CN105933932A - Real-time fault diagnosis method and method for wireless sensor network in complex environment - Google Patents

Abstract

The invention discloses a real-time fault diagnosis method and a real-time fault diagnosis method for a wireless sensor network in a complex environment. The diagnosis method comprises the steps of S1, using an improved k-medoids clustering algorithm to cluster sensor nodes in a predetermined area; S2, according to data collected by the sensor node to be diagnosed and neighbor nodes thereof, computing a Vote value of the sensor node to be diagnosed, and determining the health status of the sensor node to be determined according to the Vote value; and S3, finding a sensor body of the fault node according to the sensor network, and carrying out maintenance. According to the method, via the fault diagnosis process with three stages, after the sensor network nodes in the complex environment is subjected to the stages of clustering, diagnosing and repairing, the network maintenance can be timely and efficiently completed.

Description

The real-time fault diagnosis method of wireless sensor network and system under complex environment

Technical field

The present invention relates to wireless sensor network (WSN) technical field, particularly relate to wireless biography under a kind of complex environment The real-time fault diagnosis method of sensor network and system.

Background technology

In the epoch that current information technology is developed by leaps and bounds, the network with Internet as representative brings to the life of people Huge change.Technology of Internet of things was born from 1999, within 2005, spreads to 2009 enter develop rapidly the stage, has become as Ingredient important in current information-based industry, wherein wireless sensor network technology is the guardian technique of Internet of Things. Wireless sensor network be jumped by one by the limited sensor node of a large amount of, communication and storage capacity or multi-hop communication mode from The one organized the formation of is without static infrastructure network, for perception, collection and the various information of the object of transmission monitoring in real time, its Overall structure principle is joined shown in Fig. 1.

The application of wireless sensor network technology also initially concentrates on military field from research and has turned to civil area, including Intelligent grid, intelligent transportation, Intelligent logistics, intelligent medical, Smart Home etc., and play increasingly in these fields Big effect.But under normal circumstances, sensor node is all the least equipment, and it is deployed in unmanned adverse circumstances more Under the conditions of, network there is node there occurs fault if which results in, be difficult to timely and effectively these malfunctioning nodes be carried out Maintenance or replacement, this generates the demand of fault diagnosis technology in a kind of valid wireless sensor device network of studying.

In more existing wireless sensor networks in the technology of Fault Node Diagnosis, temporal correlation principle is used to set Meter technical scheme is a class mode of suitable main flow.But these technical schemes are at the beginning of design, it is considered to the deployed environment of network Environment the most single, conforming, here under environment, these diagnostic techniquess all can have and preferably apply performance.But in reality In the network design on border, complex environment is only modal situation, in this kind of environment, needs in network before diagnosing Sensor node effectively classify, this has vital effect to the fault diagnosis solved under complex environment.

Therefore, for above-mentioned technical problem, it is necessary to provide the real-time event of wireless sensor network under a kind of complex environment Barrier diagnostic method and system.

Summary of the invention

In view of this, it is an object of the invention to provide the real time fail of wireless sensor network under a kind of complex environment to examine Disconnected method and system, with timely to the malfunctioning node that in the wireless sensor network being deployed under complicated adverse circumstances, the phase does not occurs Diagnosis.

To achieve these goals, the technical scheme that the embodiment of the present invention provides is as follows:

The real-time fault diagnosis method of wireless sensor network under a kind of complex environment, described diagnostic method includes:

The clustering algorithm of the k-medoids that S1, employing improve carries out sub-clustering to the sensor node in presumptive area；

S2, basis treat the data that diagnostic sensor node and neighbor node thereof gather, and calculate and treat diagnostic sensor node Vote value, determines the health status treating diagnostic sensor node according to Vote value；

S3, find the sensor entity of malfunctioning node according to sensor network, and keep in repair.

