Invention
The invention aims to provide a Bayesian network construction method suitable for a reconfigurable system, and provides support for fault diagnosis based on a Bayesian network.
To achieve the above object, the present invention provides the following solutions: 1. a method for constructing a Bayesian network of a reconfigurable system comprises the following steps:
(1) establishing a model for the correlation of reconfigurable system faults and tests;
(2) representing the correlation between the reconfigurable system fault and the test;
(3) determining a Bayesian network structure of a reconfigurable system;
(4) and determining Bayesian network parameters of a reconstruction system and representing by adopting a tree structure.
Further, the model in the step (1) comprises four elements, namely a fault, a switch, a test point and a connecting line for connecting the fault, the test point and the switch; the working mode of the model is as follows: when a connection path exists between the fault and the test, the fault is related to the test point, otherwise, the fault is not related; a switch can be arranged between the fault and the test, and the correlation between the fault and the test is influenced by the opening and closing of the switch, so that the reconstruction of the system is realized.
Further, the correlation between the fault and the test in step (2) is represented by a matrix D1, D1＝[dij]m×nWherein d isijThe element belongs to {0,1}, the value of 0 indicates that the ith fault cannot be detected by the jth test in any mode, and the value of 1 indicates that the ith fault can be detected by the jth test; the correlation of the test and the switch is represented by a matrix D2, D2＝[rij]n×kWherein r isijThe element belongs to {0,1}, the value of 0 represents that the correlation of the jth test and the fault is not influenced by the state of the ith switch, and the value of 1 represents that the correlation of the jth test and the fault is correlated with the state of the ith switch; where m represents the number of faults, n represents the number of tests, and k represents the number of switches.
Further, the bayesian network in the step (3) comprises a two-layer structure, wherein the first layer comprises fault and switch nodes, and the second layer comprises test nodes.
Further, the connection relationship between the nodes of the bayesian network in the step (3) corresponds to the matrix in the step (2), and the matrix D is aimed at1＝[dij]m×nIf d isijAdding a connection relation between the ith fault and the jth test, wherein the direction is fiPoint of direction tj(ii) a For matrix D2＝[rij]n×kIf r isijAdding a connection relation between the jth test and the ith switch, wherein the direction is siPoint of direction tj(ii) a Wherein the i-th fault is defined by fiDenotes the jth test with tjDenotes that the ith switch is used as siAnd (4) showing.
Further, the Bayesian network parameters of the test nodes in the step (4) are expressed by a binary tree and generated in an optimization mode.
Further, the optimization method comprises the following steps:
the method comprises the steps that I, for a certain test node, an initial candidate node set is represented by a set E ═ { F, S }, wherein F represents a fault node relevant to the initial candidate node set, and S represents a switch node relevant to the initial candidate node set;
II, selecting the switching node with the largest number of related fault nodes and not being 0 from the set E as the next tree node; removing the switch node from E for the switch closed branch; for the branch with the switch disconnected, removing the switch node and a fault node related to the switch from E, and updating E; if the set E has no switching node, turning to the step III;
III, selecting a fault node with the highest fault rate as a next tree node, removing the fault node from the E, and updating the E; storing a parameter 1 aiming at the branch with the fault, and stopping expanding the branch; if the fault does not have the branch, if the E does not contain other fault nodes, the parameter 0 is stored, otherwise, the step II is continuously selected, and the parameter expression tree is expanded;
and IV, the finally obtained binary tree is the parameter of the test node.
The invention has the beneficial effects that: a method for constructing a programmed reconfigurable system Bayesian network is provided, the Bayesian network parameter storage and retrieval are optimized, a new solution is provided for solving the problem of constructing the reconfigurable system Bayesian network, and the application of the Bayesian network in the field of fault diagnosis can be promoted.
Drawings
For ease of illustration, the invention is described in detail by the following detailed description, examples and figures.
FIG. 1 is a correlation model of a reconfigurable system in an embodiment of the invention;
FIG. 2 is a Bayesian network constructed by a reconfigurable system in an embodiment of the present invention;
FIG. 3 is a binary node parameter tree of Bayesian network t1 constructed in the embodiment;
FIG. 4 is a binary node parameter tree of Bayesian network t2 constructed in the embodiment.
Detailed Description
The invention comprises the following steps:
step one, modeling a fault-test correlation relation of the reconfigurable system.
