CN105471632A - Auto-regression line fault detection method - Google Patents

Auto-regression line fault detection method Download PDF

Info

Publication number
CN105471632A
CN105471632A CN201510793624.8A CN201510793624A CN105471632A CN 105471632 A CN105471632 A CN 105471632A CN 201510793624 A CN201510793624 A CN 201510793624A CN 105471632 A CN105471632 A CN 105471632A
Authority
CN
China
Prior art keywords
dimension
message characteristic
gateway message
fault
gateway
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
Application number
CN201510793624.8A
Other languages
Chinese (zh)
Other versions
CN105471632B (en
Inventor
张风雨
程国振
王雨
朱圣平
张震
白冰
王志明
王鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
PLA Information Engineering University
Original Assignee
PLA Information Engineering University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by PLA Information Engineering University filed Critical PLA Information Engineering University
Priority to CN201510793624.8A priority Critical patent/CN105471632B/en
Publication of CN105471632A publication Critical patent/CN105471632A/en
Application granted granted Critical
Publication of CN105471632B publication Critical patent/CN105471632B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/06—Management of faults, events, alarms or notifications
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/06—Management of faults, events, alarms or notifications
    • H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
    • H04L41/0668—Management of faults, events, alarms or notifications using network fault recovery by dynamic selection of recovery network elements, e.g. replacement by the most appropriate element after failure
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/06—Management of faults, events, alarms or notifications
    • H04L41/0695—Management of faults, events, alarms or notifications the faulty arrangement being the maintenance, administration or management system
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0852—Delays
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L2209/00—Additional information or applications relating to cryptographic mechanisms or cryptographic arrangements for secret or secure communication H04L9/00
    • H04L2209/16—Obfuscation or hiding, e.g. involving white box

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The invention discloses an auto-regression line fault detection method comprising the following steps: acquiring equipment gateway message characteristics; determining the delay inequality of the acquired gateway message characteristic for each dimension based on the maximum cross-correlation algorithm to measure the correlation; quantifying the gateway message characteristic for each dimension based on the cosine similarity algorithm to determine the degree of similarity between the gateway message characteristics; and carrying out fault detection based on the D-S evidence theory and according to the degree of similarity between the multi-dimension and multi-cycle gateway message characteristics to determine whether there is a fault in an auto-regression line. The method provided by the invention does not rely on the running status of equipment in detection. The problem that there is no comprehensive 'black box' fault detection method in the prior art, and fault detection cannot be carried out when a 'write box' fails is solved.