As a further improvement on the present invention, described step S1 particularly as follows:

Quantity K determined bunch according to actual environment；

Initialize K cluster centre, randomly choose K sensor node；

Initialize sub-clustering, select nearest node as a bunch head according to criterion function E；

Calculate after sub-clustering the criterion function value E sum of all the sensors node and its bunch of central sensor node in K bunch For M value；

Judge that M value is the most sufficiently small, if it is not, reselect any node in each bunch as new bunch of head so that M value drops Low, and according to new leader cluster node again sub-clustering, if so, produce cluster structured, and broadcast to sensor network.

As a further improvement on the present invention, in described step S1, initialize sub-clustering particularly as follows:

Any 2 sensor node S_iWith S_jBetween use criterion functionCalculate, S_jSelect The S that criterion function is minimum_iCenter as its sub-clustering；

Wherein, α, β represent constant coefficient；d_ijIt is the distance between two sensor nodes, d_ijUse Euclidean distance calculate or Use the intensity estimation signal between two sensors；p_ijRepresent the phase of nearest t time between two sensor node historical datas DRS degree,

As a further improvement on the present invention, in described step S1, criterion function value E sum be M value particularly as follows:

$M=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\left(\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}{E}_{K^{\′}j}\right)=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\{\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}\[d_{K^{\′}j}^{\∝}\×{(1+p_{K^{\′}j})}^{\mathrm{\β}}\]\};$

Wherein, NK ' represents bunch center S_K′Sensor node number in the place bunch of node, For the M value of independent bunch.

As a further improvement on the present invention, described step S2 particularly as follows:

Diagnostic sensor node is treated in selection；

Collect the data treating diagnostic sensor node and neighbours thereof, calculate the Vote value treating diagnostic sensor node；

If Vote value is less than or equal to first threshold, then judge to treat that diagnostic sensor node is malfunctioning node；

If Vote value is more than or equal to Second Threshold, then judge to treat that diagnostic sensor node is fault-free node；

If Vote value is more than first threshold and less than Second Threshold, then judge to treat that diagnosis node, as nondeterministic statement, updates The reference state of neighbours, collects the data treating diagnostic sensor node and neighbours thereof, recalculates Vote value, if Vote value is less than Or equal to the 3rd threshold value, then judge to treat that diagnostic sensor node is malfunctioning node, if Vote value is more than the 3rd threshold value, then judge to treat Diagnostic sensor node is fault-free node.

As a further improvement on the present invention, described step S2 being treated, diagnostic sensor node uses Neighbor (S_i) and Neighbor_same(S_iIn), sensor node participates in diagnosis process.

As a further improvement on the present invention, described step S2 is treated the Vote value of diagnostic sensor node particularly as follows:

Or

Wherein, C_ijFor node S_iWith node S_jThe most consistent variate-value,θ₁ For threshold value, weight is the weight coefficient value of ballot table.

As a further improvement on the present invention, described step S2 treats diagnostic sensor node S_iState take turns by 2 Voting process carries out diagnostic analysis.

Correspondingly, the real-time fault diagnosis system of wireless sensor network under a kind of complex environment, described diagnostic system bag Include:

Sensing unit, including some sensors and D/A converter module, some sensors form sensor network, and sensing is single Unit is used for gathering physical context information information is carried out digital-to-analogue conversion around；

Processing unit, including the processor for processing sensing unit data and for storing the storage of data Device；

Communication unit, is used for carrying out data transmission；

Power subsystem, for providing energy supply for sensing unit, processing unit and communication unit.

As a further improvement on the present invention, described processor is additionally operable to:

The clustering algorithm using the k-medoids improved carries out sub-clustering to the sensor node in presumptive area；

And, according to the data treating that diagnostic sensor node and neighbor node thereof gather, calculate and treat diagnostic sensor node Vote value, determine the health status treating diagnostic sensor node according to Vote value.

The invention has the beneficial effects as follows:

By the failure diagnostic process of three phases, allow the sensor network nodes under complex environment in experience sub-clustering, examine After the stage break, keeped in repair, it is possible to complete the maintenance of network timely and effectively；

By the process of sub-clustering, it is to avoid the harmful effect that diagnostic result is produced by complex environment；

The Distributed-solution used can reduce the energy expenditure of nodes, it is to avoid sensor node because Too much communication causes the puzzlement affecting network lifecycle.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing In having technology to describe, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below is only this Some embodiments described in invention, for those of ordinary skill in the art, on the premise of not paying creative work, Other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 is the overall structure schematic diagram of wireless sensor network (WSN)；

Fig. 2 is the real-time fault diagnosis method of wireless sensor network under complex environment in the embodiment of the invention Schematic flow sheet；

Fig. 3 is the particular flow sheet of sensor node sub-clustering in the embodiment of the invention；

Fig. 4 is the particular flow sheet of sensor node diagnosis in the embodiment of the invention；

Fig. 5 is the real-time fault diagnosis system of wireless sensor network under complex environment in the embodiment of the invention Module diagram.

Detailed description of the invention

For the technical scheme making those skilled in the art be more fully understood that in the present invention, real below in conjunction with the present invention Execute the accompanying drawing in example, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described enforcement Example is only a part of embodiment of the present invention rather than whole embodiments.Based on the embodiment in the present invention, this area is common The every other embodiment that technical staff is obtained under not making creative work premise, all should belong to present invention protection Scope.

The present invention failure diagnostic process by three phases, allows the wireless sensor network node under complex environment at warp Go through sub-clustering, diagnose, after stage of keeping in repair, it is possible to complete the maintenance of network timely and effectively, specifically include following three steps:

First according to the clustering algorithm of a kind of k-mediods after optimizing, to the node in wireless sensor network Carrying out region division, make each sensor node find therewith and be in the node under like environment, they collectively form and belong to certainly Oneself group of sensor node (or cry bunch).According to the principle of the temporal correlation that wireless sensor network has, adjacent The data of node perceived should be in acceptable scope δ；

After node to be detected starts diagnosis process, begin to the neighbor node request data with surrounding, after receiving data The data gathered with oneself compare, then statistics neighbor node and the concordance of its data, using Vote value as weighing apparatus Measure conforming parameter.During determining Vote value, take into full account the state of node surrounding neighbours, include whether as fault State, whether it was diagnosed whether be in similar environment etc. condition with node to be detected recently.State at 2 bouts After decision process, obtain the health status of node to be detected, finally the diagnostic result of node is sent to base station, at base station record All of node state in network；

Finally, after being found that malfunctioning node in record, the sensing of correspondence can be found according to the route Map view of network Device entity, carries out the operation that maintenance is changed.

In the present invention during wireless sensor network runs, sensor node is required for according to the time, is diagnosed The information such as result adjust self state in time, can be given during the breakdown judge participating in other nodes and meet reality Reference value.

Specifically, shown in ginseng Fig. 2, the real-time event of wireless sensor network under the present invention one preferred implementation complex environment Barrier diagnostic method, specifically includes:

The clustering algorithm of the k-medoids that S1, employing improve carries out sub-clustering to the sensor node in presumptive area；

S2, basis treat the data that diagnostic sensor node and neighbor node thereof gather, and calculate and treat diagnostic sensor node Vote value, determines the health status treating diagnostic sensor node according to Vote value；

S3, find the sensor entity of malfunctioning node according to sensor network, and keep in repair.

Wherein, shown in Fig. 3, step S1 specifically includes following steps:

1), quantity K determined bunch according to actual environment.

With the clustering algorithm of k-medoids, the sensor node in appointment region is carried out the operation of sub-clustering, because k- The clustering algorithm of medoids it is to be appreciated that bunch quantity K, can rule of thumb specify sub-clustering quantity K of Experimental Area, example here As teaching building can determine the quantity of sub-clustering according to the region quantity such as classroom, corridor.

2), initialize K cluster centre, in region, randomly choose K sensor node as initial cluster centre.

3), initialize sub-clustering, select nearest node as a bunch head according to criterion function E.

Any 2 sensor node S_iWith S_jBetween use criterion functionCalculate, S_jSelect The S that criterion function is minimum_iCenter as its sub-clustering (or cluster)；

Wherein, α, β represent constant coefficient, can be adjusted according to practical situation；

d_ijIt is the distance between two sensor nodes, d_ijEuclidean distance can be used to calculate or use two sensors Between intensity estimation signal.Preferably, present embodiment is taked 2 internodal distances of parameter estimation of RSSI according to letter Number intensity determines, such as: unlike signal intensity divides

p_ijRepresent the different degree (i.e. contrary with similarity) of nearest t time between two sensor node historical datas, use FormulaCalculate.

4), calculate after sub-clustering in K bunch criterion function value E of all the sensors node and its bunch of central sensor node it With for M value.

M value formula is calculated as below:

$M=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\left(\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}{E}_{K^{\′}j}\right)=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\{\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}\[d_{K^{\′}j}^{\∝}\×{(1+p_{K^{\′}j})}^{\mathrm{\β}}\]\};$

Wherein, NK ' represents bunch center S_K′Sensor node number in the place bunch of node, For the M value of independent bunch.

5), judge that M value is the most sufficiently small, if it is not, reselect any node in each bunch as new bunch of head so that M Value reduces, and according to new leader cluster node again sub-clustering, if so, produces cluster structured, and broadcasts to sensor network.

For remaining N-K sensor node, the most unduplicated node of choosing replaces its affiliated bunch center Node, calculates the criterion function value E sum M ' of now network, if M ' is ＜ M, jumps to step 3), otherwise N-K sensor joint After point is attempted, M value cannot reduce further, then terminate cluster process, export the result of current sub-clustering.Specifically comprise the following steps that

For i=1 to N-K

■ Choose node S_iReplace bunch Centroid belonging to it；

Criterion function value E sum M ' in network after ■ calculating replacement；

■ If M ' ＜ M

◆ jump procedure 3)；

Return current sub-clustering result

Shown in Fig. 4, step S2 specifically includes following steps:

Diagnostic sensor node is treated in selection；

Collect the data treating diagnostic sensor node and neighbours thereof, calculate the Vote value treating diagnostic sensor node；

If Vote value is less than or equal to first threshold (threshold value 1), then judge to treat that diagnostic sensor node is malfunctioning node；

If Vote value is more than or equal to Second Threshold (threshold value 2), then judge to treat that diagnostic sensor node is fault-free node；

If Vote value is more than first threshold (threshold value 1) and less than Second Threshold (threshold value 2), then judge to treat that diagnosis node is not as Determine state, the reference state of more new neighbor, collect the data treating diagnostic sensor node and neighbours thereof, recalculate Vote Value, if Vote value is less than or equal to the 3rd threshold value (threshold value 3), then judges to treat that diagnostic sensor node is malfunctioning node, if Vote Value more than the 3rd threshold value (threshold value 3), then judges to treat that diagnostic sensor node is fault-free node.

Specifically, treat that diagnostic sensor node uses Neighbor (S in the present embodiment_i) (represent node S_iAll Neighbor node) and/or Neighbor_same(S_i) (represent S_iThe neighbor node of all same bunch) in sensor node participate in Diagnosis process, treat the Vote value of diagnostic sensor node particularly as follows:

Or

Wherein, C_ijFor node S_iWith node S_jThe most consistent variate-value,θ₁ For threshold value, weight is the weight coefficient value of ballot table.

Present embodiment step S2 treats diagnostic sensor node S_iState by 2 take turns voting process carry out diagnosis point Analysis, concrete steps and code are as follows:

Correspondingly, shown in ginseng Fig. 5, another embodiment of the present invention also discloses wireless senser under a kind of complex environment The real-time fault diagnosis system of network, specifically includes:

Sensing unit 10, including some sensors 11 and D/A converter module 12, some sensors form sensor network, Sensing unit is used for gathering physical context information information is carried out digital-to-analogue conversion around；

Processing unit 20, including the processor 21 for processing sensing unit data and for storing depositing of data Reservoir 22；

Communication unit 30, is used for carrying out data transmission, it is preferable that communication unit 30 is radio receiving transmitting module 31, to carry out Radio communication；

Power subsystem 40, for providing energy to supply for sensing unit 10, processing unit 20 and communication unit 30.

Wherein, processor 21 specifically for:

The clustering algorithm using k-medoids carries out sub-clustering to the sensor node in presumptive area；

And, according to the data treating that diagnostic sensor node and neighbor node thereof gather, calculate and treat diagnostic sensor node Vote value, determine the health status treating diagnostic sensor node according to Vote value.

In a specific embodiment of the present invention, use the wireless development board of Jennic as the joint in wireless sensor network Point, adds Temperature Humidity Sensor module, builds WSN as shown in Figure 1, is used for monitoring the environment specifying region.

First pass through PC end, according to flow process shown in Fig. 3, complete the clustering process of sensor network in base station, then, when needing When nodes health status being diagnosed, sent the order of diagnosis to network from base station by PC end.Each net Node in network diagnoses according to the flow process of Fig. 4, after node to be detected starts diagnosis process, with the neighbor node of surrounding please Seeking humiture information, the warm and humid angle value gathered with oneself after receiving data compares, then statistics neighbor node and self The concordance of data, using Vote value as weighing conforming parameter.During determining Vote value, take into full account node week Enclose the state of neighbours, include whether as malfunction, whether to be recently diagnosed and whether be in similar with node to be detected Environment etc. condition.After the state decision process of 2 bouts, obtain the health status of node to be detected, finally examining node Disconnected result sends to base station, all of node state in the record network of base station.After being found that malfunctioning node in record, permissible Route Map view according to network finds the sensor entity of correspondence, carries out the operation that maintenance is changed.

As can be seen from the above technical solutions, the method have the advantages that

By the failure diagnostic process of three phases, allow the wireless sensor network node under complex environment in experience point Bunch, diagnose, after stage of keeping in repair, it is possible to complete the maintenance of network timely and effectively；

By the process of sub-clustering, it is to avoid the harmful effect that diagnostic result is produced by complex environment；

The Distributed-solution used can reduce the energy expenditure of nodes, it is to avoid sensor node because Too much communication causes the puzzlement affecting network lifecycle.

It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie In the case of the spirit or essential attributes of the present invention, it is possible to realize the present invention in other specific forms.Therefore, no matter From the point of view of which point, all should regard embodiment as exemplary, and be nonrestrictive, the scope of the present invention is by appended power Profit requires rather than described above limits, it is intended that all by fall in the implication of equivalency and scope of claim Change is included in the present invention.Should not be considered as limiting involved claim by any reference in claim.

Although moreover, it will be appreciated that this specification is been described by according to embodiment, but the most each embodiment only wraps Containing an independent technical scheme, this narrating mode of description is only that for clarity sake those skilled in the art should Description can also be formed those skilled in the art through appropriately combined as an entirety, the technical scheme in each embodiment May be appreciated other embodiments.

Claims (10)

1. the real-time fault diagnosis method of wireless sensor network under a complex environment, it is characterised in that described diagnostic method Including:

The clustering algorithm of the k-medoids that S1, employing improve carries out sub-clustering to the sensor node in presumptive area；

S2, basis treat the data that diagnostic sensor node and neighbor node thereof gather, and calculate the Vote treating diagnostic sensor node Value, determines the health status treating diagnostic sensor node according to Vote value；

S3, find the sensor entity of malfunctioning node according to sensor network, and keep in repair.

Diagnostic method the most according to claim 1, it is characterised in that described step S1 particularly as follows:

Quantity K determined bunch according to actual environment；

Initialize K cluster centre, randomly choose K sensor node；

Initialize sub-clustering, select nearest node as a bunch head according to criterion function E；

After calculating sub-clustering, in K bunch, the criterion function value E sum of all the sensors node and its bunch of central sensor node is M Value；

Judge that M value is the most sufficiently small, if it is not, reselect any node in each bunch as new bunch of head so that M value reduces, And according to new leader cluster node again sub-clustering, if so, produce cluster structured, and broadcast to sensor network.

Diagnostic method the most according to claim 2, it is characterised in that in described step S1, initialize sub-clustering particularly as follows:

Any 2 sensor node S_iWith S_jBetween use criterion functionCalculate, S_jSelection criterion The S that function is minimum_iCenter as its sub-clustering；

Wherein, α, β represent constant coefficient；d_ijIt is two sensor node spacings, d_ijUse Euclidean distance to calculate or use two Intensity estimation signal between individual sensor；p_ijRepresent the different degree of nearest t time between two sensor node historical datas,

Diagnostic method the most according to claim 3, it is characterised in that in described step S1, criterion function value E sum is M Value particularly as follows:

$M=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\left(\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}{E}_{K^{\′}j}\right)=\underset{K^{\′}=1}{\overset{K^{\′}=K}{\Σ}}\{\underset{j=1}{\overset{j={\mathrm{NK}}^{\′}}{\Σ}}\[d_{K^{\′}j}^{\∝}\×{(1+p_{K^{\′}j})}^{\mathrm{\β}}\]\};$

Wherein, NK ' represents bunch center S_K′Sensor node number in the place bunch of node,For The M value of independent one bunch.

Diagnostic method the most according to claim 1, it is characterised in that described step S2 particularly as follows:

Diagnostic sensor node is treated in selection；

Collect the data treating diagnostic sensor node and neighbours thereof, calculate the Vote value treating diagnostic sensor node；

If Vote value is less than or equal to first threshold, then judge to treat that diagnostic sensor node is malfunctioning node；

If Vote value is more than or equal to Second Threshold, then judge to treat that diagnostic sensor node is fault-free node；

If Vote value is more than first threshold and less than Second Threshold, then judge to treat diagnosis node as nondeterministic statement, more new neighbor Reference state, collect and treat the data of diagnostic sensor node and neighbours thereof, recalculate Vote value, if Vote value is less than or etc. In the 3rd threshold value, then judge to treat that diagnostic sensor node is malfunctioning node, if Vote value is more than the 3rd threshold value, then judge to wait to diagnose Sensor node is fault-free node.

Diagnostic method the most according to claim 5, it is characterised in that treat in described step S2 that diagnostic sensor node uses Neighbor(S_i) and Neighbor_same(S_iIn), sensor node participates in diagnosis process.

Diagnostic method the most according to claim 6, it is characterised in that treat diagnostic sensor node in described step S2 Vote value particularly as follows:

Or

Wherein, C_ijFor node S_iWith node S_jThe most consistent variate-value,θ₁For threshold Value, weight is the weight coefficient value of ballot table.

Diagnostic method the most according to claim 1, it is characterised in that treat diagnostic sensor node S in described step S2_i State take turns voting process carry out diagnostic analysis by 2.

9. the real-time fault diagnosis system of wireless sensor network under a complex environment, it is characterised in that described diagnostic system Including:

Sensing unit, including some sensors and D/A converter module, some sensors form sensor network, and sensing unit is used In collection physical context information around and information is carried out digital-to-analogue conversion；

Processing unit, including the processor for processing sensing unit data and for storing the memorizer of data；

Communication unit, is used for carrying out data transmission；

Power subsystem, for providing energy supply for sensing unit, processing unit and communication unit.

Diagnostic system the most according to claim 9, it is characterised in that described processor is additionally operable to:

The clustering algorithm using the k-medoids improved carries out sub-clustering to the sensor node in presumptive area；

And, according to the data treating that diagnostic sensor node and neighbor node thereof gather, calculate and treat diagnostic sensor node Vote value, determines the health status treating diagnostic sensor node according to Vote value.