Unlike a general single mode system, the reconfigurable system may be changed according to a change in environment due to a system configuration during use. This means that the relationship between faults and tests will also vary. Here, the characterization is performed by means of a multimode, multi-signal model. The model contains the following four elements: the device comprises a fault, a switch, a test point and a connecting line, wherein the connecting line is used for connecting the fault, the test and the switch. The meaning of the model is: if the connection path exists between the fault and the test, the fault and the test are related, otherwise, the fault and the test are not related. A switch can be arranged between the fault detection device and the test device, and the correlation relationship between the fault and the test can be influenced by opening and closing the switch, so that the system reconfiguration is achieved.
Step two, representing the correlation of the reconfigurable system
And analyzing the system based on the model in the step one. A set of two correlation matrices is obtained. Assume that the system has a total of m failure modes, n tests, and k switches. Matrix one is the correlation between fault and test, i.e. D1＝[dij]m×nWherein d isijThe element is {0,1}, and the value of 0 indicates that the jth test cannot detect the ith fault in any mode; a value of 1 indicates that the jth test can detect the ith fault. Matrix two is the correlation between the switch and the test, using D2＝[rij]n×kIs represented by the formula (I) in which rijThe element belongs to {0,1}, and the value of 0 represents that the correlation of the jth test and the fault is not influenced by the ith switch state; a value of 1 indicates a j-th test and fault correlationSex is related to the state of the ith switch.
Step three, determining the structure of the Bayesian network of the reconfigurable system
According to the definition of the Bayesian network, the structure of the Bayesian network comprises two parts of nodes and connection relations among the nodes. The Bayesian network of the reconfigurable system constructed by the invention consists of three types of nodes including two layers. The first layer is the fault and switch layer, represented by the fault node and switch node, respectively. The second layer is a test layer, represented by test nodes. The failure nodes correspond to the failure modes in the step two, and the number of the failure nodes is m. The switch nodes correspond to the switches in the step two, and the total number of the switch nodes is k. The test nodes correspond to the test in the step two, and the number of the test nodes is n.
And after the number and the type of the nodes are determined, adding connection relations among the nodes. The adding mode is carried out according to the two correlation matrixes in the second step. The method specifically comprises the following steps: for matrix D1＝[dij]m×nIf d isijAt the ith fault (using f) when 1iDenoted by t) and the jth test (with t)jRepresented) in a direction of fiPoint of direction tj(ii) a For matrix D2＝[rij]n×kIf r isijTest at jth (with t) 1jDenoted) and ith switch (by s)iRepresented) in a direction of siPoint of direction tj。
Step four, determining parameters of the Bayesian network of the reconfigurable system and representing by adopting a tree structure
(1) And setting normal and abnormal states aiming at the fault node, wherein the abnormal probability represents the fault rate.
(2) And setting two states of on and off aiming at the switch node, wherein the probability is respectively set to be 0.5.
(3) And for one test node, representing an initial candidate node set by using a set E ═ { F, S }, wherein F represents a fault node related to the initial candidate node set, and S represents a switch node related to the initial candidate node set. And selecting the switch node with the largest number of related fault nodes and not being 0 from the set E as the next tree node. Removing the switch node from E for the switch closed branch; e is updated for the branch with the switch open, removing the switch node and the fault node associated with the switch from E. And if no switching node exists, selecting the fault node with the highest fault rate as the next tree node, removing the fault node from the E, and updating the E. Storing a parameter 1 aiming at the branch with the fault, and stopping expanding the branch; and aiming at the fault, if the fault does not have the branch, if the E does not contain other fault nodes, storing the parameter 0, otherwise, continuously selecting the nodes and expanding the parameter expression tree until a stop condition is reached. And finally obtaining the binary tree which is the parameter of the test node.
Examples
Step one, as shown in fig. 1, a correlation model is established for a certain reconfigurable system. The system comprises three fault modes of f1, f2 and f3, two switches of s1 and s2 and two tests of t1 and t 2. Through the change of the switch state, the reconfiguration of the system can be realized.
And step two, characterizing the reconfigurable system.
In this system, m is 3, n is 2, and k is 2.
Step three, determining the structure of the Bayesian network of the reconfigurable system
The resulting bayesian network structure is shown in fig. 2, according to the bayesian network structure determination rule.
Step four, determining parameters of the Bayesian network of the reconfigurable system
(1) For the failed nodes (i.e., f1, f2, f3), each node has two states: failure and normal. The probability of failure, i.e., failure rate, assumes three failure rates of 0.1, 0.3, and 0.6, respectively.
(2) For the switch nodes (i.e., s1, s2), each node has two states: open and close. The occurrence probability of both was set to 0.5.
(3) For test node t1, the initial candidate node is E ═ { f1, f2, s1, s2 }. The number of failed nodes associated with s1 is 1 and the number of failed nodes associated with s2 is 2, for which s2 is selected as the next tree node. For the switch closing branch, updated E' ═ { f1, f2, s1 }; for the switch open branch, E ″ { f1, s1 }. Then, by analogy, a parameter expression tree of t1 is obtained as shown in FIG. 3.
For test node t2, the initial candidate node is E ═ { f3}, and its binary parameter tree is shown in fig. 4.
The foregoing shows that the basic principles and essential features of the invention, as well as features of the invention, are described. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are merely illustrative of the principles of the invention, but that various changes and modifications may be made without departing from the spirit and scope of the invention, which shall fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (4)
1. A method for constructing a Bayesian network of a reconfigurable system is characterized by comprising the following steps: the method comprises the following steps:
(1) establishing a model for the correlation of reconfigurable system faults and tests;
(2) representing the correlation between the reconfigurable system fault and the test; the correlation between the fault and the test is represented by a matrix D1, D1＝[dij]m×nWherein d isijThe element belongs to {0,1}, the value of 0 indicates that the ith fault cannot be detected by the jth test in any mode, and the value of 1 indicates that the ith fault can be detected by the jth test; the correlation of the test and the switch is represented by a matrix D2, D2＝[rij]n×kWherein r isijThe element belongs to {0,1}, the value of 0 represents that the correlation of the jth test and the fault is not influenced by the state of the ith switch, and the value of 1 represents that the correlation of the jth test and the fault is correlated with the state of the ith switch; wherein m represents the number of faults, n represents the number of tests, and k represents the number of switches;
(3) determining a Bayesian network structure of a reconfigurable system; the Bayesian network comprises two layers of structures, the first layer is composed of fault and switch nodes, the second layer is composed of test nodes, the connection relation between the Bayesian network nodes corresponds to the matrix in the step (2), and the matrix D is aimed at1＝[dij]m×nIf d isijAdding a connection relation between the ith fault and the jth test, wherein the direction is fiPoint of direction tj(ii) a For matrix D2＝[rij]n×kIf r isijAdding a connection relation between the jth test and the ith switch, wherein the direction is siPoint of direction tj(ii) a Wherein the i-th fault is defined by fiDenotes the jth test with tjDenotes that the ith switch is used as siRepresents;
(4) and determining Bayesian network parameters of a reconstruction system and representing by adopting a tree structure.
2. The method of claim 1, wherein the model in step (1) comprises four elements, namely, a fault, a switch, a test point, and a connection line for connecting the fault, the test point, and the switch; the working mode of the model is as follows: when a connection path exists between the fault and the test, the fault is related to the test point, otherwise, the fault is not related; a switch can be arranged between the fault and the test, and the correlation between the fault and the test is influenced by the opening and closing of the switch, so that the reconstruction of the system is realized.
3. The method of claim 1, wherein: and (4) expressing the Bayesian network parameters of the test nodes in the step (4) by using a binary tree and generating the parameters in an optimization mode.
4. A method according to claim 3, characterized in that the optimization method comprises the steps of:
the method comprises the steps that I, for a certain test node, an initial candidate node set is represented by a set E ═ { F, S }, wherein F represents a fault node relevant to the initial candidate node set, and S represents a switch node relevant to the initial candidate node set;
II, selecting the switching node with the largest number of related fault nodes and not being 0 from the set E as the next tree node; removing the switch node from E for the switch closed branch; for the branch with the switch disconnected, removing the switch node and a fault node related to the switch from E, and updating E; if the set E has no switching node, turning to the step III;
III, selecting a fault node with the highest fault rate as a next tree node, removing the fault node from the E, and updating the E; storing a parameter 1 aiming at the branch with the fault, and stopping expanding the branch; if the fault does not have the branch, if the E does not contain other fault nodes, the parameter 0 is stored, otherwise, the step II is continuously selected, and the parameter expression tree is expanded;
and IV, the finally obtained binary tree is the parameter of the test node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710168449.2A CN106951707B (en) | 2017-03-21 | 2017-03-21 | Method for constructing Bayesian network of reconfigurable system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710168449.2A CN106951707B (en) | 2017-03-21 | 2017-03-21 | Method for constructing Bayesian network of reconfigurable system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106951707A CN106951707A (en) | 2017-07-14 |
| CN106951707B true CN106951707B (en) | 2020-04-24 |
Family
ID=59472903
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710168449.2A Active CN106951707B (en) | 2017-03-21 | 2017-03-21 | Method for constructing Bayesian network of reconfigurable system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951707B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108020781B (en) * | 2017-12-19 | 2020-05-29 | 上海电机学院 | Breaker fault diagnosis method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101266592A (en) * | 2007-03-16 | 2008-09-17 | 松下电器产业株式会社 | Reconfigurable circuit, reconfigurable circuit system, and reconfigurable circuit setting method |
| CN101303584A (en) * | 2008-06-26 | 2008-11-12 | 上海交通大学 | Method for reconfiguring control module of reconfigurable digital controller |
| CN102530270A (en) * | 2012-02-09 | 2012-07-04 | 北京控制工程研究所 | Restructurable satellite propulsion system |
| CN106021036A (en) * | 2016-05-26 | 2016-10-12 | 工业和信息化部电子第五研究所 | Reconfigurable system fault analyzing method and device |
| US9652213B2 (en) * | 2014-10-23 | 2017-05-16 | National Instruments Corporation | Global optimization and verification of cyber-physical systems using floating point math functionality on a system with heterogeneous hardware components |
-
2017
- 2017-03-21 CN CN201710168449.2A patent/CN106951707B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101266592A (en) * | 2007-03-16 | 2008-09-17 | 松下电器产业株式会社 | Reconfigurable circuit, reconfigurable circuit system, and reconfigurable circuit setting method |
| CN101303584A (en) * | 2008-06-26 | 2008-11-12 | 上海交通大学 | Method for reconfiguring control module of reconfigurable digital controller |
| CN102530270A (en) * | 2012-02-09 | 2012-07-04 | 北京控制工程研究所 | Restructurable satellite propulsion system |
| US9652213B2 (en) * | 2014-10-23 | 2017-05-16 | National Instruments Corporation | Global optimization and verification of cyber-physical systems using floating point math functionality on a system with heterogeneous hardware components |
| CN106021036A (en) * | 2016-05-26 | 2016-10-12 | 工业和信息化部电子第五研究所 | Reconfigurable system fault analyzing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106951707A (en) | 2017-07-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Aydin et al. | Using LSTM networks to predict engine condition on large scale data processing framework | |
| CN107704392A (en) | The processing method and server of a kind of test case | |
| CN105117772B (en) | A kind of method for parameter estimation of multi-state System Reliability model | |
| CN105871636A (en) | Reconstruction method and system for unmanned-aerial-vehicle formation communication topology based on minimum arborescence | |
| JP7568211B2 (en) | Techniques for generating network simulation scenarios | |
| CN105530122A (en) | Network failure diagnosis method based on selective hidden Naive Bayesian classifier | |
| CN106650942A (en) | Fault diagnosis method based on electronic equipment testability model | |
| Cui et al. | An analytical model of electronic fault diagnosis on extension of the dependency theory | |
| CN110018390B (en) | Hierarchical fuzzy petri network fault diagnosis method based on comprehensive variable weight | |
| CN106951707B (en) | Method for constructing Bayesian network of reconfigurable system | |
| Yang et al. | Supervisory fault tolerant control with integrated fault detection and isolation: A switched system approach | |
| Nam et al. | Virtual machine failure prediction using log analysis | |
| Alweshah et al. | Evolution of software reliability growth models: a comparison of auto-regression and genetic programming models | |
| Schmidt et al. | Active diagnosis of deterministic I/O automata | |
| Riaz et al. | Challenges with providing reliability assurance for self-adaptive cyber-physical systems | |
| Roy et al. | Identification of the K-most Vulnerable Entities in a Smart Grid System | |
| Abdelhameed et al. | Neural network based design of fault-tolerant controllers for automated sequential manufacturing systems | |
| CN113484685B (en) | Power grid fault diagnosis method based on time sequence organization type P system | |
| Kogeda et al. | A probabilistic approach to faults prediction in cellular networks | |
| Mohagheghi et al. | Fuzzy cognitive maps for identifying fault activation patterns in automation systems | |
| WO2023157280A1 (en) | Searching device, searching method, and searching program | |
| Shrobe | Model-based diagnosis for information survivability | |
| Maghsoodi et al. | Predicting cascading failure with machine learning methods in the interdependent networks | |
| Qu et al. | Optimal design of fault-tolerant Petri net controllers | |
| Kroese et al. | Applications of the Cross-Entropy Method in Reliability. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |