CN109861859A - The Agent system fault detection method of comprehensive judgement is tested based on frontier inspection - Google Patents
The Agent system fault detection method of comprehensive judgement is tested based on frontier inspection Download PDFInfo
- Publication number
- CN109861859A CN109861859A CN201910083345.0A CN201910083345A CN109861859A CN 109861859 A CN109861859 A CN 109861859A CN 201910083345 A CN201910083345 A CN 201910083345A CN 109861859 A CN109861859 A CN 109861859A
- Authority
- CN
- China
- Prior art keywords
- vertex
- inspection
- failure
- agent
- inconsistent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a kind of Agent system fault detection methods that comprehensive judgement is tested based on frontier inspection, using each Agent in Agent system as vertex, using the cooperation measurement relationship between Agent as side, constitute frontier inspection and test network.It is whether normal to detect the property measurement on two vertex associated with side based on the inspection of side attribute measurement using the functional relation of side attribute and Agent vertex attribute.Malfunction using inspection result polygon in network, in conjunction with the polygon inspection result conditional probability calculated based on single side test false alarm rate and leakage alarm rate, according to each Agent of maximum likelihood principle comprehensive judgement.This method be confirmed using the cooperation measurement between multiple Agent each Agent itself property measurement whether failure, be capable of detecting when only to be not easy the failure that detects by Agent itself measurement；Its maximum likelihood determination method is more more accurate than simple vote；It can guarantee that any vertex failure to quantity less than vertex minimum neighbours' number can detect completely, therefore side attribute measurement topological structure selection has clear foundation.
Description
Technical field
It is especially a kind of to pass through the comprehensive of polygon inspection the present invention relates to a kind of Agent system fault detection technique field
Close the method determined to detect Agent system property measurement failure.
Background technique
Agent system is that the unification to the various distributed systems for completing task by multi-agent synergy work is abstracted table
It states, has many different instances because the specific object of wherein Agent is different, such as unmanned plane bee colony, satellites formation, robot cluster
Equal movable bodies group system and wireless sensor network distributed detection system.With artificial intelligence, sensor, network
The fast development of the technologies such as communication, device miniaturization is had become a trend using a large amount of microminiature Agent collaborative work,
The many advantages such as performance boost, reliability increase, adaptability enhances and cost reduces can be brought.
Agent is the primary condition and form that Agent system cooperates to the external world or the measurement of itself.Work as Agent
Measurement break down when, will affect multi-agent synergy work performance.When breaking down such as the sensor of wireless sensor network
It will affect system service quality, it is possible that after the calamity such as collision damage when movable body cluster self poisoning measures failure
Fruit.Therefore the Agent measurement failure effectively in detection Agent system is to system service quality, functional reliability and robustness
It is particularly significant.
Currently, for Agent system fault detection, the method for having each Agent individually to detect faults itself also has utilization
Cooperation between multiple Agent is come the method that detects Agent failure.Microminiature movable body field, is widely used GNSS (global satellite
Navigation system) measurement movement body position.Determine for GNSS fault detection, current research or using single machine itself measurement from GNSS
Failure is detected in the mechanism of position, or is mutually confirmed with the other measurement means of single machine itself and detects the measurement event of movement body position
Barrier.In terms of multi-Agent Cooperation detects failure, some research and utilization movable body Agent detect the observation of outer scene
The positioning failure of Agent also has research to detect position or the posture survey of movable body Agent based on the relative measurement between Agent
Measure failure.Wireless sensor network fault detection field, some single sensor of research and utilization itself measurements and its time correlation
Property detects failure, also has the spatial coherence of the multiple sensor measurements of research and utilization to detect failure.Multi-Agent Cooperation failure
Detection method is substantially mutually to be confirmed using the measurement of different Agent and the opposite or cooperation measurement between them, therefore
Only the cooperative detection (i.e. unilateral detection) between a couple Agent is only it can be found that failure, but cannot identify the source of trouble.In benefit
It is identified in the source of trouble with the synthesis (utilizing polygon testing result comprehensive judgement) of multipair youngster Agent cooperation measurement, current
Research is all not account for the error probability in unilateral detection to polygon comprehensive using the simple method that the minority is subordinate to the majority votes
The quantitative effect determined is closed, influence of the cooperative detection network topology structure to fault detectability is not accounted for yet, it cannot be guaranteed that
Failure can be detected completely.
Summary of the invention
The technical problem to be solved by the present invention is to sentence in view of the shortcomings of the prior art, providing one kind and testing synthesis based on frontier inspection
Fixed Agent system fault detection method, the false alarm rate tested according to frontier inspection and leakage alarm rate, it is comprehensive based on maximum likelihood principle
Analysis identification source of trouble Agent is closed, and can determine that frontier inspection tests network topology structure condition according to expected failure Agent quantity, protect
Demonstrate,prove to failure can detection property completely.
In order to solve the above technical problems, the technical scheme adopted by the invention is that: it is a kind of that comprehensive judgement is tested based on frontier inspection
Agent system fault detection method, comprising the following steps:
1) setting frontier inspection tests the topological structure of network；
2) malfunction initial value is set；
3) single side test is carried out to each side for examining figure G=(V, E) respectively, obtains each frontier inspection and tests function of state value C (ek)；
Determine whether each side is normal；The vertex on side normal in inspection figure is determined as normally；It is to the failure side examined in figure, i.e., all full
Sufficient C (ekThe side e of)=0k=(vi, vj), if S (vi)=1 and S (vj)=- 1 then enables S (vj)=0；If S (vi)=- 1 and S (vj)
=1, then enable S (vi)=0；S(vi) it is vertex viThree value malfunction functions；K=1,2 ..., m；I=1,2 ..., n；J=1,
2 ..., n；V is the set for examining vertex in figure G；E is the set for examining side in figure G；viAnd vjIt is to examine in figure summit set conjunction
Two different vertex.
After step 3), also to inconsistent vertex cluster subgraph Gc=(Vc, Ec) in the malfunction on all vertex sentence again
Determine, wherein GcFor the connected subgraph for examining figure G；VcFor inconsistent vertex set,EcFor inconsistent line set,
Decision process includes:
4) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, count as the following formula respectively
Calculate examine figure G in VcIn the associated all consistent normal sides in all vertex the summer condition probability that occur of inspection resultWherein,NoteTo examine inconsistent vertex；ω is inconsistent vertex cluster subgraph GcIn number of vertices；Indicate the conditional probability that the inspection result on the associated all consistent normal sides in single vertex occurs；nrFor examine figure G in top
Point set VcIn r-th of vertexThe associated number for examining consistent normal side；PfaFor the false alarm rate of single side test；PmaFor
The leakage alarm rate of single side test；
5) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, count as the following formula respectively
Calculate examine figure G in VcIn all associated all consistent failure sides in vertex inspection result occur summer condition probabilityarFor examine figure G in vertex set VcIn r-th of vertexAssociated inspection
The number on consistent failure side；
6) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, count as the following formula respectively
Calculate EcIn all sides inspection result occur summer condition probabilityWherein,|Ec| indicate EcThe sum on middle side,Respectively indicate inconsistent side ek∈EcTwo
Vertex,Indicate the conditional probability that single inconsistent side inspection result occurs；
7) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, count as the following formula respectively
Calculate examine figure G in GcThe associated all sides of inconsistent vertex cluster inspection result occur summer condition probability P:
P(S1, S2..., Sω)=PN(S1, S2..., Sω)·PA(S1, S2..., Sω)·PF(S1, S2..., Sω)；
8) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, ask inspection result to go out
The maximum value of existing conditional probability function:Take conditional probability P (S1, S2..., Sω) maximum value pair
Each vertex failure state variable value answered is the final judgement to the inconsistent vertex cluster malfunction；
9) to other inconsistent vertex clusters in figure G are examined, step 4)~step 8) is repeated, it is all in figure G until examining
Inconsistent vertex cluster determines to finish again.
In step 1), the topological structure meets the following conditions: to the network with n vertex, when expected maximum failure
Number of vertex is nfaultWhen, design frontier inspection tests network, and neighbours' number on each vertex is made all to be not less than nfault+1；nfault≤n-2。
The specific implementation process of step 2) includes: to all vertex vsi, malfunction is initially taken as S (vi)=- 1 indicates
Not yet distinguish state；To all side ek, arbitrarily setting C (ek) initial value.
In step 2), single side test is carried out to each side in inspection figure using hypothesis testing method respectively.
Compared with prior art, the advantageous effect of present invention is that: the invention proposes one kind based on frontier inspection test it is comprehensive
The Agent system fault detection method determined is closed, using Agent as network vertex, the cooperation measurement relationship between Agent is made
For side, gives through side attribute inspection and detect and identify the method frame of network vertex attribute measurement failure；It can be based on
Frontier inspection tests error probability and network graph topological structure to calculate the conditional probability of polygon inspection result appearance and make vertex event
The maximum likelihood of barrier state determines, to determine network vertex failure state more more scientificly than simple vote energy；This method can be from
Guarantee in network topology structure to any no more than given quantity vertex simultaneous faults can detection property completely.
Detailed description of the invention
Fig. 1 is the inconsistent vertex cluster exemplary diagram examined in figure；
Fig. 2 is that frontier inspection tests network and wherein 6 Agent distribution schematic diagrams.
Specific embodiment
Consider the fault detection problem measured multiple Agent dynamic attributes.Certain dynamic attribute of each Agent is carried out
Real-time measurement, measurement are likely to occur failure.If existing has function with their attribute to be measured to any two Agent
Relationship and measurable variable, then the variable can be used to whether be normally carried out mutual confirmation to the property measurement of two Agent.
Using the Agent in Agent system as vertex, using the cooperation measurement relationship between two Agent as side, network is constituted, the present invention claims
To examine network or examining figure, by the integrated treatment of inspection and inspection result to each edge attribute value, to determine in network
Whether each vertex has occurred failure.
Examine figure that can be indicated with non-directed graph G=(V, E).Wherein V is vertex set, V={ v1, v2..., vn, vi(i=1,
2 ..., n) be G i-th of vertex, n is number of vertices；E is side collection, E={ e1, e2..., em, m is the number on side in E, ek(k
=1,2 ..., m) be G kth side, ekIt can be e with two vertex representationk=(vi, vj), wherein viAnd vjIt is two in V
Different vertex.
Real number space R is tieed up to l in vertex in definition inspection figurelA mapping p (v), illustrate a certain to be detected of vertex
Attribute, such as coordinate position；Define the side e in figurek=(vi, vj) arrive real number space R mapping f (ek), indicate the category to be detected on side
Property value, the attribute to be detected on it and two vertex on the side has functional relation, f (ek)=f (vi, vj)=f (p (vi), p
(vj)), such as when side attribute between its two vertex apart from when, have f (vi, vj)=| | p (vi)-p(vj)||.Opposite vertexes attribute
Measured value is denoted asWherein εiAnd εjTo measure random error.By vertex attribute measured value
The side attribute value of calculating is denoted asSide attribute measured value is denoted asWherein η is
Side attribute measures random error.Whether the method judgement for having hypothesis testing using certain is unilateral normal, that is, passes through vertex attribute
The side attribute value that measured value is calculatedWith side attribute measured valueWhether unanimously (k=1,2 ..., m), in the present invention
Referred to as single side test.And the false alarm rate of known single side test is Pfa, leakage alarm rate is Pma。
The technical problem to be solved by the present invention is in the false alarm rate P of given single side testfaWith leakage alarm rate Pma's
Under the conditions of, by polygon inspection result come each vertex v in comprehensive judgement inspection figureiThe property measurement of (i=1,2 ..., n)
Whether failure.
The technical scheme is that a kind of provided Agent system failure inspection for testing comprehensive judgement based on frontier inspection
Survey method is tested each single side test result in network and single side test false alarm rate and leakage alarm rate information using frontier inspection, is based on
Maximum likelihood principle, each network vertex, that is, Agent of comprehensive judgement whether failure.Specifically, this method includes the following steps:
(1) setting frontier inspection tests the topological structure of network.For guarantee to network institute faulty node can detection property completely, it is right
Network with n vertex, when expected maximum failure number of vertex is nfault(nfault≤ n-2) when, design frontier inspection tests network, makes
Neighbours' number on each vertex is all not less than nfault+ 1, then it can guarantee and n be not more than to failure number of vertexfaultArbitrary Fault situation
It can detect completely.
(2) fault detection initializes.
Define the single side test function of state C (e of two-valuek) (k=1,2 ..., m), if side ekIt is verified as normally, then C
(ek)=1, otherwise C (ek)=0.If ek=(vi, vj), for convenience by C (ek) equivalently it is expressed as C (vi, vj), or write a Chinese character in simplified form
For Cij, and C (ek)=C (vi, vj)=Cij=C (vj, vi)=Cji。
Define the three value malfunction function S (v on vertexi) (i=1,2 ..., n), if it is determined that vertex viTo be normal, then S
(vi)=1；If it is determined that vertex viFor failure, then S (vi)=0；If vertex viWhether failure is unknown, then enables S (vi)=- 1.
When algorithm is initial, to all vertex vsi(i=1,2 ..., n), malfunction is initially taken as S (vi)=- 1 indicates
Not yet distinguish state.To all side ek(k=1,2 ..., m) can arbitrarily set C (ek) initial value.
(3) each side independence test in network.Using existing hypothesis testing method, each side in inspection figure is carried out respectively
Single side test determines whether each side is normal, obtains each frontier inspection and tests function of state value C (ek) (k=1,2 ..., m).
(4) determine the state on normal side vertex.The vertex on normal sides all in inspection figure is determined as normally.Examining
All C (e are found in figurekThe side of)=1, for ek=(vi, vj), enable S (vi)=S (vj)=1.
(5) determine the state on failure side vertex.It is to the every failure side examined in figure, i.e., all to meet C (ekThe side of)=0
ek=(vi, vj), if S (vi)=1 and S (vj)=- 1 then enables S (vj)=0；If S (vi)=- 1 and S (vj)=1 then enables S (vi)=
0。
(6) vertex and side determine inconsistent amendment.
Since single side test is there are false alarm and police is failed to report, judged according to different matched edges certain vertex whether failure can
It can will appear inconsistent.If the inspection result for examining certain side in figure is failure, and two vertex associated with it all has determined that and is positive
Often, then this is when being referred to as to examine inconsistent, referred to as inconsistent side, its associated two vertex be the inconsistent top of inspection
Point, referred to as inconsistent vertex.In algorithm, vertex and side determine inconsistent amendment, be exactly to the associated top in inconsistent side
The comprehensive judgement again of point failure state.
For the inconsistent vertex of multiple inspections being connected in inspection figure with inconsistent side, the present invention is known as them and forms one
A inconsistent vertex cluster.Fig. 1 gives an inconsistent vertex cluster example, and circle represents vertex in figure, the line generation between circle
Table side, the number in circle are the judgement of current opposite vertexes malfunction, and the digital representative edge marked beside each edge examines knot
Fruit.V in Fig. 1i、vjAnd vkThree vertex are connected by two inconsistent sides, and vi、vjAnd vkWith examine in figure between other vertex
All without inconsistent side, therefore vi、vjAnd vkConstitute an inconsistent vertex cluster.Pay attention in figure with vi、vjAnd vkAdjacent its
Its vertex is possible to and vi、vjAnd vkIn two or three simultaneously it is adjacent, a relationship is not shown in figure, because it is not influenced
Method and result.
For examining certain inconsistent vertex cluster in figure G=(V, E), its all vertex set is denoted asAndTo examine inconsistent vertex.It will be between each vertex of the inconsistent vertex cluster
The set on all inconsistent sides be denoted as Ec, meet:It is rightekTo examine inconsistent side, and there is vi
∈Vc, vj∈Vc；It is rightIfTo examine inconsistent side, then there must be v ∈ VcWithThen, by not
Consistent vertex set VcWith inconsistent side collection EcConstitute a connected subgraph G for examining figure Gc=(Vc, Ec), referred to as inconsistent vertex
Cluster subgraph.For vertex set VcIn r-th of vertexAssuming that associated with it in inspection figure G have arIt is a
Consistent failure side is examined, there is nrIt is a to examine consistent normal side, the inconsistent number of edges w of inspection associated with itrEqual to it in Gc
In neighbours' number
To examine figure in inconsistent vertex cluster, wherein the malfunction of institute's faulty node determine again must integrally into
Row.To inconsistent vertex cluster subgraph GcIn the malfunction on all vertex determine that specific step is as follows again:
(6.1) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, press respectively
Formula calculate examine figure G in VcIn the associated all consistent normal sides in all vertex the summer condition probability Ps that occur of inspection resultN:
WhereinIndicate the conditional probability that the inspection result on the associated all consistent normal sides in single vertex occurs.
(6.2) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, press respectively
Formula calculate examine figure G in VcIn all associated all consistent failure sides in vertex inspection result occur summer condition probability PA:
(6.3) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, press respectively
Formula calculates EcIn all sides inspection result occur summer condition probability PF:
Wherein | Ec| indicate EcThe sum on middle side,Respectively indicate inconsistent side ek∈EcTwo vertex,Indicate single
The conditional probability that a inconsistent side inspection result occurs.
(6.4) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, press respectively
Formula calculate examine figure G in GcThe associated all sides of inconsistent vertex cluster inspection result occur summer condition probability P:
P(S1, S2..., Sω)=PN(S1, S2..., Sω)·PA(S1, S2..., Sω)·PF(S1, S2..., Sω) (4)
(6.5) to VcIn each vertex variables associated with failure S1, S2..., SωAll different valued combinations, ask inspection to tie
The maximum value for the conditional probability function that fruit occurs:
Then according to maximum likelihood principle, conditional probability P (S is taken1, S2..., Sω) the corresponding each vertex failure shape of maximum value
State variable value is the final judgement to the inconsistent vertex cluster malfunction.
(6.6) to other inconsistent vertex clusters in figure G are examined, the decision process again of (6.1)-(6.5), Zhi Daojian are repeated
All inconsistent vertex clusters tested in figure G determine to finish again.
Method and step of the invention leaves it at that.
The fault detection effect of the method for the present invention is verified with Agent system positioning failure test problems below.
By taking six Agent networks of regular hexagon distribution as an example, the side length of the regular hexagon where each Agent distribution is
1000m, as shown in Figure 2.It is based on to examine each Agent self-position to measure failure using the method for mutual distance measurement between Agent
Ranging frontier inspection tests the malfunction that Agent in network is determined with comprehensive descision.If the position measurement of each Agent itself uses satellite
The modes such as laser ranging, radio distance-measuring, range accuracy ratio can be used in navigation positioning system, the distance between each Agent measurement
Position measurement is much higher, therefore ignores range error.The network vertex failure tested based on frontier inspection is verified by Monte-Carlo Simulation
Detection algorithm.With fault detection accuracy CDR (Correct Detection Rate) evaluation algorithms performance of statistics, CDR=
NC/NS, wherein NCRepresent the simulation times that can be appropriately determined all vertex failure states, NSRepresent total time of Monte-Carlo Simulation
Number.Emulation carries out fault detection based on the random sampling that each position Agent measures every time.Fault mode is mean value failure,
Mean bias size δ takes different given values respectively, and the direction of mean bias vector is uniformly distributed sampling by random.If each
The normal position measurement error standard deviation of Agent is σ0=5m, single side test method, which uses, to be had based on non-central chi square distribution
Unilateral Distance Test Method, single side test confidence level takes α=0.02, and then unilateral false alarm rate is Pfa=0.02, the distance
The unilateral leakage alarm rate of the method for inspection has passed through emulation and has obtained, and is listed in table 1, and wherein d indicates the actual range between Agent.
1 single side test of table leaks alarm rate statistics
No. 6 Agent are respectively set to break down, No. 4 and No. 6 two Agent break down and No. 2, No. 4 and 6 simultaneously
Numbers three Agent break down simultaneously, and totally three failure scenarios, Monte-Carlo Simulation result are listed in table 2.Each example covers special
Caro simulation times are 10000 times.
The fault detection accuracy of each failure scenario example of table 2
According to the result of table 2 as can be seen that when unilateral detection leakage alarm rate is lower, failure judgment method energy of the invention
Enough realize high fault detection accuracy.The inconsistent comprehensive modification of the inspection based on maximum likelihood principle that the present invention provides is to calculation
Method detection accuracy is obviously improved effect, and when unilateral leakage alarm rate is higher, amendment bring accuracy promotion is more aobvious
It writes.This is because the probability that the inconsistent situation of polygon inspection occurs is higher when unilateral leakage alarm rate is higher.
Claims (5)
1. a kind of Agent system fault detection method for testing comprehensive judgement based on frontier inspection, which is characterized in that including following step
It is rapid:
1) setting frontier inspection tests the topological structure of network；
2) malfunction initial value is set；
3) single side test is carried out to each side for examining figure G=(V, E) respectively, obtains each frontier inspection and tests function of state value C (ek)；Determine
Whether each side is normal；The vertex on side normal in inspection figure is determined as normally；It is to the failure side examined in figure, i.e., all to meet C
(ekThe side e of)=0k=(vi,vj), if S (vi)=1 and S (vj)=- 1 then enables S (vj)=0；If S (vi)=- 1 and S (vj)=1,
Then enable S (vi)=0；S(vi) it is vertex viThree value malfunction functions；K=1,2, L, m；I=1,2, L, n；J=1,2, L, n；
V is the set for examining vertex in figure G；E is the set for examining side in figure G；viAnd vjIt is two differences examined in figure summit set conjunction
Vertex.
2. the Agent system fault detection method according to claim 1 for being tested comprehensive judgement based on frontier inspection, feature are existed
In after step 3), also to inconsistent vertex cluster subgraph Gc=(Vc,Ec) in the malfunction on all vertex determine again, wherein
GcFor the connected subgraph for examining figure G；VcFor inconsistent vertex set,EcFor inconsistent line set,Determined
Journey includes:
4) to VcIn each vertex variables associated with failure S1,S2,L,SωAll different valued combinations, inspection is calculated as follows respectively
Test in figure G with VcIn the associated all consistent normal sides in all vertex the summer condition probability Ps that occur of inspection resultN:Wherein,Note For
Examine inconsistent vertex；ω is inconsistent vertex cluster subgraph GcIn number of vertices；Indicate single vertex associated all one
The conditional probability for causing the inspection result on normal side to occur；nrFor examine figure G in vertex set VcIn r-th of vertexIt is associated
Examine the number on consistent normal side；PfaFor the false alarm rate of single side test；PmaFor the leakage alarm rate of single side test；
5) to VcIn each vertex variables associated with failure S1,S2,L,SωAll different valued combinations, inspection is calculated as follows respectively
Test in figure G with VcIn all associated all consistent failure sides in vertex inspection result occur summer condition probability PA:arFor examine figure G in vertex set VcIn r-th of vertexIt is associated to examine unanimously
Failure side number；
6) to VcIn each vertex variables associated with failure S1,S2,L,SωAll different valued combinations, E is calculated as follows respectivelycIn
The summer condition probability P that the inspection result on all sides occursF:Wherein,|Ec| indicate EcThe sum on middle side,Respectively indicate inconsistent side ek∈EcTwo
Vertex,Indicate the conditional probability that single inconsistent side inspection result occurs；
7) to VcIn each vertex variables associated with failure S1,S2,L,SωAll different valued combinations, inspection is calculated as follows respectively
Test in figure G with GcThe associated all sides of inconsistent vertex cluster inspection result occur summer condition probability P:
P(S1,S2,L,Sω)=PN(S1,S2,L,Sω)·PA(S1,S2,L,Sω)·PF(S1,S2,L,Sω)；
8) to VcIn each vertex variables associated with failure S1,S2,L,SωAll different valued combinations, ask inspection result to occur
The maximum value of conditional probability function:Take conditional probability P (S1,S2,L,Sω) maximum value it is corresponding each
Vertex failure state variable value is the final judgement to the inconsistent vertex cluster malfunction；
9) to other inconsistent vertex clusters in figure G are examined, step 4)~step 8) is repeated, it is all different in figure G until examining
Vertex cluster is caused to determine to finish again.
3. the Agent system fault detection method according to claim 1 for being tested comprehensive judgement based on frontier inspection, feature are existed
In in step 1), the topological structure meets the following conditions: to the network with n vertex, when expected maximum failure number of vertex
For nfaultWhen, design frontier inspection tests network, and neighbours' number on each vertex is made all to be not less than nfault+1；nfault≤n-2。
4. the Agent system fault detection method according to claim 1 for being tested comprehensive judgement based on frontier inspection, feature are existed
In the specific implementation process of step 2) includes: to all vertex vsi, malfunction is initially taken as S (vi)=- 1 indicates not yet
Distinguish state；To all side ek, arbitrarily setting C (ek) initial value.
5. the Agent system fault detection method according to claim 1 for being tested comprehensive judgement based on frontier inspection, feature are existed
In carrying out single side test respectively to each side in inspection figure using hypothesis testing method in step 2).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910083345.0A CN109861859B (en) | 2019-01-28 | 2019-01-28 | Multi-Agent system fault detection method based on side inspection comprehensive judgment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910083345.0A CN109861859B (en) | 2019-01-28 | 2019-01-28 | Multi-Agent system fault detection method based on side inspection comprehensive judgment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109861859A true CN109861859A (en) | 2019-06-07 |
| CN109861859B CN109861859B (en) | 2021-08-27 |
Family
ID=66896673
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910083345.0A Active CN109861859B (en) | 2019-01-28 | 2019-01-28 | Multi-Agent system fault detection method based on side inspection comprehensive judgment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109861859B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111562794A (en) * | 2020-04-08 | 2020-08-21 | 中南大学 | Spacecraft attitude control method for actuator fault and input quantization |
| CN115903885A (en) * | 2022-10-26 | 2023-04-04 | 中国人民解放军陆军炮兵防空兵学院 | Unmanned aerial vehicle flight control method based on task traction bee colony Agent model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130188762A1 (en) * | 2012-01-24 | 2013-07-25 | Texas Instruments Incorporated | Clock Data Recovery With Out-Of-Lock Detection |
| CN108062567A (en) * | 2017-12-13 | 2018-05-22 | 中国人民解放军陆军工程大学 | A kind of Evidence, module and multi-Agent Diagnostic System |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| CN108709426A (en) * | 2018-05-31 | 2018-10-26 | 中冶华天工程技术有限公司 | It is leaked out on-line fault diagnosis method based on the bilateral detection method sintering machine of spectrum signature |
| CN109116188A (en) * | 2018-09-05 | 2019-01-01 | 广东电网有限责任公司 | A kind of electric power system fault deduction method, device, equipment and readable storage medium storing program for executing |
-
2019
- 2019-01-28 CN CN201910083345.0A patent/CN109861859B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130188762A1 (en) * | 2012-01-24 | 2013-07-25 | Texas Instruments Incorporated | Clock Data Recovery With Out-Of-Lock Detection |
| CN108320040A (en) * | 2017-01-17 | 2018-07-24 | 国网重庆市电力公司 | Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm |
| CN108062567A (en) * | 2017-12-13 | 2018-05-22 | 中国人民解放军陆军工程大学 | A kind of Evidence, module and multi-Agent Diagnostic System |
| CN108709426A (en) * | 2018-05-31 | 2018-10-26 | 中冶华天工程技术有限公司 | It is leaked out on-line fault diagnosis method based on the bilateral detection method sintering machine of spectrum signature |
| CN109116188A (en) * | 2018-09-05 | 2019-01-01 | 广东电网有限责任公司 | A kind of electric power system fault deduction method, device, equipment and readable storage medium storing program for executing |
Non-Patent Citations (6)
| Title |
|---|
| C.SPINU等: ""A multi-agent approach to edge detection as a distributed optimization"", 《IEEE PROCEEDINGS OF 13TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION》 * |
| LIGUO QIN等: ""A New Local-Model-Based Distributed Fault Diagnosis Scheme for Multi-Agent Systems with Actuator Faults"", 《ELSEVIER IFAC-PAPERS ONLINE》 * |
| 刘敏华: ""基于SDG模型的故障诊断及应用研究"", 《中国优秀博硕士学位论文全文数据库 (博士)》 * |
| 宋其江: ""基于有向图模型的故障诊断方法研究及其在航天中的应用"", 《中国博士学位论文全文数据库》 * |
| 张文波: ""基于多代理协作的卫星网络故障检测算法"", 《兵工学报》 * |
| 李千目: ""基于多代理的网络故障检测新方法"", 《计算机工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111562794A (en) * | 2020-04-08 | 2020-08-21 | 中南大学 | Spacecraft attitude control method for actuator fault and input quantization |
| CN115903885A (en) * | 2022-10-26 | 2023-04-04 | 中国人民解放军陆军炮兵防空兵学院 | Unmanned aerial vehicle flight control method based on task traction bee colony Agent model |
| CN115903885B (en) * | 2022-10-26 | 2023-09-29 | 中国人民解放军陆军炮兵防空兵学院 | Unmanned aerial vehicle flight control method of swarm Agent model based on task traction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109861859B (en) | 2021-08-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106323635B (en) | A kind of rolling bearing fault on-line checking and state evaluating method | |
| CN105425123B (en) | A kind of ultraviolet imagery and the method and system of infrared imaging cooperation detection electrical equipment fault | |
| CN106646526B (en) | A kind of receiver-autonomous integrity detection method that can detect identification various faults simultaneously | |
| CN108682140B (en) | Enhanced anomaly detection method based on compressed sensing and autoregressive model | |
| CN104125112B (en) | Physical-information fuzzy inference based smart power grid attack detection method | |
| CN109861859A (en) | The Agent system fault detection method of comprehensive judgement is tested based on frontier inspection | |
| CN105512011B (en) | A kind of electronics testability modeling appraisal procedure | |
| CN108199891B (en) | Cps network attack identification method based on artificial neural network multi-angle comprehensive decision | |
| CN104199051B (en) | Method for detecting and identifying satellite navigation RAIM (Receiver Autonomous Integrity Monitoring) multi-satellite faults | |
| CN103810382B (en) | Method for choosing two-level data fusion strategy of airborne distributed multi-sensor | |
| CN103458504B (en) | A kind of sensor node positioning method based on sequence and device | |
| CN106198020A (en) | Wind turbines bearing failure diagnosis method based on subspace and fuzzy C-means clustering | |
| CN106052962A (en) | Gas source positioning technology in wireless sensor network without wind speed and wind direction measuring | |
| CN108508458B (en) | Unmanned aerial vehicle GPS positioning fault detection and reconstruction method based on inter-aircraft ranging | |
| CN107966648B (en) | A kind of embedded failure diagnosis method based on correlation matrix | |
| US20230417372A1 (en) | Methods and internet of things (iot) systems for managing smart gas safety hazard items based on geographic information systems | |
| CN104504247B (en) | A kind of RAIM methods of the double satellite failures of GPS | |
| CN104537581B (en) | The voltage sag source tuning on-line method matched using fuzzy similarity | |
| CN104331602A (en) | GNSS (global navigation satellite system) completeness detection rate estimation method based on Markov chain | |
| CN111536437A (en) | Wisdom water affairs terminal that multisensor fuses | |
| CN111027721A (en) | System fault positioning method | |
| Zhang et al. | Detection and identification of bad data based on neural network and k-means clustering | |
| CN116234005A (en) | Indoor positioning method for multiple buildings | |
| Saha et al. | Distributed fault diagnosis in wireless sensor networks | |
| JP4868471B2 (en) | Real-time earthquake risk prediction method |
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 |