Description

A kind of detection method of autoregression line fault
Technical field
The present invention relates to detection field, particularly relate to a kind of detection method of autoregression line fault.
Background technology
The Internet management and control devices is serially connected with discrepancy critical point, the Internet, if equipment breaks down, will have a strong impact on network operation, and cause extremely severe social influence.Therefore, need to protect the Internet management and control devices, with normal operation equipment being carried out to fault detect, suppress fault impact scope, ensure backhaul.
Current fault detect is mainly divided into " white box " to detect and black box detects two kinds.The former runs on device interior, detects each subsystem, the fault of each module, Timeliness coverage fast processing." white box " detection technique usually used as the standard configuration of equipment with deployed with devices on circuit." white box " detection technique can grasp detailed device interior equipment operation information, can fast detecting localizing faults reason, but " white box " detection technique runs in equipment as equipment standard configuration, when equipment itself breaks down, white box detects and also will break down thereupon, and at this moment " white box " detects and will lose efficacy.
The current fault detect for backbone network safeguard faces two and challenges greatly.First, the Internet management and control devices function is complicated, and be made up of different processing units, each unit is made up of multiple subsystem again, and along with the increase of system function module, the factor sharp increase of initiating failure, proposes severe challenge to fault detect.Secondly, the Internet management and control devices is deployed in international critical point, and electromagnetic environment is complicated and changeable, and bursts of traffic degree is high, very easily initiating system fault.Therefore, solve above-mentioned two hang-ups, system not only will possess conventional " white box " fault detect means, and will possess the black box fault detection method outside independent of Business Processing.But prior art is a kind of comprehensive black box fault detection method not, when " white box " lost efficacy, cannot detect fault.
Summary of the invention
The invention provides a kind of detection method of autoregression line fault, there is no a kind of comprehensive black box fault detection method in order to solve prior art, when " white box " lost efficacy, the problem that cannot detect fault.
For solving the problems of the technologies described above, on the one hand, the invention provides a kind of detection method of autoregression line fault, comprising: obtain equipment gateway message characteristic; The delay inequality of the gateway message characteristic of each dimension got is determined, to measure its correlation based on maximum cross correlation algorithm; The gateway message characteristic of cosine similarity algorithm to each dimension is adopted to quantize, to determine the similarity degree of entrance message feature; Based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic described gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault.
Further, the delay inequality of the gateway message characteristic of each dimension got is determined based on maximum cross correlation algorithm, to measure its correlation, comprising: respectively cross-correlation function calculating is carried out to the gateway message characteristic of different dimensions, to determine the cross correlation value that each dimension is maximum; The mean value getting all maximum cross correlation values is as to the delay inequality between the message of gateway.
Further, the gateway message characteristic of cosine similarity algorithm to each dimension is adopted to quantize, to determine the similarity degree of entrance message feature, comprising: adopt cosine similarity formula to calculate the gateway message characteristic value of each dimension in N number of dimension respectively; Using the gateway message characteristic value of dimension each in N number of dimension as a point in N dimension coordinate system, the point corresponding by the gateway message characteristic connecting each dimension and the initial point of coordinate system form straight line, wherein, to get over similarity degree high for this straight line and the horizontal coordinate angle number of degrees little corresponding gateway message characteristic; Or, judge whether described gateway message characteristic value value is between [-1,1], and when being between [-1,1], determine value close to 1 corresponding gateway message characteristic to get over similarity degree high.
Further, based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic described gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault, comprise: the similarity degree obtaining various dimensions multiply periodic described gateway message characteristic, using as evidence; Calculate the basic probability assignment of each described evidence; The described basic probability assignment corresponding according to each dimension calculates belief function Bel; Based on D-S evidence theory, brief combination is carried out to described belief function Bel, with to synthesis result, determine whether autoregression circuit exists fault.
Further, after carrying out fault detect, also comprise: deposit in the case of a fault at autoregression circuit according to the similarity degree of various dimensions multiply periodic described gateway message characteristic, startup separator suppresses operation, so that equipment is switched to bypass condition.
The present invention is measured the correlation between the gateway message characteristic of each dimension by maximum cross correlation algorithm, quantized by the gateway message characteristic of cosine similarity algorithm to each dimension, finally whether exist based on D-S evidence theory detection failure, this process does not rely on equipment running status and detects, solve prior art and there is no a kind of comprehensive black box fault detection method, when " white box " lost efficacy, the problem that cannot detect fault.
Accompanying drawing explanation
Fig. 1 is the flow chart of the detection method of autoregression line fault in the embodiment of the present invention;
Fig. 2 is based on configuration diagram residing for the detection method of the autoregression line fault of D-S evidence theory in the preferred embodiment of the present invention;
Fig. 3 suppresses schematic diagram based on the autoregression line fault of D-S evidence theory in the preferred embodiment of the present invention;
Fig. 4 is the information fusion block diagram based on D-S means of proof in the preferred embodiment of the present invention;
Fig. 5 is compartmention schematic diagram in D-S evidence theory in the preferred embodiment of the present invention;
Fig. 6 is the fault detect flow chart based on D-S evidence theory in the preferred embodiment of the present invention;
Fig. 7 is the autoregression failure restraint flow chart based on D-S evidence theory in the preferred embodiment of the present invention.
Embodiment
A kind of comprehensive black box fault detection method is not had in order to solve prior art, when " white box " lost efficacy, the problem that cannot detect fault, the invention provides a kind of detection method of autoregression line fault, below in conjunction with accompanying drawing and embodiment, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, do not limit the present invention.
The flow process of the detection method of the autoregression line fault that the embodiment of the present invention provides as shown in Figure 1, comprises step S102 to S108:
S102, obtains equipment gateway message characteristic;
S104, determines the delay inequality of the gateway message characteristic of each dimension got based on maximum cross correlation algorithm, to measure its correlation;
S106, adopts the gateway message characteristic of cosine similarity algorithm to each dimension to quantize, to determine the similarity degree of entrance message feature;
S108, based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault.
The embodiment of the present invention is measured the correlation between the gateway message characteristic of each dimension by maximum cross correlation algorithm, quantized by the gateway message characteristic of cosine similarity algorithm to each dimension, finally whether exist based on D-S evidence theory detection failure, this process does not rely on equipment running status and detects, solve prior art and there is no a kind of comprehensive black box fault detection method, when " white box " lost efficacy, the problem that cannot detect fault.
In implementation procedure, determine that based on maximum cross correlation algorithm the process of the delay inequality of the gateway message characteristic of each dimension got can comprise: respectively cross-correlation function calculating is carried out to the gateway message characteristic of different dimensions, to determine the cross correlation value that each dimension is maximum; The mean value getting all maximum cross correlation values is as to the delay inequality between the message of gateway.
In implementation process, the gateway message characteristic of cosine similarity algorithm to each dimension is adopted to quantize, to determine the similarity degree of entrance message feature, comprising: adopt cosine similarity formula to calculate the gateway message characteristic value of each dimension in N number of dimension respectively; Using the gateway message characteristic value of dimension each in N number of dimension as a point in N dimension coordinate system, the point corresponding by the gateway message characteristic connecting each dimension and the initial point of coordinate system form straight line, wherein, to get over similarity degree high for this straight line and the horizontal coordinate angle number of degrees little corresponding gateway message characteristic; Or, judge whether entrance message characteristic value value is between [-1,1], and when being between [-1,1], determine value close to 1 corresponding gateway message characteristic to get over similarity degree high.
Finally, based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault, comprise: the similarity degree obtaining various dimensions multiply periodic gateway message characteristic, using as evidence; Calculate the basic probability assignment of each evidence; The basic probability assignment corresponding according to each dimension calculates belief function Bel; Based on D-S evidence theory, brief combination is carried out to belief function Bel, with to synthesis result, determine whether autoregression circuit exists fault.
After carrying out fault detect according to the similarity degree of various dimensions multiply periodic gateway message characteristic, if autoregression circuit is deposited in the case of a fault, startup separator suppresses operation, so that equipment is switched to bypass condition.
Preferred embodiment
For ensure white box detect lost efficacy time, still can checkout equipment occur fault and take failure restraint mechanism, the present invention proposes a kind of black box fault detection method, and black box fault detect does not rely on equipment running status.
The black box fault detection method that the embodiment of the present invention provides is the detection method of the autoregression line fault based on D-S evidence theory, to solve the problem of existing the Internet management and control devices difficult fault diagnosis.For solving the problems of the technologies described above, the present invention program is as follows:
First based on the delay inequality of maximum crosscorrelation estimation equipment gateway message characteristic, its correlation is measured, accurately to portray the uncertainty of equipment state; Secondly, cosine similarity is adopted to quantize the similarity degree of gateway feature; Again introduce D-S evidence theory, Comprehensive Evaluation is carried out to the multiply periodic fault detection information of various dimensions, and starts corresponding failure handling mechanisms according to court verdict.
Framework signal residing for the detection method of the autoregression line fault based on D-S evidence theory that the present embodiment provides as shown in Figure 2, in implementation procedure, first, failure monitoring agency is disposed in equipment each processing unit, the turnover message of equipment plate card and data processing unit is sampled, extracts the multidimensional characteristics such as bag quantity, packet length and inter-packet gap; Secondly, calculate the correlation of inlet features and corresponding exporting features, obtain cross-correlation function, and estimate that proper phase is poor, and then obtain cross-correlation coefficient; Finally, to the multi-source multidimensional cross-correlation coefficient obtained, D-S evidence theory is adopted to carry out amalgamation judging.
When realizing, if find or after perceiving fault, the Internet management and control devices startup separator suppression mechanism, link automatic returning to the process of initial condition, its illustrate as shown in Figure 3.
Multidimensional characteristic (bag size, bag length, inter-packet gap etc.) in the gateway net bag time domain that fault detect adopts can regard time varying signal as, Correlation Theory is utilized to contrast gateway net bag feature, determine the time delay between entrance and similitude, detect according to similarity degree and whether break down, and startup separator suppresses operation in good time.
For cross-correlation function, it is used to a tolerance of similitude between expression two signals, usually by comparing the characteristic for finding in unknown signaling with known signal.It is a function relative to the time between two signals.
If x (n), y (n) are two groups of length is this sequence signal of N full pattern, relevant between them is discrete correlation, and its discrete correlation is defined as:
m＝-N+1,…,0,…,N-1；
For bag size characteristic, if x (n) is inlet features, y (n) is corresponding exporting features, considers the late effect of equipment, and above-mentioned feature is respectively x (n) and y (n+m).
If two groups of these sequence signals of full pattern have certain similitude, the peak point of so discrete cross-correlation function just reflects their similitude, and the position obtaining peak point is exactly time delay.Therefore, correlation function R (m) amplitude size reflects the similitude of gateway flow, considers again equipment delay difference simultaneously.
The Internet management and control devices has carried out multi-dimension feature extraction to gateway net bag, and the characteristic information of different dimensions can be used for calculating the m value making cross-correlation function maximum respectively, and the mean value getting m under all situations is estimated as to the time delay between the net bag of gateway.
Although the peak value of cross-correlation function also can indicate the similitude between entrance net bag, the cross-correlation function peak value of different bag feature may differ greatly, and is difficult in order to information fusion.Based on the gateway feature delay inequality m that above-mentioned method of estimation obtains, vectorial similitude is utilized to weigh inlet features X=[x (k), x (k+1), L, x (k+N)] and exporting features Y=[y (k+m), y (k+m+1), L, y (k+m+N)] between difference, wherein, N represents the fault detect cycle.
Conventional vectorial similarity calculation method mainly contains Pearson correlation coefficient, Euclidean distance and cosine similarity etc.Wherein, Pearson correlation coefficient needs compute vector variance and covariance, and amount of calculation is large, Euclidean distance does not consider the correlation between each vector element, what cosine similarity was weighed is the angle of space vector, is more the difference be embodied on direction, instead of position.Therefore the embodiment of the present invention adopts cosine similarity to weigh the otherness of gateway feature.
Cosine similarity computing formula is as follows:
T ( x , y ) = x · y | | x | | 2 × | | y | | 2 = Σx i y i Σx i 2 Σy i 2 .
Computational methods based on cosine similarity are that n-is tieed up gateway bag feature as a point in n-dimension coordinate system, and form straight line (vector) by connecting this point with the initial point of coordinate system, the less representative of angle is more similar.
The Internet management and control devices adopts cosine similarity formula to calculate the similitude of entrance different characteristic respectively, and value is between [-1,1].Close to 1, similitude value more represents that system is more stable, the possibility more departing from 1 expression system failure is larger.
As shown in Figure 4, the feature of each dimension is divided into multiple cycle by the embodiment of the present invention in time domain, using the corroboration body of the similarity of the gateway bag feature in each cycle as D-S evidence theory judgement framework, carry out probability assignment according to similarity mode degree, and then complete amalgamation judging.
D-S evidence theory is a kind of reasoning method under uncertainty, its maximum feature is the method that the description of uncertain information be have employed to " interval estimation " instead of " point estimation ", differentiation do not know with uncertain in and accurately reflect and demonstrate very large flexibility in evidence-gathering, this characteristic can portray the uncertainty of failure detection result.
D-S evidence theory meets the condition more weak than Bayes probability theory, does not namely need to know prior probability, has the ability of direct expression " uncertain " and " not knowing ".The difference of D-S method and other probabilistic methods is: its evidence ambiguity of D-S method makes single evidence lack adequacy, needs comprehensively to adjudicate by merging multiple evidence body, thus reaches higher accuracy.Provide following concept below:
Basic reliability value: set Θ as identification framework, if set function m:2 Θ→ [0,1] (2 Θpower set for Θ) meet Σ m (A)=1, then title m is the basic brief inference on identification framework Θ; m (A) is called the basic reliability value of A.
M (A) reflects the size of the reliability to A itself.In fault detect of the present invention, basic reliability value reflects the credibility of the gateway feature as equipment fault evidence.Fig. 5 is compartmention schematic diagram in D-S evidence theory.
Belief function: if m is a basic brief inference, then b ≠ φ, then defined function Bel is a belief function, and Bel (A) reflects the reliability that on A, all subsets are total.
If there is A to belong to identification framework Θ, definition Dou (A)=Bel (A); Bel (A)=1-Bel (A), then claim the likelihood degree of the Pl likelihood function that (A) is Bel and A, namely describes the likelihood of A or reliable degree.Dou is the suspection function of Bel, and the suspicious degree that Dou (A) is A, what describe A does not be sure of degree.In fact, [Bel (A), Pl (A)] illustrates the indeterminacy section of A, i.e. the bound of probability.Belief function is then to the probabilistic quantitative description in process fault detection.
About the synthesis of reliability, D-S evidence theory proposes following rule:
If m 1, m 2belief function respectively on corresponding same identification framework Θ distributes, and burnt unit is respectively A 1, A 2... ,a kand B 1, B 2..., B k; If: Σ A i∩ B j{ m 1(A i) m 2(B j) < 1, then the function m:2 defined by following formula → (0,1) be associating after belief function distribute:
m ( A ) = &Sigma; A = A i &cap; B j m 1 ( A i ) m 2 ( B j ) 1 - C , A &NotEqual; Q 0 , A = Q ;
C=Σ A i∩ B j{ m 1(A i) m 2(B j) < 1, represent the conflict spectrum of evidence.The building-up process of reliability reflects the fusion of D-S evidence theory to multi-source evidence.In actual applications, this process model building carries out the information fusion of multi-source various dimensions to the data (evidence of fault detect) that multiple failure monitoring agency collects.
The acquisition of basic probability assignment function is a very important link in D-S evidence theory, and it directly affects accuracy and the reliability of fusion results, but in D-S evidence theory, do not have concrete description.In general, acquisition and the application of Basic Probability As-signment are closely related, and its defining method is with larger subjectivity.The present invention is directed to fault detect and the actual conditions suppressed, being constructed as follows Basic probability assignment function:
If the cosine similarity that failure monitoring acts on behalf of i output is X i, target monitoring type is A j(N c=3, A 1=-1, A 2=0, A 3=1), N crepresent number of targets.Then both distances are: d ij(X i, A j)=| X i-A j|, distance d ij(X i, A j) larger, then failure monitoring acts on behalf of i and target A jdegree of correlation lower; Otherwise, distance d ij(X i, A j) less, then failure monitoring acts on behalf of i and target A j, degree of correlation higher, definition C i(A j)=1/d ij(X i, A j).
Failure monitoring agency with the maximum correlation of target is:
α i＝max{C i(A j)}＝1/min{d ij(X i,A j)}；
The breadth coefficient that failure monitoring acts on behalf of i and each target coefficient correlation is:
&beta; i = &lsqb; N C w i &Sigma; j C i ( A j ) - 1 &rsqb; / ( N C - 1 ) ;
The safety factor that failure monitoring acts on behalf of i is:
Comprehensive formula above obtains failure monitoring and acts on behalf of i imparting target A jbasic probability assignment be:
m i ( A j ) = w i C i ( A j ) &Sigma; i w i C i ( A j ) + N s ( 1 - R i ) ( 1 - &alpha; i &beta; i ) ;
The formula that failure monitoring acts on behalf of the uncertain probable value of i is:
m i ( &theta; ) = 1 - &Sigma; j m i ( A ( j ) ) ;
In above formula, N s--failure monitoring proxy number; w i---weight coefficient, according to coefficient correlation C i(A j) size value, and 0<w i<l.Above formula meets the definite condition of mass function; When calculating the basic probability assignment that failure monitoring is acted on behalf of target (comprising Θ), first failure monitoring agency and the spacing of target (monitoring type) and the corresponding relation of correlation and the distribution of correlation is set up, and monitoring agent safety factor, weight coefficient is introduced with the correlation of target again according to failure monitoring agency, the introducing of weight coefficient, effectively enhances the correctness of the result of decision.
Based on D-S evidence theory fault detect process step as shown in Figure 6:
Step 601: to the similarity detecting the gateway feature obtained from failure monitoring agency, as evidence, calculate its basic probability assignment;
Step 602: according to the basic probability assignment of evidence, calculates belief function Bel;
Step 603: according to the composition rule of D-S evidence theory, brief combination is carried out to the belief function that step 602 obtains, obtains synthesis result.
Autoregression line fault suppression technology based on D-S evidence theory utilizes multidimensional information to describe the running status of the Internet management and control devices from different aspects, by merging multidimensional characteristic, effective raising detection accuracy, thus utilize autoregression line fault suppression technology to be switched to bypass condition rapidly when the system failure, ensure circuit stable operation.
Provide the fault detect flow process of the autoregression line fault suppressing method based on D-S evidence theory below, as shown in Figure 7, comprise following process.
Step 701: fault agency collects the gateway grouping feature of corresponding board;
Step 702: for the gateway grouping feature of same board, based on maximum correlation, estimates inlet features due to equipment and postpones the delay inequality that causes, to correct gateway characteristic error;
Step 703: calculate the gateway feature cosine similarity after correcting;
Step 704: using various dimensions cosine similarity as evidence, in input D-S evidence theory, carries out various dimensions multicycle information fusion, generates final judging result;
Step 705: according to the in good time startup separator suppression mechanism of court verdict.
Although be example object, disclose the preferred embodiments of the present invention, it is also possible for those skilled in the art will recognize various improvement, increase and replacement, and therefore, scope of the present invention should be not limited to above-described embodiment.

Claims (5)

1. a detection method for autoregression line fault, is characterized in that, comprising:
Acquisition equipment gateway message characteristic;
The delay inequality of the gateway message characteristic of each dimension got is determined, to measure its correlation based on maximum cross correlation algorithm;
The gateway message characteristic of cosine similarity algorithm to each dimension is adopted to quantize, to determine the similarity degree of entrance message feature;
Based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic described gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault.
2. detection method as claimed in claim 1, is characterized in that, determine the delay inequality of the gateway message characteristic of each dimension got based on maximum cross correlation algorithm, to measure its correlation, comprising:
Respectively cross-correlation function calculating is carried out to the gateway message characteristic of different dimensions, to determine the cross correlation value that each dimension is maximum;
The mean value getting all maximum cross correlation values is as to the delay inequality between the message of gateway.
3. detection method as claimed in claim 1, is characterized in that, adopts the gateway message characteristic of cosine similarity algorithm to each dimension to quantize, to determine the similarity degree of entrance message feature, comprising:
Cosine similarity formula is adopted to calculate the gateway message characteristic value of each dimension in N number of dimension respectively;
Using the gateway message characteristic value of dimension each in N number of dimension as a point in N dimension coordinate system, the point corresponding by the gateway message characteristic connecting each dimension and the initial point of coordinate system form straight line, wherein, to get over similarity degree high for this straight line and the horizontal coordinate angle number of degrees little corresponding gateway message characteristic;
Or, judge whether described gateway message characteristic value value is between [-1,1], and when being between [-1,1], determine value close to 1 corresponding gateway message characteristic to get over similarity degree high.
4. detection method as claimed in claim 1, it is characterized in that, based on D-S evidence theory, the similarity degree according to various dimensions multiply periodic described gateway message characteristic carries out fault detect, to determine whether autoregression circuit exists fault, comprising:
Obtain the similarity degree of various dimensions multiply periodic described gateway message characteristic, using as evidence;
Calculate the basic probability assignment of each described evidence;
The described basic probability assignment corresponding according to each dimension calculates belief function Bel;
Based on D-S evidence theory, brief combination is carried out to described belief function Bel, with to synthesis result, determine whether autoregression circuit exists fault.
5. the detection method according to any one of Claims 1-4, is characterized in that, after carrying out fault detect, also comprises according to the similarity degree of various dimensions multiply periodic described gateway message characteristic:
Deposit in the case of a fault at autoregression circuit, startup separator suppresses operation, so that equipment is switched to bypass condition.
CN201510793624.8A 2015-11-18 2015-11-18 A kind of detection method of autoregression line fault Active CN105471632B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510793624.8A CN105471632B (en) 2015-11-18 2015-11-18 A kind of detection method of autoregression line fault

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510793624.8A CN105471632B (en) 2015-11-18 2015-11-18 A kind of detection method of autoregression line fault

Publications (2)

Publication Number Publication Date
CN105471632A true CN105471632A (en) 2016-04-06
CN105471632B CN105471632B (en) 2018-11-06

Family

ID=55608946

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510793624.8A Active CN105471632B (en) 2015-11-18 2015-11-18 A kind of detection method of autoregression line fault

Country Status (1)

Country Link
CN (1) CN105471632B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114866396A (en) * 2022-07-07 2022-08-05 浩鲸云计算科技股份有限公司 Method for realizing network fault location under inaccurate resources based on text similarity

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104714537A (en) * 2015-01-10 2015-06-17 浙江大学 Fault prediction method based on joint relative change analysis and autoregression model
CN104899762A (en) * 2015-04-09 2015-09-09 哈尔滨工程大学 Trust management method based on backward inference

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104714537A (en) * 2015-01-10 2015-06-17 浙江大学 Fault prediction method based on joint relative change analysis and autoregression model
CN104899762A (en) * 2015-04-09 2015-09-09 哈尔滨工程大学 Trust management method based on backward inference

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114866396A (en) * 2022-07-07 2022-08-05 浩鲸云计算科技股份有限公司 Method for realizing network fault location under inaccurate resources based on text similarity

Also Published As

Publication number Publication date
CN105471632B (en) 2018-11-06

Similar Documents

Publication Publication Date Title
CN109474472A (en) A kind of fault detection method based on the more cell space filtering of holohedral symmetry
CN103116688B (en) For the multi-source Dissimilar sensors targetpath correlating method of airborne avionics system
US11480594B2 (en) Systems and methods of bad data identification and recovery for electric power systems
CN103728598B (en) The method of track spoofing interference is suppressed with the active radar and passive radar net of other place configure
CN105717505B (en) The data correlation method of multiple target tracking is carried out using Sensor Network
CN108089183B (en) Detection and tracking integrated method for asynchronous multi-base radar system
CN106054169B (en) Multistation Radar Signal Fusion detection method based on tracking information
CN103728599B (en) The method of false targets interference is suppressed with the active radar and passive radar net of other place configure
Chen et al. Cyber-physical systems: Dynamic sensor attacks and strong observability
CN110942109A (en) PMU false data injection attack prevention method based on machine learning
CN103970997A (en) Method for rapidly diagnosing unmanned helicopter sensor faults
CN104267439A (en) Unsupervised human detecting and positioning method
CN103905440A (en) Network security situation awareness analysis method based on log and SNMP information fusion
CN103338451B (en) Distributed malicious node detection method in a kind of wireless sensor network
CN110889862B (en) Combined measurement method for multi-target tracking in network transmission attack environment
CN106707304A (en) Satellite navigation receiver fault satellite detection method
CN104125112A (en) Physical-information fuzzy inference based smart power grid attack detection method
CN108682140B (en) Enhanced anomaly detection method based on compressed sensing and autoregressive model
CN109936568B (en) Malicious attack prevention sensor data acquisition method based on recurrent neural network
CN104133983A (en) Multiple-target Bernoulli distribution type fusion method based on weighting negative first-order RD (Renyi Divergence)
CN103458504B (en) A kind of sensor node positioning method based on sequence and device
CN108037493A (en) Double threshold CFAR detection method under clutter subspace based on multistatic radar
CN113534199B (en) Self-adaptive generalized accumulation and GPS spoofing attack detection method
CN107085167A (en) A kind of transmission line failure localization method based on big data
CN104199051A (en) Method for detecting and identifying satellite navigation RAIM (Receiver Autonomous Integrity Monitoring) multi-satellite faults

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant