CN102802182B - A kind of wireless sensor network fault diagnostic device and method - Google Patents
A kind of wireless sensor network fault diagnostic device and method Download PDFInfo
- Publication number
- CN102802182B CN102802182B CN201210282597.4A CN201210282597A CN102802182B CN 102802182 B CN102802182 B CN 102802182B CN 201210282597 A CN201210282597 A CN 201210282597A CN 102802182 B CN102802182 B CN 102802182B
- Authority
- CN
- China
- Prior art keywords
- node
- diagnostic
- diagnostic evidence
- evidence
- diagnosis
- 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
Abstract
The present invention discloses a kind of wireless sensor network fault diagnostic device and method, comprises diagnosis trigger element, merges tree construction unit, diagnostic evidence generation unit, diagnostic evidence integrated unit and failure diagnosis unit.For different fault types, automatic foundation merges tree accordingly, then carries out the fusion diagnosis of fault in local problem region.Compared with traditional wireless sensor network fault diagnosis method, the diagnosis expense that the invention solves the existence of traditional wireless sensor networks method for diagnosing faults is large, diagnosis state that is not real-time and local problem region obtains the problems such as inaccurate, achieve the local real-time diagnosis of wireless sensor network fault, and diagnosis speed is fast, diagnostic accuracy is high, is with good expansibility.
Description
Technical field
The present invention relates to wireless sensor network field, particularly relate to a kind of wireless sensor network fault diagnostic device and method.
Background technology
Along with the fast development of the communication technology, sensor technology and embedding assembly technology and increasingly mature, wireless sensor network is widely used in that environmental monitoring, coal mining accident detection, desperate situation navigation, magnitude of traffic flow monitoring etc. are many to be had in the scene of significant associations with national economy.Due to the restriction of current scientific and technological level and manufacturing process, the software restraint of sensor node (abbreviation node) usually ideally can not agree with and carry out work, generally has the feature of easily makeing mistakes.Simultaneously, transfer of data is carried out owing to adopting wireless signal between node, multipath (multi-path), the interference factor such as (interference) further increase the unsteadiness of wireless sensor network: the appearance of barrier can weaken wireless signal, cause instantaneous between node or permanently lose connection; And the communication simultaneously of multiple node can cause occurring competition to the use of channel, finally only has a node successfully can seize channel and carries out data transmission.
In order to strengthen the availability of wireless sensor network, promote the reliability of wireless sensor network, many research institutions all expand the research of wireless sensor network fault diagnostic techniques simultaneously, in order to the malfunctioning node in supervising network fault and fixer network.Traditional wireless sensor network fault diagnostic techniques can be divided into two classes.The first kind is Software correction technology.Typical method is the debugging acid being similar to GDB at the source code layer building of node procedure, code error correction is carried out by interfaces such as breakpoint execution, variable observation, heap stack addressings, wherein, GDB is the program debugging tool that GNU increases income under a powerful UNIX that tissue issues, and UNIX is a kind of title of widely used commercial operating systems.These class methods can the logic error of determining program, but can not fault in the wireless sensor networks such as identification communication link is obstructed, nodal function is disorderly.Equations of The Second Kind technology carries out profound data analysis by the relevant information of collecting in wireless sensor network, diagnosis wireless sensor network fault.These class methods can be good at the situation of makeing mistakes identifying wireless sensor network, but usually need deep domain knowledge.For the rule-based diagnostic method that the researcher of University of California in Los Angeles proposes, base station first can the information of the neighbor node and next node etc. of each node in active collection wireless sensor network, then with reference to the decision-tree model set up with this, analysis result, thus there is fault and fault rootstock in rapid positioned radio sensor network.But, the diagnosis expense that these class methods are brought is large, and the foundation of the statistical model such as decision tree depends critically upon the practical operating experiences of researcher to wireless sensor network and the degree of understanding to wireless sensor network fault, does not thus have stronger extensibility.In other words, one all likely causes these class methods to lose efficacy to the unfamiliar personnel of wireless sensor network or a network error do not occurred.
As the above analysis, traditional wireless sensor network fault diagnostic techniques depends on the domain knowledge of designer for wireless sensor network and the experience of actual motion to a great extent, makes current diagnostic tool have significant limitation.In fact, due to wireless sensor network from group interconnected, remote deployment, circumstance complication, the characteristic such as extensive, failure cause, impact etc. that management maintenance personnel are difficult to wireless sensor network inside may occur have comprehensive understanding, especially appear at fault in multi-point interaction process and mistake is difficult to detect more.Along with the development of wireless sensor network and the continuous increase of application type, the extensibility of traditional wireless sensor network fault diagnostic techniques is faced with stern challenge.
Summary of the invention
For above-mentioned technical problem, the object of the present invention is to provide a kind of wireless sensor network fault diagnostic device and method, the diagnosis expense state that is large and local problem region which solving the existence of traditional wireless sensor networks method for diagnosing faults obtains the problems such as inaccurate, achieve the local diagnosis of wireless sensor network fault, be with good expansibility.
For reaching this object, the present invention by the following technical solutions:
A kind of wireless sensor network fault diagnostic device, this diagnostic device is installed in each node of wireless sensor network, comprising:
Diagnosis trigger element, for analyzing the state information of this diagnostic device place node, and according to analysis result, judges whether to generate diagnosis process;
Merge tree construction unit, for when diagnosing trigger element to generate diagnosis process, the fault type corresponding according to this diagnosis process builds and merges tree;
Diagnostic evidence generation unit, for the state information of this diagnostic device place node and probability correlation being joined, generates probability assignment and the diagnostic evidence of described node preset failure type;
Diagnostic evidence integrated unit, for receiving the described diagnostic evidence merging all child nodes input of this diagnostic device place node in tree, and merges the diagnostic evidence of itself and this diagnostic device place node self;
Failure diagnosis unit, for the diagnostic evidence fusion results according to described diagnostic evidence integrated unit, judges whether fault corresponding to described diagnosis process exists.
Especially, described fusion tree construction unit comprises:
Diagnosis request signal transmitting element, for broadcast diagnostics request signal in wireless sensor network, when not diagnosing request signal to propagate in wireless sensor network, represents that merging tree structure completes; Wherein, described diagnosis request signal comprises the identify label number (ID) of the node of the diagnostic area corresponding with fault type and this diagnosis request signal of transmission;
Diagnosis request signal response unit, for receiving the diagnosis request signal of this diagnostic device place node of input, and parses diagnostic area from this diagnosis request signal, judges whether this node belongs to the fusion tree that will set up according to described diagnostic area; If the determination result is YES, then will send the father node of node as oneself of described diagnosis request signal, and extract its identify label number, and notice diagnosis request signal transmitting element continues broadcast diagnostics request signal.
Especially, described diagnostic evidence integrated unit comprises:
Diagnostic evidence receiving element, for receive this diagnostic device place node child list in the diagnostic evidence of all child node inputs;
Merge arithmetic element, merge with the diagnostic evidence of this diagnostic device place node self for the diagnostic evidence that diagnostic evidence receiving element is received;
Diagnostic evidence transmitting element, for when diagnostic evidence receiving element does not receive the diagnostic evidence of a certain child node in described child list, notify that this child node retransmits diagnostic evidence, and the diagnostic evidence after fusion arithmetic element being merged sends to the father node of this diagnostic device place node.
The invention also discloses a kind of wireless sensor network fault diagnosis method, each node of this wireless sensor network is all provided with above-mentioned wireless sensor network fault diagnostic device, comprises the steps:
A, diagnosis trigger element analyze the state information of its place node, and according to analysis result, judge whether to generate diagnosis process;
B, when the diagnosis trigger element of a certain node generates diagnosis process, the fusion tree construction unit of this node builds according to fault type corresponding to this diagnosis process and merges tree, and wherein, described node is the root node merging tree;
The state information of self and probability correlation are joined by its diagnostic evidence generation unit by the leaf node of C, fusion tree, generate probability assignment and the diagnostic evidence of preset failure type, and import described diagnostic evidence into father node;
The diagnostic evidence that described leaf node inputs by the diagnostic evidence integrated unit of D, described father node and the diagnostic evidence of self merge, and import the diagnostic evidence after merging into node in father node list;
Diagnostic evidence after the described fusion received and the diagnostic evidence of self merge by the diagnostic evidence integrated unit of the node in E, described father node list, and import the diagnostic evidence after fusion into node in the father node list of this node;
F, repeated execution of steps E, finally, the failure diagnosis unit of described root node, according to the diagnostic evidence fusion results of diagnostic evidence integrated unit, judges whether fault corresponding to described diagnosis process exists.
Especially, the fault type that the fusion tree construction unit of this node described in described step B is corresponding according to this diagnosis process builds fusion tree, specifically comprises:
Diagnosis request signal transmitting element broadcast diagnostics request signal in wireless sensor network of B1, root node; Wherein, described diagnosis request signal comprises the diagnostic area corresponding with fault type, the identify label number of root node and the regular set that is made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node;
After the diagnosis request signal response unit of B2, other node except root node receives described diagnosis request signal, from this diagnosis request signal, parse diagnostic area, and according to described diagnostic area, judge whether this node belongs to the fusion tree that will set up;
B3, when the judged result of step B2 is for being, node extracts the identity recognition number of root node and described regular set from diagnosis request signal, using the father node of root node as oneself, and the identity recognition number of self is inserted in described diagnosis request signal obtain new diagnosis request signal, by the described diagnosis request signal newly of diagnosis request signal transmitting element broadcast;
B4, repeated execution of steps B2 and B3, root node constructs the fusion tree corresponding with fault type.
Especially, described step C specifically comprises:
In C1, fusion tree, all child list are that empty node passes through diagnosis request signal transmitting element to neighbor node transmission leaf request signal, be confirmed whether the leaf node for merging tree, if the diagnosis request signal response unit sending the node of leaf request signal does not receive back-signalling, illustrate that this node is the leaf node merging tree; If the node sending leaf request signal receives back-signalling, illustrate that this node is not the leaf node merging tree, then this node will send the child node of node as oneself of described back-signalling, upgrade the child list of oneself;
The diagnostic evidence generation unit of the leaf node of C2, fusion tree utilizes Naive Bayes Classifier the state information of self and probability correlation to be joined, and calculates probability assignment and the diagnostic evidence of preset failure type; Computational process is as follows:
Wherein, R is default fault type R
0, R
1, R
2... R
nany one, R
0represent without exception, (F
1, F
2... F
n) be data parameter F
1, F
2... F
nset and the state information of node, P (F
1, F
2... F
n) be conversion coefficient, P (R) be the training stage estimate fault occur probability, P (F
i| R) (i gets 1,2...n) data parameter F of state information when fault occurs of estimating for the training stage
ithe probability existed, P (R|F
1, F
2... F
n) be the probability of happening of various types of faults corresponding to state information and diagnostic evidence, also referred to as basic trust partition function m (R); So, node N
kdiagnostic evidence be designated as m
k(R
j), be simply designated as m
k, and ∑
0≤j≤nm
k(R
j)=1;
The diagnostic evidence generation unit of C3, described leaf node utilizes the diagnostic evidence m obtained in step C2
kwith the described regular set S be made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node, calculate diagnostic evidence m according to formula (G1), (G2) and (G3)
kbasic trust degree β
k, and according to basic trust degree β
kby formula (G4) and (G5) to diagnostic evidence m
kbe weighted process, calculate diagnostic evidence m
kweighting diagnostic evidence m '
k(R
j), be simply designated as m '
k;
S(m
k,m
l)＝1-d(m
k,m
l) （G2）
Wherein, d (m
k, m
l) be diagnostic evidence m
kwith m
ldistance, m
lfor any one diagnostic evidence in described regular set S, M
k=[m
k(R
0), m
k(R
1) ... m
k(R
n)]
t, 0≤d (m
k, m
l)≤1; S (m
k, m
l) be diagnostic evidence m
kwith m
lsimilarity; P is the quantity of diagnostic evidence in regular set S;
The weighting diagnostic evidence m ' that the diagnostic evidence generation unit of C4, described leaf node will calculate in step C3
kthe father node of this leaf node is sent to by diagnostic evidence transmitting element.
Especially, described step C2 also comprises:
After the diagnostic evidence generation unit of the leaf node merging tree utilizes Naive Bayes Classifier to generate diagnostic evidence m (R), judge whether merge the quantity comprising node in tree reaches predetermined threshold value, if reach, then the diagnostic evidence generation unit of described leaf node performs step C3, if do not reach, then the diagnostic evidence generation unit of described leaf node does not perform step C3, direct execution step C4, crosses diagnostic evidence m (R) father node that diagnostic evidence transmitting element sends to this leaf node.
Especially, described step D specifically comprises:
The diagnostic evidence receiving element of father node described in D1, step C4 receives the weighting diagnostic evidence m ' of leaf node input
k;
The diagnostic evidence transmitting element of D2, described father node, when not receiving the weighting diagnostic evidence of a certain child node in child list, notifies that this child node retransmits weighting diagnostic evidence;
The diagnostic evidence generation unit of D3, described father node performs identical operation with the diagnostic evidence generation unit of leaf node: utilize the described regular set S be made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node the diagnostic evidence of self to be processed, obtain weighting diagnostic evidence m '
q;
The fusion arithmetic element of D3, described father node according to formula (G6) by weighting diagnostic evidence m '
kwith weighting diagnostic evidence m '
qmerge, obtain fusion diagnosis evidence m '
kq;
Wherein, the fault type R preset
0, R
1, R
2... R
nthe set formed is called identification framework, is designated as Θ, merges the identification framework of each node in tree identical, and the set of all subsets compositions of Θ is called and the power set of Θ is denoted as 2
Θ; X
i, Y, Z ∈ 2
Θ;
The diagnostic evidence transmitting element of D4, described father node will merge the fusion diagnosis evidence m ' of arithmetic element input
kqimport the node in father node list into.
Especially, the diagnostic evidence of self described in described step e is the weighting diagnostic evidence that diagnostic evidence generation unit generates.
Especially, in described step F, the diagnostic evidence that the fusion arithmetic element of the root node and its child node that merge tree is added in described regular set directly merges with the fusion diagnosis evidence received.
The present invention is directed to different fault types, automatic foundation merges tree accordingly, then carries out the fusion diagnosis of fault in local problem region.Compared with traditional wireless sensor network fault diagnosis method, the present invention only needs the information merging tree interior joint just can complete the real-time diagnosis of network failure, and effectively reduce diagnosis expense, and diagnose speed fast, diagnostic accuracy is high.
Accompanying drawing explanation
The wireless sensor network fault diagnostic device block diagram that Fig. 1 provides for the embodiment of the present invention;
The wireless sensor network fault diagnosis method flow chart that Fig. 2 provides for the embodiment of the present invention;
The particular flow sheet of the step S102 that Fig. 3 provides for the embodiment of the present invention;
The particular flow sheet of the step S103 that Fig. 4 provides for the embodiment of the present invention;
The particular flow sheet of the step S104 that Fig. 5 provides for the embodiment of the present invention.
Embodiment
For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with drawings and Examples, the invention will be further described.
Please refer to shown in Fig. 1, in the present embodiment, wireless sensor network fault diagnostic device comprises: diagnosis trigger element 101, fusion tree construction unit 102, diagnostic evidence generation unit 103, diagnostic evidence integrated unit 104 and failure diagnosis unit 105.Described fusion tree construction unit 102 comprises diagnosis request signal transmitting element 1021 and diagnosis request signal response unit 1022.Described diagnostic evidence generation unit 103 comprises diagnostic evidence transmitting element 1031, merges arithmetic element 1032 and diagnostic evidence receiving element 1033.
Described diagnosis trigger element 101 for analyzing the state information of this diagnostic device place node, and according to analysis result, judges whether to generate diagnosis process.
To diagnose the node failure fault in wireless sensor network.Such as, some node node1 are always in the neighbor list of its neighbor node node2, sometime, node node1 is suddenly deleted in the neighbor list of node node2, after then the diagnosis trigger element 101 of node node2 detects this state information of neighbor table, start to calculate this state information (namely neighbor table interior nodes A is deleted) lasting time, if the duration exceedes preset value such as 20 minutes, then by generation one diagnosis process, whether lost efficacy with diagnosis node node1, if the duration does not exceed preset value, then do not generate diagnosis process.Wherein, preset value can adjust as required flexibly.
Described fusion tree construction unit 102 is for when diagnosing trigger element 101 to generate diagnosis process, and the fault type corresponding according to this diagnosis process builds and merge tree.
The diagnosis process that diagnosis trigger element 101 generates is all corresponding with dissimilar fault phase, and merging tree construction unit 102 be that each fault type is all set up a corresponding fusion and set.Such as, we wonder whether a node lost efficacy, so we can utilize the diagnostic message of its neighbours to infer, the fusion tree of now setting up is made up of all neighbor nodes of this node, for another example we want to judge whether there is route loop in wireless sensor network, so we just should access all nodes transmitted on the path of certain associated packet, and tree is made up of all nodes on described path in the fusion of now setting up.When the diagnosis process diagnosing trigger element 101 to generate, the fault type foundation of merging tree construction unit 102 corresponding according to this diagnosis process merges tree accordingly.
Setting up in the process merging tree, described diagnosis request signal transmitting element 1021, for broadcast diagnostics request signal (DREQ) in wireless sensor network, when not diagnosing request signal to propagate in wireless sensor network, represents that merging tree structure completes; Wherein, described diagnosis request signal comprises the identify label number (ID) of the node of the diagnostic area corresponding with fault type and this diagnosis request signal of transmission.
Described diagnosis request signal response unit 1022 for receiving the diagnosis request signal of this diagnostic device place node of input, and parses diagnostic area from this diagnosis request signal, judges whether this node belongs to the fusion tree that will set up according to described diagnostic area; If the determination result is YES, then will send the father node of node as oneself of described diagnosis request signal, and extract its identify label number, and notice diagnosis request signal transmitting element 1021 continues broadcast diagnostics request signal.
Described diagnostic evidence generation unit 103, for the state information of this diagnostic device place node and probability correlation being joined, generates probability assignment and the diagnostic evidence of described node preset failure type.
The state information of this diagnostic device place node and probability correlation are joined by Naive Bayes Classifier by diagnostic evidence generation unit 103 in the present embodiment, through some conversion, generate probability assignment and the diagnostic evidence of node preset failure type.The quantity of described preset failure type can adjust according to actual needs.
The diagnostic evidence of itself and this diagnostic device place node self for receiving the described diagnostic evidence merging all child nodes input of this diagnostic device place node in tree, and merges by described diagnostic evidence integrated unit 104.
Wherein, described diagnostic evidence receiving element 1033 for receive this diagnostic device place node child list in the diagnostic evidence of all child node inputs.
Described fusion arithmetic element 1032 merges with the diagnostic evidence of this diagnostic device place node self for the diagnostic evidence received by diagnostic evidence receiving element 1033.
Described diagnostic evidence transmitting element 1031 is not for when diagnostic evidence receiving element 1033 receives the diagnostic evidence of a certain child node in described child list, notify that this child node retransmits diagnostic evidence, and the diagnostic evidence after fusion arithmetic element 1032 being merged sends to the father node of this diagnostic device place node.
Described failure diagnosis unit 105, for the diagnostic evidence fusion results according to described diagnostic evidence integrated unit 104, judges whether fault corresponding to described diagnosis process exists.
It should be noted that, in each node of wireless sensor network, above-mentioned wireless sensor network fault diagnostic device is all installed.
As shown in Figure 2, in the present embodiment, wireless sensor network fault diagnosis method comprises the steps:
Step S201, diagnosis trigger element 101 analyze the state information of its place node, and according to analysis result, judge whether to generate diagnosis process.
Step S202, when the diagnosis trigger element 101 of a certain node generates diagnosis process, the fault type that the fusion of this node tree construction unit 102 is corresponding according to this diagnosis process builds and merges tree, and wherein, described node is the root node merging tree.
The state information of self and probability correlation are joined by its diagnostic evidence generation unit 103 by the leaf node of step S203, fusion tree, generate probability assignment and the diagnostic evidence of preset failure type, and import described diagnostic evidence into father node.
The diagnostic evidence that described leaf node inputs by the diagnostic evidence integrated unit 104 of step S204, described father node and the diagnostic evidence of self merge, and import the diagnostic evidence after merging into node in father node list.
Diagnostic evidence after the described fusion received and the diagnostic evidence of self merge by the diagnostic evidence integrated unit 104 of the node in step S205, described father node list, and import the diagnostic evidence after fusion into node in the father node list of this node.
Step S206, repeated execution of steps S205, finally, the failure diagnosis unit 105 of described root node, according to the diagnostic evidence fusion results of diagnostic evidence integrated unit 104, judges whether fault corresponding to described diagnosis process exists.
As shown in Figure 3, the detailed process merging tree foundation is as follows:
Diagnosis request signal transmitting element 1021 broadcast diagnostics request signal in wireless sensor network of step S2021, root node; Wherein, described diagnosis request signal comprises the diagnostic area corresponding with fault type, the identify label number of root node and the regular set that is made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node.
The diagnostic area that dissimilar fault is corresponding different, such as, we wonder whether a node lost efficacy, and so we can utilize the diagnostic message of its neighbours to infer, diagnostic area is now exactly all neighbor nodes of this node.
After the diagnosis request signal response unit 1022 of step S2022, other node except root node receives described diagnosis request signal, diagnostic area is parsed from this diagnosis request signal, and according to described diagnostic area, judge whether this node belongs to the fusion tree that will set up.
Step S2023, when the judged result of step S2022 is for being, node extracts the identity recognition number of root node and described regular set from diagnosis request signal, using the father node of root node as oneself, and the identity recognition number of self is inserted in described diagnosis request signal obtain new diagnosis request signal, broadcast described diagnosis request signal newly by diagnosis request signal transmitting element 1021.It should be noted that, in order to ensure the normal foundation of merging tree, all nodes in diagnostic area receive only once diagnoses request signal, and when again there being diagnosis request signal to import into, this node will abandon automatically, and does not resolve it.
Step S2024, repeated execution of steps S2022 and step S2023, root node constructs the fusion tree corresponding with fault type.
As shown in Figure 4, the generation of diagnostic evidence and later stage conversion process as follows:
In step S2031, fusion tree, all child list are that empty node passes through diagnosis request signal transmitting element 1021 to neighbor node transmission leaf request signal (LQUE), be confirmed whether the leaf node for merging tree, if the diagnosis request signal response unit 1022 sending the node of leaf request signal does not receive back-signalling, illustrate that this node is the leaf node merging tree; If the node sending leaf request signal receives back-signalling, illustrate that this node is not the leaf node merging tree, then this node will send the child node of node as oneself of described back-signalling, upgrade the child list of oneself.
Child node list is that empty node is when receiving the back-signalling of neighbor node, illustrate in the father node list of this neighbor node and have described child list to be the identity recognition number of empty node, but, when setting up fusion tree, described neighbor node is caused not join in the child list of its father node due to the problem of the aspects such as data communication.
The diagnostic evidence generation unit 103 of the leaf node of step S2032, fusion tree utilizes Naive Bayes Classifier the state information of self and probability correlation to be joined, and calculates probability assignment and the diagnostic evidence of preset failure type; Computational process is as follows:
Wherein, R is default fault type R
0, R
1, R
2... R
nany one, R
0represent without exception, (F
1, F
2... F
n) be data parameter F
1, F
2... F
nset and the state information of node, P (F
1, F
2... F
n) be conversion coefficient, P (R) be the training stage estimate fault occur probability, P (F
i| R) (i gets 1,2...n) data parameter F of state information when fault occurs of estimating for the training stage
ithe probability existed, P (R|F
1, F
2... F
n) be the probability of happening of various types of faults corresponding to state information and diagnostic evidence, also referred to as basic trust partition function m (R); So, node N
kdiagnostic evidence be designated as m
k(R
j), be simply designated as m
k, and ∑
0≤j≤nm
k(R
j)=1.
Step S2033, after the diagnostic evidence generation unit 103 of leaf node merging tree utilizes Naive Bayes Classifier to generate diagnostic evidence m (R), judge to merge in setting the quantity comprising node and whether reach predetermined threshold value.
Step S2034, when the judged result of step S2033 is for being, the diagnostic evidence generation unit 103 of described leaf node utilizes the diagnostic evidence m obtained in step S2032
kwith the described regular set S be made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node, calculate diagnostic evidence m according to formula (G1), (G2) and (G3)
kbasic trust degree β
k, and according to basic trust degree β
kby formula (G4) and (G5) to diagnostic evidence m
kbe weighted process, calculate diagnostic evidence m
kweighting diagnostic evidence m '
k(R
j), be simply designated as m '
k; And by the diagnostic evidence generation unit 103 of described leaf node by weighting diagnostic evidence m '
kthe father node of this leaf node is sent to by diagnostic evidence transmitting element 1031;
S(m
k,m
l)＝1-d(m
k,m
l) （G2）
Wherein, d (m
k, m
l) be diagnostic evidence m
kwith m
ldistance, m
lfor any one diagnostic evidence in described regular set S, M
k=[m
k(R
0), m
k(R
1) ... m
k(R
n)]
t, 0≤d (m
k, m
l)≤1; S (m
k, m
l) be diagnostic evidence m
kwith m
lsimilarity; P is the quantity of diagnostic evidence in regular set S.
The principle of setting threshold: if merge the m of only design two diagnostic evidence in tree
iand m
jmerge, because s is (m
i, m
j)=s (m
j, m
i), even if so wherein any one diagnostic evidence be inaccurate, they also have identical basic trust degree.In order to address this problem, we set a threshold value, when only having the fusion tree quantity of interior joint and the quantity of diagnostic data to be greater than threshold value, Cai we are weighted process to diagnostic data.
Step S2035, when the judged result of step S2033 is no, diagnostic evidence generation unit 103 pairs of diagnostic evidence m (R) of described leaf node are not weighted process, are sent it to the father node of this leaf node by diagnostic evidence transmitting element 1031.
It should be noted that no matter whether diagnostic evidence m (R) is weighted process, the method that diagnostic evidence merges is identical.In the present embodiment, be weighted process with diagnostic evidence generation unit 103 pairs of diagnostic evidence m (R), obtain weighting diagnostic evidence m '
kfor example.
As shown in Figure 5, the detailed process that in tree, diagnostic evidence merges is merged as follows:
The diagnostic evidence receiving element 1033 of the father node of step S2041, leaf node receives the weighting diagnostic evidence m ' of leaf node input
k.
The diagnostic evidence transmitting element 1031 of step S2042, described father node, when not receiving the weighting diagnostic evidence of a certain child node in child list, notifies that this child node retransmits weighting diagnostic evidence.
In order to avoid diagnostic evidence is lost, each intermediate node (node except leaf node) merging tree needs to go " prompting " child node to transmit data by child node inquiry request signal (CQUE), and the information of losing so just can utilize retransmission mechanism to be retracted.
The fusion arithmetic element 1032 of step S2043, described father node according to formula (G6) by weighting diagnostic evidence m '
kwith weighting diagnostic evidence m '
qmerge, obtain fusion diagnosis evidence m '
kq;
Wherein, the fault type R preset
0, R
1, R
2... R
nthe set formed is called identification framework, is designated as Θ, merges the identification framework of each node in tree identical, and the set of all subsets compositions of Θ is called and the power set of Θ is denoted as 2
Θ; X
i, Y, Z ∈ 2
Θ.
The fusion diagnosis evidence m ' that the diagnostic evidence transmitting element 1031 of step S2044, described father node will merge arithmetic element 1032 and inputs
kqimport the node in father node list into.
In addition, when the fusion arithmetic element 1032 of the root node and its child node that merge tree carries out diagnostic evidence fusion,
Because the weights of the diagnostic evidence in regular set are 1, so the diagnostic evidence generation unit 103 of these nodes does not need the diagnostic evidence to it joins in regular set to be weighted process, only diagnostic evidence corresponding in regular set with these nodes for the fusion diagnosis evidence received directly need be carried out merging.
Below to using the principle of regular set to be briefly described in the present invention:
Formula (G1), (G2), (G3), (G4) and (G5) are for the present invention is for the improvement of D-S evidence theory.Suppose n diagnostic evidence m
1, m
2... m
nthe set formed is ASS, for wherein any one diagnostic evidence m
ithe weighting processing procedure of (1≤i≤n) is as follows: one, calculate m by formula (G1)
iwith the distance d (m of other diagnostic evidence in set A SS
i, m
l) (i ≠ l); Two, m is calculated by formula (G2)
iwith the similarity S (m of other diagnostic evidence in set A SS
i, m
l); Three, by formula β
i=∑
1≤l≤n, i ≠ ls (m
i, m
l), calculate m
iwith the similarity of other all diagnostic evidence in set A SS and; Four, m
iand in set A SS other all diagnostic evidence similarity and the similarity degree that reflects between diagnostic evidence, choose diagnostic evidence based on some diagnostic evidence according to this similarity degree, this basic diagnostic evidence forms regular set.But only, the method calculating of this selection standard collection expends too large, transfer of data burden is heavy, and build in the process of fusion tree, we cannot obtain whole diagnostic evidence, therefore cannot calculate to the diagnostic evidence of each node the similarity that it arrives the diagnostic evidence of other all nodes, therefore we need the diagnostic evidence selecting some nodes thought as regular set.
Generally we select the set of the diagnostic evidence of the root node merging tree and the diagnostic evidence formation of its child node as regular set.Just from level to level regular set can be dealt into each node merging tree like this to get on when contributing, therefore each node could utilize formula (G4), (G5) and (G6) calculate the basic trust degree of oneself diagnostic evidence, the weights of this diagnostic evidence are obtained, final acquisition weighting diagnostic evidence by basic trust degree.And the weights of the diagnostic evidence that our required standard is concentrated are 1.
We have on the test envelope of the wireless sensor network of 50 nodes at one and verify function of the present invention.In the accuracy rate of failure diagnosis, we manually add the mistake of three types in test envelope, are that node hardware damages, flow blocks and Routing Loop respectively.By the comparison with traditional wireless sensor networks fault diagnosis technology, the present invention can detect the node hardware damage of 92%, the flow blocking of 86% and the Routing Loop fault of 95%, average specific traditional wireless sensor networks method for diagnosing faults detects the mistake of about 5% more.Meanwhile, in the middle of the fault determining to make mistakes, on average only has the rate of false alarm of 8%.In failure diagnosis time, the present invention has only spent 95 milliseconds to damage for successfully diagnosing the node hardware of 80%, 101 milliseconds of Routing Loop faults for successfully diagnosing 80%, maximumly also only occupies 110 milliseconds.For the flow clogging diagnoses of more complicated, the present invention has on average spent 140 milliseconds.And all every 200 milliseconds of the present invention creates 13 signal bags lesser than traditional wireless sensor networks fault diagnosis technology, thus less impact is created on the transmission application of former network.
Technical scheme of the present invention is for different fault types, and automatic foundation merges tree accordingly, then carries out the fusion diagnosis of fault in local problem region, and not only diagnose speed fast, diagnostic accuracy is high, and is with good expansibility.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. a wireless sensor network fault diagnostic device, is characterized in that, this diagnostic device is installed in each node of wireless sensor network, comprising:
Diagnosis trigger element, for analyzing the state information of this diagnostic device place node, and according to analysis result, judges whether to generate diagnosis process;
Merge tree construction unit, for when diagnosing trigger element to generate diagnosis process, the fault type corresponding according to this diagnosis process builds and merges tree;
Diagnostic evidence generation unit, for the state information of this diagnostic device place node and probability correlation being joined, generates probability assignment and the diagnostic evidence of described node preset failure type;
Diagnostic evidence integrated unit, for receiving the described diagnostic evidence merging all child nodes input of this diagnostic device place node in tree, and merges the diagnostic evidence of itself and this diagnostic device place node self;
Failure diagnosis unit, for the diagnostic evidence fusion results according to described diagnostic evidence integrated unit, judges whether fault corresponding to described diagnosis process exists.
2. wireless sensor network fault diagnostic device according to claim 1, is characterized in that, described fusion tree construction unit comprises:
Diagnosis request signal transmitting element, for broadcast diagnostics request signal in wireless sensor network, when not diagnosing request signal to propagate in wireless sensor network, represents that merging tree structure completes; Wherein, described diagnosis request signal comprises the identify label number (ID) of the node of the diagnostic area corresponding with fault type and this diagnosis request signal of transmission;
Diagnosis request signal response unit, for receiving the diagnosis request signal of this diagnostic device place node of input, and parses diagnostic area from this diagnosis request signal, judges whether this node belongs to the fusion tree that will set up according to described diagnostic area; If the determination result is YES, then will send the father node of node as oneself of described diagnosis request signal, and extract its identify label number, and notice diagnosis request signal transmitting element continues broadcast diagnostics request signal.
3. wireless sensor network fault diagnostic device according to claim 2, it is characterized in that, described diagnostic evidence integrated unit comprises:
Diagnostic evidence receiving element, for receive this diagnostic device place node child list in the diagnostic evidence of all child node inputs;
Merge arithmetic element, merge with the diagnostic evidence of this diagnostic device place node self for the diagnostic evidence that diagnostic evidence receiving element is received;
Diagnostic evidence transmitting element, for when diagnostic evidence receiving element does not receive the diagnostic evidence of a certain child node in described child list, notify that this child node retransmits diagnostic evidence, and the diagnostic evidence after fusion arithmetic element being merged sends to the father node of this diagnostic device place node.
4. a wireless sensor network fault diagnosis method, is characterized in that, each node of this wireless sensor network is all provided with wireless sensor network fault diagnostic device as claimed in claim 3, comprises the steps:
A, diagnosis trigger element analyze the state information of its place node, and according to analysis result, judge whether to generate diagnosis process;
B, when the diagnosis trigger element of a certain node generates diagnosis process, the fusion tree construction unit of this node builds according to fault type corresponding to this diagnosis process and merges tree, and wherein, described node is the root node merging tree;
The state information of self and probability correlation are joined by its diagnostic evidence generation unit by the leaf node of C, fusion tree, generate probability assignment and the diagnostic evidence of preset failure type, and import described diagnostic evidence into father node;
The diagnostic evidence that described leaf node inputs by the diagnostic evidence integrated unit of D, described father node and the diagnostic evidence of self merge, and import the diagnostic evidence after merging into node in father node list;
Diagnostic evidence after the described fusion received and the diagnostic evidence of self merge by the diagnostic evidence integrated unit of the node in E, described father node list, and import the diagnostic evidence after fusion into node in the father node list of this node;
F, repeated execution of steps E, finally, the failure diagnosis unit of described root node, according to the diagnostic evidence fusion results of diagnostic evidence integrated unit, judges whether fault corresponding to described diagnosis process exists.
5. wireless sensor network fault diagnosis method according to claim 4, is characterized in that, the fault type that the fusion tree construction unit of this node described in described step B is corresponding according to this diagnosis process builds fusion tree, specifically comprises:
Diagnosis request signal transmitting element broadcast diagnostics request signal in wireless sensor network of B1, root node; Wherein, described diagnosis request signal comprises the diagnostic area corresponding with fault type, the identify label number of root node and the regular set that is made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node;
After the diagnosis request signal response unit of B2, other node except root node receives described diagnosis request signal, from this diagnosis request signal, parse diagnostic area, and according to described diagnostic area, judge whether this node belongs to the fusion tree that will set up;
B3, when the judged result of step B2 is for being, node extracts the identity recognition number of root node and described regular set from diagnosis request signal, using the father node of root node as oneself, and the identity recognition number of self is inserted in described diagnosis request signal obtain new diagnosis request signal, by the described diagnosis request signal newly of diagnosis request signal transmitting element broadcast;
B4, repeated execution of steps B2 and B3, root node constructs the fusion tree corresponding with fault type.
6. wireless sensor network fault diagnosis method according to claim 5, is characterized in that, described step C specifically comprises:
In C1, fusion tree, all child list are that empty node passes through diagnosis request signal transmitting element to neighbor node transmission leaf request signal, be confirmed whether the leaf node for merging tree, if the diagnosis request signal response unit sending the node of leaf request signal does not receive back-signalling, illustrate that this node is the leaf node merging tree; If the node sending leaf request signal receives back-signalling, illustrate that this node is not the leaf node merging tree, then this node will send the child node of node as oneself of described back-signalling, upgrade the child list of oneself;
The diagnostic evidence generation unit of the leaf node of C2, fusion tree utilizes Naive Bayes Classifier the state information of self and probability correlation to be joined, and calculates probability assignment and the diagnostic evidence of preset failure type; Computational process is as follows:
Wherein, R is default fault type R
0, R
1, R
2... R
nany one, R
0represent without exception, (F
1, F
2... F
n) be data parameter F
1, F
2... F
nset and the state information of node, P (F
1, F
2... F
n) be conversion coefficient, P (R) be the training stage estimate fault occur probability; P (F
i| R) the data parameter F of state information when fault occurs that estimates for the training stage
ithe probability existed, wherein, i gets 1,2...n; P (R|F
1, F
2... F
n) be the probability of happening of various types of faults corresponding to state information and diagnostic evidence, also referred to as basic trust partition function m (R); So, node N
kdiagnostic evidence be designated as m
k(R
j), be simply designated as m
k, and Σ
0≤j≤nm
k(R
j)=1;
The diagnostic evidence generation unit of C3, described leaf node utilizes the diagnostic evidence m obtained in step C2
kwith the described regular set S be made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node, calculate diagnostic evidence m according to formula (G1), (G2) and (G3)
kbasic trust degree β
k, and according to basic trust degree β
kby formula (G4) and (G5) to diagnostic evidence m
kbe weighted process, calculate diagnostic evidence m
kweighting diagnostic evidence m '
k(R
j), be simply designated as m '
k;
S(m
k,m
l)＝1-d(m
k,m
l) (G2)
Wherein, d (m
k, m
l) be diagnostic evidence m
kwith m
ldistance, m
lfor any one diagnostic evidence in described regular set S, M
k=[m
k(R
0), m
k(R
1) ... m
k(R
n)]
t, 0≤d (m
k, m
l)≤1; S (m
k, m
l) be diagnostic evidence m
kwith m
lsimilarity; P is the quantity of diagnostic evidence in regular set S;
The weighting diagnostic evidence m ' that the diagnostic evidence generation unit of C4, described leaf node will calculate in step C3
kthe father node of this leaf node is sent to by diagnostic evidence transmitting element.
7. wireless sensor network fault diagnosis method according to claim 6, is characterized in that, described step C2 also comprises:
After the diagnostic evidence generation unit of the leaf node merging tree utilizes Naive Bayes Classifier to generate diagnostic evidence m (R), judge whether merge the quantity comprising node in tree reaches predetermined threshold value, if reach, then the diagnostic evidence generation unit of described leaf node performs step C3, if do not reach, then the diagnostic evidence generation unit of described leaf node does not perform step C3, diagnostic evidence m (R) is sent to the father node of this leaf node by diagnostic evidence transmitting element.
8. wireless sensor network fault diagnosis method according to claim 7, is characterized in that, described step D specifically comprises:
The diagnostic evidence receiving element of father node described in D1, step C4 receives the weighting diagnostic evidence m ' of leaf node input
k;
The diagnostic evidence transmitting element of D2, described father node, when not receiving the weighting diagnostic evidence of a certain child node in child list, notifies that this child node retransmits weighting diagnostic evidence;
The diagnostic evidence generation unit of D3, described father node performs identical operation with the diagnostic evidence generation unit of leaf node: utilize the described regular set S be made up of the diagnostic evidence of the diagnostic evidence of root node and the child node of root node the diagnostic evidence of self to be processed, obtain weighting diagnostic evidence m '
q;
The fusion arithmetic element of D3, described father node according to formula (G6) by weighting diagnostic evidence m '
kwith weighting diagnostic evidence m '
qmerge, obtain fusion diagnosis evidence
Wherein, the fault type R preset
0, R
1, R
2... R
nthe set formed is called identification framework, is designated as Θ, merges the identification framework of each node in tree identical, and the set of all subsets compositions of Θ is called and the power set of Θ is denoted as 2
Θ; X
i, Y, Z ∈ 2
Θ;
The diagnostic evidence transmitting element of D4, described father node will merge the fusion diagnosis evidence of arithmetic element input
import the node in father node list into.
9. wireless sensor network fault diagnosis method according to claim 8, is characterized in that, the diagnostic evidence of self described in described step e is the weighting diagnostic evidence that diagnostic evidence generation unit generates.
10. wireless sensor network fault diagnosis method according to claim 9, it is characterized in that, in described step F, the diagnostic evidence that the fusion arithmetic element of the root node and its child node that merge tree is added in described regular set directly merges with the fusion diagnosis evidence received.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210282597.4A CN102802182B (en) | 2012-08-09 | 2012-08-09 | A kind of wireless sensor network fault diagnostic device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210282597.4A CN102802182B (en) | 2012-08-09 | 2012-08-09 | A kind of wireless sensor network fault diagnostic device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102802182A CN102802182A (en) | 2012-11-28 |
| CN102802182B true CN102802182B (en) | 2015-08-26 |
Family
ID=47201124
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210282597.4A Active CN102802182B (en) | 2012-08-09 | 2012-08-09 | A kind of wireless sensor network fault diagnostic device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102802182B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104661185B (en) * | 2013-11-21 | 2019-03-01 | 富士通株式会社 | A kind of fused data transmission method and multihop network |
| CN104865956B (en) * | 2015-03-27 | 2017-07-07 | 重庆大学 | Sensor fault diagnosis method in a kind of complication system based on Bayesian network |
| CN107690155A (en) * | 2016-08-05 | 2018-02-13 | 富士通株式会社 | Diagnostic device, method and the portable terminal device of malfunctioning node |
| CN107295536B (en) * | 2017-07-05 | 2020-07-07 | 苏州大学 | Parallel diagnosis test method |
| CN109246736B (en) * | 2018-08-31 | 2021-11-26 | 中建科技集团有限公司 | Sensor network monitoring system, monitor and fault analysis terminal |
| CN110070118A (en) * | 2019-04-10 | 2019-07-30 | 广东电网有限责任公司 | A kind of multi-space data fusion method |
| CN110572822B (en) * | 2019-08-29 | 2022-09-02 | 河海大学常州校区 | Trust updating method in underwater wireless sensor network based on decision tree |
| CN112801171A (en) * | 2021-01-25 | 2021-05-14 | 中国商用飞机有限责任公司北京民用飞机技术研究中心 | Sensor false alarm identification method and device, computer equipment and storage medium |
| CN112911625B (en) * | 2021-02-04 | 2022-06-03 | 重庆邮电大学 | Fault diagnosis method for deterministic time slot communication sensing node |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101814114A (en) * | 2010-01-07 | 2010-08-25 | 北京航空航天大学 | Method for performing fault diagnosis by using model conversion |
| CN102033789A (en) * | 2010-12-03 | 2011-04-27 | 北京理工大学 | Reliability analysis method for embedded safety-critical system |
| CN102170648A (en) * | 2011-01-28 | 2011-08-31 | 北京浩阳华夏科技有限公司 | Passive diagnosis method of wireless sensor network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7158958B2 (en) * | 2003-12-24 | 2007-01-02 | The Boeing Company | Automatic generation of baysian diagnostics from fault trees |
-
2012
- 2012-08-09 CN CN201210282597.4A patent/CN102802182B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101814114A (en) * | 2010-01-07 | 2010-08-25 | 北京航空航天大学 | Method for performing fault diagnosis by using model conversion |
| CN102033789A (en) * | 2010-12-03 | 2011-04-27 | 北京理工大学 | Reliability analysis method for embedded safety-critical system |
| CN102170648A (en) * | 2011-01-28 | 2011-08-31 | 北京浩阳华夏科技有限公司 | Passive diagnosis method of wireless sensor network |
Non-Patent Citations (1)
| Title |
|---|
| 宫义山.基于故障树和贝叶斯网络的故障诊断模型.《沈阳工业大学学报》.2009, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102802182A (en) | 2012-11-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102802182B (en) | A kind of wireless sensor network fault diagnostic device and method | |
| Sahoo | Faulty Node Detection in Wireless Sensor Networks Using Cluster | |
| US7113988B2 (en) | Proactive on-line diagnostics in a manageable network | |
| CN102170648A (en) | Passive diagnosis method of wireless sensor network | |
| CN102684902B (en) | Based on the network failure locating method of probe prediction | |
| CN103840967A (en) | Method for locating faults in power communication network | |
| CN101783749B (en) | Network fault positioning method and device | |
| CN102255764A (en) | Method and device for diagnosing transmission network failure | |
| CN106056269A (en) | NanoSat satellite house-keeping health management system based on Bayes network model | |
| CN101170447A (en) | Service failure diagnosis system based on active probe and its method | |
| Lu et al. | A new fault detection method for computer networks | |
| CN102883359B (en) | A kind of method of measurement of the key node for wireless sensor network, device and system | |
| CN106559166A (en) | For condition detection method and equipment in distributed processing system(DPS) based on fingerprint | |
| Babaie et al. | A new self-diagnosing approach based on petri nets and correlation graphs for fault management in wireless sensor networks | |
| CN111600805B (en) | Bayes-based power data network congestion link inference method | |
| CN102281103B (en) | Optical network multi-fault recovering method based on fuzzy set calculation | |
| CN102739527B (en) | Network packet loss rate detection method | |
| Mohapatra et al. | Artificial immune system based fault diagnosis in large wireless sensor network topology | |
| CN103347279B (en) | Based on transmission method and the system of risk assessment | |
| CN106130780A (en) | A kind of IP network Fault Locating Method based on static Bayesian model | |
| Chen et al. | An efficient solution to locate sparsely congested links by network tomography | |
| JP2003032253A (en) | In-advance countermeasure on-line diagnosis in manageable network | |
| Mahapatroa et al. | Transient fault tolerant wireless sensor networks | |
| Khilar et al. | Intermittent fault diagnosis in wireless sensor networks | |
| CN103096412B (en) | Data fusion method with time delay restriction in delay tolerant network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |