CN105844023B - A kind of probabilistic testability modeling method of consideration test point - Google Patents

A kind of probabilistic testability modeling method of consideration test point Download PDF

Info

Publication number
CN105844023B
CN105844023B CN201610177270.9A CN201610177270A CN105844023B CN 105844023 B CN105844023 B CN 105844023B CN 201610177270 A CN201610177270 A CN 201610177270A CN 105844023 B CN105844023 B CN 105844023B
Authority
CN
China
Prior art keywords
disturbance degree
test point
rate
false
test
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
Application number
CN201610177270.9A
Other languages
Chinese (zh)
Other versions
CN105844023A (en
Inventor
侯文魁
范小林
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.)
Beijing Tianhang Changying Technology Co.,Ltd.
Original Assignee
Beihang 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 Beihang University filed Critical Beihang University
Priority to CN201610177270.9A priority Critical patent/CN105844023B/en
Publication of CN105844023A publication Critical patent/CN105844023A/en
Application granted granted Critical
Publication of CN105844023B publication Critical patent/CN105844023B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/30—Circuit design
    • G06F30/36—Circuit design at the analogue level
    • G06F30/367—Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Alarm Systems (AREA)

Abstract

A kind of probabilistic testability modeling method of consideration test point, this method include four steps: (1) after carrying out description system using dependency graph representation model, the failure rate of each component and the failure rate of each test point are found according to the historical summary of system；(2) the undetectable rate of computing system, the second class false alarm rate and untrustworthy rate；(3) finding out corresponding system when each test point is not known can detect disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree；(4) the test correlation matrix (detectable disturbance degree correlation matrix, anti-false-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix, devoid of risk disturbance degree test correlation matrix) when test point is not known is established.After obtaining test correlation matrix by aforementioned four step, intelligent algorithm can be used and carry out the preferred of test point, so that testability index is targetedly improved, optimal inspection design.Therefore the present invention has filled up a blank of testability technical field.

Description

A kind of probabilistic testability modeling method of consideration test point
Technical field
The invention belongs to testability technical fields, the in particular to a kind of probabilistic testability modeling side of consideration test point Method.
Background technique
Existing testability modeling, which is frequently in, to be carried out under the conditions of discriminating hypotheses, without consider in practice not really Qualitative factor.This model often establish a large amount of certainty assume on the basis of, as long as such as it is faulty, in signal flow Test point can measure, i.e., do not consider the reliability and the influence caused by test result of other uncertain factors of test point.But In fact, the effect of estimating for causing testability to design is better than actual effect since a large amount of uncertain factors are artificially ignored.
For the uncertain problem of testability, carry out many researchs both at home and abroad at present, proposes many new tests Property modeling method.Wang Baolong, Huang Kaoli et al. analyze the uncertain problem of testability in terms of the two from test and failure, Uncertain test problem is modeled and analyzed based on Bayesian network testability model, is based on hybrid diagnosis model pair Failure uncertain problem is modeled and has been analyzed, and finally by Bayesian network testability model and hybrid diagnosis model phase Fusion, proposes testability modeling and method for predicting based on hybrid diagnosis Bayesian network, and make testability design estimates effect Fruit is more credible.Chen Xiyang, Qiu Jing et al. are for equipment diagnosis and test uncertain problem generally existing in real process, Uncertainty probability is tested by introducing, establishes the testability analysis model based on Bayesian network, obtains test on this basis Failure-test correlation matrix under condition of uncertainty calculates testability index parameter through Bayesian inference, establishes test item Collect Optimized model, and is solved using mixing binary system population-genetic algorithm.Case, which is verified, to be shown, the analysis and calculating Process is uncertain due to considering test so that result and actual conditions are more coincide, with traditional certainty optimization method Compared to higher confidence level.Dai Jing et al. proposes to be associated with based on object-oriented Bayesian network (OOBN) with state-test The new method of the system testing modeling and analysis of sensitivity index.The modeling method can clearly describing system failure and test Between correlation degree, reflect the hierarchical relationship of complication system.Intersection Entropy principle based on information theory proposes that state-test is closed Join sensitivity index, and provides calculation method.The index reflects the uncertain influence in complex electromechanical systems test, overcomes Measuring point based on Shannon entropy is judged the shortcomings that analysis method, and the model information that binding test modeling obtains makes inferences calculating, It can be used for the quantitative analysis of testability.Testability modeling and analysis are carried out to aircraft fuel system with the comprehensive analysis method, The result shows that the method and index that are proposed have practicability in Aviation ElctroMechanical system DFT.In addition to above-mentioned testability modeling side Outside method, fuzzy diagnosis, petri model based on machine learning etc. are also used in testability modeling by people.
Above-mentioned testability modeling method all considers the uncertain problem of side test from different aspect, and passes through various intelligence Can model solve the problems, such as that but they all ignore the uncertain problem of test point.In systems in practice, test point has not been Complete reliable, because it can break down, its power of test is not infinitely that the power of test of each test point is also variant 's.
For test point uncertain problem, set forth herein a kind of probabilistic testability modeling sides of consideration test point Method.
Summary of the invention
In view of the above-mentioned problems, the purpose of the present invention is to provide a kind of probabilistic testability modeling sides of consideration test point Method.
In order to achieve the above object, technical solution proposed by the present invention are as follows:
1. a kind of probabilistic testability modeling method of consideration test point, it is characterised in that: this method is based on correlation Model construction correlation matrix is illustrated to carry out testability modeling；Any system can be described using dependency graph representation model The correlative relationship of each building block and each test point, any dependency graph representation model is all nothing more than cascaded structure, parallel-connection structure Or the mixed structure connected and composed in parallel by (or being equivalent to)；This method thinks test point it is possible that failure, therefore surveys The power of test of pilot is not exclusively reliable, and which results in the uncertainties of test point；It, may when test point itself breaks down It will appear first failure Times second failure, faulty do not report but failure or given result that cannot correspond to any one to survey That is, there is the second class false alarm condition, undetectable situation and untrustworthy situation in the normal instruction state of pilot in the event of failure. Include following four step when this method progress testability modeling:
Step 1: after describing system using dependency graph representation model, find each according to the historical summary of system The failure rate of the corresponding failure rate of component and corresponding test point；
Step 2: the undetectable rate of system of the dependency graph representation model, the second class false alarm rate and untrustworthy are calculated Rate；
Step 3: corresponding undetectable influence when each test point is not known is found out in the dependency graph representation model Degree, false-alarm disturbance degree, untrustworthy disturbance degree and system risk disturbance degree, and finally find out its detectable disturbance degree, anti-void Alert disturbance degree, believable disturbance degree and devoid of risk disturbance degree；
Step 4: establishing test correlation matrix when test point is not known, and such as detectable disturbance degree correlation matrix is prevented False-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree correlation matrix.
Wherein, " dependency graph representation model " described in step 1 refers to the diagram expression side of one of testability correlation Method.It is on the basis of functional block diagram, to clearly show functional information stream direction after UUT function and structure classifying rationally It with the interconnected relationship of each building block, and marks and understands position and the number of initial testing point, show each to form portion with this The correlative relationship of part and each test point, as shown in Figure of description 2,3,4,5,6.Wherein: box represents each building block, Circle represents test point, and arrow shows the direction of functional information transmitting.
Wherein, " component " described in step 1 refers to the functional unit in system, it can be component, components, group Part, equipment, subsystem etc..It is any of component, components, component, equipment, subsystem etc. on earth as it, depends on How the division of function and structure is carried out to UUT.
Wherein, " the undetectable rate of system, the second class false alarm rate of the dependency graph representation model are calculated described in step 2 And untrustworthy rate ", this is in the dependency graph representation model of parallel-connection structure (such as Figure of description 4) and actually only needs to count The undetectable rate of calculation system；And the undetectable rate of system, the second class false-alarm in the dependency graph representation model of non-parallel-connection structure Rate and untrustworthy rate " requires to calculate.Because being not in false alarm condition in the dependency graph representation model of parallel-connection structure With untrustworthy situation, it is only possible to undetectable situation occur.
Wherein, " the undetectable rate of system, the second class false alarm rate of the dependency graph representation model are calculated described in step 2 And untrustworthy rate ", calculation method is as follows:
(1) if dependency graph representation model is the cascaded structure of n component composition
1. undetectable situation:
If certain component malfunction, corresponding to the corresponding test point of test point and its components downstream also all occur therefore Barrier, the failure of the component is undetectable at this time.Such as (see specification in the dependency graph representation model of 3 components composition Attached drawing 2), if U2 failure, only if T2 failure, the failure of U2 remains to detect, because T3 will call the police；If but T2, T3 All failures, then 3 test points not will call the police, and the failure of U2 cannot detect, and here it is undetectable situations.
If saying that the probability of malfunction of 3 components is respectively λU1, λU2, λU3, and the probability of malfunction of 3 test points is λT1, λT2, λT3.The then undetectable rate (UFDR) of system are as follows:
UFDR=λU1λT1λT2λT3+λU2λT2λT3+λU3λT3
To sum up analysis is as can be seen that the undetectable condition of its system of the dependency graph representation model of cascaded structure generation is: Some component malfunction in system, then from the component, all signal stream downstreams test point includes the corresponding test of the component Point all failures, then the failure is undetectable.
So for the dependency graph representation model of cascaded structure, it is assumed that part count is n, and test point number is also n, The then undetectable rate (UFDR of systemS):
In formula: UFDRSThe undetectable rate of finger system, λUiRefer to the failure rate of i-th of component, λTjRefer to the event of j-th of test point Barrier rate
2. the second class false alarm condition:
Do not consider that no failure but reports out of order this false alarm condition.In Figure of description 2, if U1 failure, Test point T1 also lucky failure, and T2, T3 be it is normal, then can normally indicate the failure of U1.T2, T3 are indicated under normal circumstances What is indicated is U2 failure, and U1 failure should be that T1, T2, T3 are indicated.It should suspect U1 failure, suspect now and arrived on U2 head, Think that failure has occurred in U2, thus cause the second class false-alarm, is i.e. generation first failure but quotes second failure.
For dependency graph representation model shown in Figure of description 2, if U1 failure, T1, T2 all failures can also be with, But T3 is unable to failure, otherwise just at the non-detectable situation of failure.Certainly, if U3 failure, T3 also failure, Belong to undetectable situation.
In fact, judging false-alarm method in the dependency graph representation model of all cascaded structures are as follows: in non-least significant end When component malfunction:
The corresponding test point of A most end end pieces cannot centainly break down；
Its corresponding test point of the B component centainly breaks down, or from the corresponding test point of the component in signal stream Continuous several test points (containing itself) are played all to break down.Two conditions all meet just can be with.
Thus it releases, for the dependency graph representation model of cascaded structure, it is assumed that part count is n, test point number It is also n, then the false alarm rate index (FAR of systemS) are as follows:
In formula: FARSFinger system false alarm rate index, λUiRefer to the failure rate of i-th of component, λTjRefer to the failure of j-th of test point Rate
3. untrustworthy situation:
It is so-called it is untrustworthy refer to when the lucky failure of test point, given result cannot correspond to any one survey The normal instruction state of pilot in the event of failure.Such as (see explanation in the cascaded structure dependency graph representation model of 3 components composition Book attached drawing 2), if T1, T3 failure when U1 failure, but T2 is worked normally, what does not all illustrate for result at this time, because of this person This diagnostic result " cannot be accepted and believed ".But untrustworthy rate is directly relatively difficult because it need by it is all not A possibility that additional state in normal condition circle, is cumulative.Therefore it needs by indirectly calculating.
We provide the abnormal rate of system below.So-called normality, be exactly otherwise system unit all without failure or go out Show failure and is correctly indicated.In addition to this state is referred to as abnormal.Under single fault hypothesis, event occurs in some component When barrier, all test points (including oneself) positioned at signal stream downstream are as long as have more than or equal to 1 since its corresponding test point It is a to break down, it is taken as abnormal.(according to correlation models, the test point of upstream is unrelated, institute with this object Not within limit of consideration.) therefore abnormal including undetectable situation, the second class false alarm condition, untrustworthy situation.
It can release in the cascaded structure correlation diagram flow model of n component composition, the abnormal rate of system (ANSRS) are as follows:
In formula: ANSRSThe abnormal rate of finger system, λUiRefer to the failure rate of i-th of component, λTkRefer to the event of k-th of test point Barrier rate
Obviously, (3) are not equal to (1)+(2), and extra part corresponds to the untrustworthy rate CNRR of systemS, i.e.,
CNRRS=ANSRS-UFDRS-FARS (4)
In formula: CNRRSThe untrustworthy rate of finger system, ANSRSThe abnormal rate of finger system, UFDRSFinger system is undetectable Rate, FARSFinger system false alarm rate index.
(2) if dependency graph representation model is the parallel-connection structure of n component composition
In the parallel-connection structure dependency graph representation model of n component composition, the component of any branch road is surveyed with corresponding When pilot all breaks down, the appearance of undetectable situation is all only resulted in.In Figure of description 4, any part Ui and survey Can all occur undetectable situation when pilot Ti breaks down.So for the parallel signal flow model of n component composition, Undetectable rate (the UFDR of systemS):
In formula: UFDRSThe undetectable rate of finger system, λUiRefer to the failure rate of i-th of component, λTiRefer to the event of i-th of test point Barrier rate.
(3) if dependency graph representation model is the mixed structure by connecting with composing in parallel, and there is n component
1. undetectable situation:
For the mixed structure dependency graph representation model of n component composition, if certain component breaks down and cannot all be believed Number stream on test point detect, then will appear undetectable situation.
There are many kinds of structures for the mixed structure dependency graph representation model formed due to n component, so its system can not be examined Survey rate UFDRSIt to be found out in conjunction with specific dependency graph representation model.Seeking the undetectable rate UFDR of systemSWhen need it is comprehensive simultaneously It is coupled the method for solving of structure dependency graph representation model and cascaded structure dependency graph representation model.
2. the second class false alarm condition:
Second class false-alarm occurs first failure and but quotes second failure.
For the mixed structure dependency graph representation model of n component composition, due to its various structures, it is therefore desirable in conjunction with tool The dependency graph representation model of body finds out its system false alarm rate index (FDRS).Seeking system false alarm rate index FARSWhen need it is comprehensive Close the method for solving of parallel-connection structure dependency graph representation model and cascaded structure dependency graph representation model.
When there is false-alarm, following two situation centainly will appear:
A centainly has at least one test point normal work in the end of signal stream
Its corresponding test point of the B component centainly breaks down
3. untrustworthy situation:
The untrustworthy rate of system, correlation of the method with cascaded structure are found out in the dependency graph representation model of mixed structure Property diagram model and the method for dependency graph representation model of parallel-connection structure be the same, be all by finding out the abnormal of system Rate seeks the untrustworthy rate of its system indirectly.
For the mixed structure dependency graph representation model of n component composition, if having found out the abnormal rate of its system (ANSRS), the undetectable rate UFDR of systemSWith system false alarm rate index (FARS), then its corresponding untrustworthy rate of system CNRRSIt is as follows:
CNRRS=ANSRS-UFDRS-FARS
In formula: CNRRSThe untrustworthy rate of finger system, ANSRSThe abnormal rate of finger system, UFDRSFinger system is undetectable Rate, FARSFinger system false alarm rate index.
Wherein, " undetectable disturbance degree " described in step 3, " false-alarm disturbance degree ", " untrustworthy disturbance degree ", point To system false alarm rate index when not referring to when testing node failure to the influence degree of the undetectable rate of system, test node failure To the influence degree of the untrustworthy rate of system when influence degree, test node failure.Disturbance degree illustrates test point to testability The influence degree of energy, reflects the test point in the case where existing dependency graph representation model and dependability parameter, influences system The weight of undetectable rate, false alarm rate and untrustworthy rate.
Wherein, " system risk disturbance degree " described in step 3, refer to shadow when testing node failure to system risk The degree of sound.Can be respectively to undetectable situation, false alarm condition, untrustworthy situation assigns different weighted values, to indicate to be System risk.Such as assume undetectable rate, false alarm rate, untrustworthy rate weight be respectively α, β, γ, then we set one A risk function L, then can order:
L=α × UFDRS+β×FARS+γ×CNRRS, alpha+beta+γ=1.
Wherein, " detectable disturbance degree " described in step 3, " anti-false-alarm disturbance degree ", " believable influence " degree and " nothing Venture influence degree ", respectively successively with " undetectable disturbance degree ", " false-alarm disturbance degree ", " untrustworthy disturbance degree ", " system wind Dangerous disturbance degree " is corresponding and the probability of both and be 1.The sum that can be detected disturbance degree and undetectable disturbance degree is 1, is prevented The sum of false-alarm disturbance degree and false-alarm disturbance degree is 1, and the sum of untrustworthy disturbance degree and believable influence is 1, devoid of risk disturbance degree With system risk disturbance degree and be also 1.
Wherein, " false-alarm disturbance degree " described in step 3, " untrustworthy disturbance degree ", " system risk disturbance degree ", " anti- False-alarm disturbance degree ", " believable disturbance degree " and " devoid of risk disturbance degree ", they in the dependency graph representation model of parallel-connection structure not It needs to solve.It only needs to solve " undetectable disturbance degree " and " detectable influence in the dependency graph representation model of parallel-connection structure Degree ", because only existing undetectable situation in the dependency graph representation model of parallel-connection structure, there is no the second class false alarm condition and not Believable situation.
Wherein, it " is found out in the dependency graph representation model corresponding when each test point is not known described in step 3 Undetectable disturbance degree, false-alarm disturbance degree, untrustworthy disturbance degree and system risk disturbance degree, and finally find out it and can examine Survey disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree ", method for solving is as follows:
According to relevant mathematical theory, following test point disturbance degree formula is defined.
(1) undetectable disturbance degree:
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point.
First by parameter UFDRSLocal derviation is sought, obtains its position disturbance degree, then passes through failure rate with itself again Product has obtained undetectable disturbance degree.
For the dependency graph representation model of n component composition, the undetectable disturbance degree of n test point can use row vector WFDIt is following to indicate:
WFD=[WFD1, WFD2, WFD3..., WFDn]
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point.
The detectable disturbance degree of n test pointAre as follows:
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point.
(2) false-alarm disturbance degree
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point.
First by parameter FARSLocal derviation is sought, obtains its position disturbance degree, then passes through failure rate with itself again Product has obtained test point false-alarm disturbance degree.
For the dependency graph representation model of n component composition, the false-alarm disturbance degree of n test point can use row vector WFASuch as Lower expression:
WFA=[WFA1, WFA2, WFA3..., WFAn]
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point.
The anti-false-alarm of n test point surveys disturbance degreeAre as follows:
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point.
(3) untrustworthy disturbance degree:
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point.
First by parameter CNRRSLocal derviation is sought, obtains its position disturbance degree, then passes through failure rate with itself again Product has obtained the untrustworthy disturbance degree of test point.
For the dependency graph representation model of n component composition, the untrustworthy disturbance degree of n test point can use row vector WCNIt is following to indicate:
WCN=[WCN1, WCN2, WCN3..., WCNn]
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point.
The believable disturbance degree of n test pointAre as follows:
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point.
(4) system risk disturbance degree
Assuming that undetectable rate, false alarm rate, untrustworthy rate weight are respectively α, β, γ, then we set a risk Spend function L.
Order L=α × UFDRS+β×FARS+γ×CNRRS, alpha+beta+γ=1.
At this time we can define test point system risk disturbance degree it is as follows:
In formula: WLiIndicate the system risk disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point.
For the dependency graph representation model of n component composition, the system risk disturbance degree of n test point can use row vector WLIt is following to indicate:
WL=[WL1, WL2, WL3..., WLn]
In formula: WLiIndicate the system risk disturbance degree of i-th of test point.
The devoid of risk disturbance degree of n test pointAre as follows:
In formula: WLiIndicate the system risk disturbance degree of i-th of test point.
In systems, it when there is a situation where undetectable, since we can not take counter-measure in advance, and thus leads The consequence of cause be also it is uncontrollable, the loss of generation is also maximum；In the case where untrustworthy, although we do not know it is specific which Failure has occurred in component, but the component finally to break down will be found out by various investigation means by having known faulty, this Although causing the waste on time and resource, bigger loss is avoided；For false alarm condition, due to the original of wrong report Cause, the component that we may hold disassemble, this has resulted in the waste of component.After part replacement, if continuing to show Indicating fault, then we just should be aware that possible false-alarm also can be true at this time by artificial detection and other various means Failure is made, this loss is between undetectable and untrustworthy.Therefore the value of α, β, γ should be α > β > γ and α + β+γ=1.
Wherein, described in step 4 " detectable disturbance degree correlation matrix ", " anti-false-alarm disturbance degree correlation matrix ", " believable disturbance degree correlation matrix " and " system risk disturbance degree correlation matrix ", they are all test point in essence Corresponding test correlation matrix when uncertain.Value not instead of Boolean in these correlation matrixes, between 0 and 1 it Between probability value.
Wherein, " anti-false-alarm disturbance degree correlation matrix ", " believable disturbance degree correlation matrix " described in step 4 With " system risk disturbance degree correlation matrix ", they do not need to solve in the dependency graph representation model of parallel-connection structure.Simultaneously It is coupled in the dependency graph representation model of structure and only needs to find out " detectable disturbance degree correlation matrix ", because of the correlation of parallel-connection structure Property diagram model in only exist undetectable situation, there is no the second class false alarm condition and untrustworthy situations.
Wherein, " test correlation matrix when test point is not known is established, such as detectable disturbance degree described in step 4 Correlation matrix, anti-false-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree phase Closing property matrix ", method for building up operation is as follows:
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, tested Correlation matrix D.
(2) by row vectorIt is multiplied respectively with correlation matrix D (premultiplication), respectively obtains detectable shadow Loudness tests correlation matrix DFD, anti-false-alarm disturbance degree test correlation matrix DFA, Reliability disturbance degree test correlation matrix DCN, devoid of risk disturbance degree test correlation matrix DL。
Detectable disturbance degree, which is obtained, by above-described four steps tests correlation matrix DFD, anti-false-alarm disturbance degree surveys Try correlation matrix DFA, Reliability disturbance degree test correlation matrix DCNCorrelation matrix D is tested with devoid of risk disturbance degreeLAfterwards, Intelligent algorithm can be used and carry out the preferred of test point, so that testability index is targetedly improved, optimal inspection design. Such as, if it is desired to verification and measurement ratio is higher, just should test correlation matrix D in detectable disturbance degreeFDOn the basis of tested That puts is preferred.It is lower if necessary to false alarm rate, it should to test correlation matrix D in anti-false-alarm disturbance degreeFAOn the basis of surveyed Pilot is preferred.If comprehensively considering overall risk, correlation matrix D just is tested in devoid of risk disturbance degreeLOn the basis of tested That puts is preferred.
Beneficial effect
The present invention, which compares prior art, following innovative point:
(1) consider that test is uncertain from test point not exclusively reliably
(2) detectable disturbance degree correlation matrix, anti-false-alarm disturbance degree correlation square are obtained based on test point uncertainty Battle array, believable disturbance degree correlation matrix and system risk disturbance degree correlation matrix.
The present invention compares prior art and has the advantage that
Testability modeling is carried out based on test point uncertainty, and thus to obtain detectable disturbance degree correlation matrix, anti- False-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree correlation matrix.At this The preferred of test point set is carried out on the basis of a little correlation matrixes, and it is (such as empty more targeted to improve certain testability indexes Alert rate).
Detailed description of the invention
Fig. 1 is modeling method flow chart of the present invention.
Fig. 2 is the cascaded structure dependency graph representation model being made of 3 components.
Fig. 3 is the cascaded structure dependency graph representation model being made of n component.
Fig. 4 is the parallel-connection structure dependency graph representation model being made of n component.
Fig. 5 is a kind of parallel-connection structure dependency graph representation model being made of 6 components.
Fig. 6 is a kind of mixed structure dependency graph representation model being made of 7 components.
Symbol description is as follows in figure: in Fig. 2,3,4,5,6, U1, U2, U3, U4, U5, U6, U7 ..., Un be all system Building block, T1, T2, T3, T4, T5, T6, T7 ..., Tn be all test point.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, right below in conjunction with the accompanying drawings and the specific embodiments The present invention is described in further detail.
(1) case 1
For the dependency graph representation model of certain avionics system local circuit as shown in Figure of description 2, which is 3 component groups At cascaded structure, be common diagnosis and signalling game model.Wherein each component distribution has a test point.
A kind of probabilistic testability modeling method of consideration test point of the present invention, as shown in Figure 1, implementation step is such as Under:
Step 1: according to historical summary, each component corresponds to failure rate and is shown in Table 1, and corresponding test point failure rate is shown in Table 2.
1 unit failure rate of table
U(i) U1 U2 U3
λ 0.3 0.3 0.3
2 test point failure rate of table
T(i) T1 T2 T3
λ 0.2 0.1 0.1
Step 2: the undetectable rate UFDR of system that dependency graph representation model is calculated according to Figure of description 2S, false-alarm Rate FARS, untrustworthy rate CNRRS。
CNRRS=ANSRS-UFDRS-FARS
Step 3: respectively to corresponding test point failure rate derivation, find out each test node failure when it is corresponding not Detectable disturbance degree, false-alarm disturbance degree, untrustworthy disturbance degree and system risk disturbance degree, and finally find out its detectable shadow Loudness, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree.
Undetectable disturbance degree, false-alarm disturbance degree and the untrustworthy disturbance degree of required test point see the table below 3:
3 disturbance degree index table of table
T(i) T1 T2 T3
WFDi 0.003 0.0036 0.336
WFA 0.27 0.27 0
WCN 0.006 0.018 0.025
According to analysis, the value of α, β, γ should be α > β > γ and alpha+beta+γ=1.If determining respectively, the value of α, β, γ are 0.5,0.3,0.2, then it is as shown in table 4 to calculate system risk disturbance degree.
4 system risk disturbance degree of table
T(i) T1 T2 T3
WL 0.0177 0.0135 0.0218
By table 3, the detectable disturbance degree of 4 available test points, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk Disturbance degree such as the following table 5.
Table 5 can detect disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree
Step 4: establishing test correlation matrix when test point is not known according to Figure of description 2, such as detectable to influence Spend correlation matrix, anti-false-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree Correlation matrix.Its operation is as follows:
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, tested Correlation matrix D is as shown in table 6.
Table 6 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3
T1 1 0 0
T2 1 1 0
T3 1 1 1
(2) by row vectorIt is multiplied respectively with correlation matrix D (premultiplication), respectively obtains detectable shadow Loudness tests correlation matrix DFD, anti-false-alarm disturbance degree test correlation matrix DFA, Reliability disturbance degree test correlation matrix DCN, devoid of risk disturbance degree test correlation matrix DLSuch as the following table 7,8,9,10.
Table 7 can detect disturbance degree test correlation matrix DFD
U1 U2 U3
T1 0.997 0 0
T2 0.9964 0.9964 0
T3 0.664 0.664 0.664
The anti-false-alarm disturbance degree of table 8 tests correlation matrix DFA
The believable disturbance degree of table 9 tests correlation matrix DCN
U1 U2 U3
T1 0.994 0 0
T2 0.982 0.982 0
T3 0.975 0.975 0.975
10 devoid of risk disturbance degree of table tests correlation matrix DL
U1 U2 U3
T1 0.9823 0 0
T2 0.9865 0.9865 0
T3 0.9782 0.9782 0.9782
After obtaining above-mentioned test correlation matrix, intelligent algorithm such as particle swarm algorithm can be used and carry out the excellent of test point Choosing.Such as, if it is desired to verification and measurement ratio is higher, just should test correlation matrix D in detectable disturbance degreeFDOn the basis of surveyed Pilot it is preferred.It is lower if necessary to false alarm rate, it should to test correlation matrix D in anti-false-alarm disturbance degreeFAOn the basis of carry out Test point is preferred.If comprehensively considering overall risk, correlation matrix D just is tested in devoid of risk disturbance degreeLOn the basis of surveyed Pilot it is preferred.
If the dependency graph representation model of certain avionics system local circuit as shown in Figure of description 3, is that n component forms Cascaded structure, then (such as specification is attached for the cascaded structure dependency graph representation model that its testability modeling process is formed with 3 components Fig. 2) be it is the same, the calculation amount only modeled will it is a little greatly for.
(2) case 2
The dependency graph representation model of certain avionics system local circuit is as shown in Fig. 5, the model be 6 components composition and It is coupled structure, is common diagnosis and signalling game model.Wherein each component distribution has a test point.
A kind of probabilistic testability modeling method of consideration test point of the present invention, as shown in Figure 1, implementation step is such as Under:
Step 1: according to historical summary, each component corresponds to failure rate and is shown in Table 11, and corresponding test point failure rate is shown in Table 12.
11 unit failure rate of table
U(i) U1 U2 U3 U4 U5 U6
λ 0.5 0.4 0.5 0.3 0.5 0.5
12 test point failure rate of table
T(i) T1 T2 T3 T4 T5 T6
λ 0.2 0.2 0.1 0.1 0.1 0.1
Step 2: the undetectable rate UFDR of system that dependency graph representation model is calculated according to Figure of description 5S
Step 3: respectively to corresponding test point failure rate derivation, find out each test node failure when it is corresponding not Detectable disturbance degree, and finally find out its detectable disturbance degree
The undetectable disturbance degree of required test point see the table below 13:
13 disturbance degree index table of table
T(i) T1 T2 T3 T4 T5 T6
WFDi 0.1 0.08 0.05 0.03 0.05 0.05
By detectable disturbance degree such as the following table 14 of the available test point of table 13.
Table 14 can detect disturbance degree
Step 4: establishing detectable disturbance degree when test point is not known according to Figure of description 5 and test correlation matrix, Its operation is as follows:
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, tested Correlation matrix D is as shown in Table 15.
Table 15 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3 U4 U5 U6
T1 1 0 0 0 0 0
T2 0 1 0 0 0 0
T3 0 0 1 0 0 0
T4 0 0 0 1 0 0
T5 0 0 0 0 1 0
T6 0 0 0 0 0 1
(2) by row vectorBe multiplied (premultiplication) with correlation matrix D, respectively obtains detectable disturbance degree test correlation square Battle array DFDSuch as the following table 16.
Table 16 can detect disturbance degree test correlation matrix DFD
U1 U2 U3 U4 U5 U6
T1 0.9 0 0 0 0 0
T2 0 0.92 0 0 0 0
T3 0 0 0.95 0 0 0
T4 0 0 0 0.97 0 0
T5 0 0 0 0 0.95 0
T6 0 0 0 0 0 0.95
Obtaining above-mentioned detectable disturbance degree test correlation matrix DFDAfterwards, intelligent algorithm such as particle swarm algorithm can be used The preferred of test point is carried out, the verification and measurement ratio obtained in this way can be higher.
If the dependency graph representation model of certain avionics system local circuit as shown in Figure of description 4, is that n component forms Parallel-connection structure, then the parallel-connection structure dependency graph representation model (Figure of description that its testability modeling process and 6 components form It 5) is the same.
(3) case 3
The dependency graph representation model of certain avionics system local circuit is as shown in Fig. 6, which is the mixed of 7 component compositions Structure is closed, is common diagnosis and signalling game model.Wherein each component distribution has a test point.
A kind of probabilistic testability modeling method of consideration test point of the present invention, as shown in Figure 1, implementation step is such as Under:
Step 1: according to historical summary, each component corresponds to failure rate and is shown in Table 17, and corresponding test point failure rate is shown in Table 18.
17 unit failure rate of table
U(i) U1 U2 U3 U4 U5 U6 U7
λ 0.5 0.5 0.5 0.5 0.5 0.5 0.5
18 test point failure rate of table
T(i) T1 T2 T3 T4 T5 T6 T7
λ 0.1 0.2 0.1 0.1 0.1 0.2 0.1
Step 2: the undetectable rate UFDR of system that dependency graph representation model is calculated according to Figure of description 6S, false-alarm Rate FARS, untrustworthy rate CNRRS。
CNRRS=ANSRS-UFDRS-FARS
Step 3: respectively to corresponding test point failure rate derivation, find out each test node failure when it is corresponding not Detectable disturbance degree, false-alarm disturbance degree, untrustworthy disturbance degree and system risk disturbance degree, and finally find out its detectable shadow Loudness, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree.
Undetectable disturbance degree, false-alarm disturbance degree and the untrustworthy disturbance degree of required test point see the table below 19:
19 disturbance degree index table of table
According to analysis, the value of α, β, γ should be α > β > γ and alpha+beta+γ=1.If determining respectively, the value of α, β, γ are 0.5,0.3,0.2, then it is as shown in table 20 to calculate system risk disturbance degree.
20 system risk disturbance degree of table
By table 19, the detectable disturbance degree of 20 available test points, anti-false-alarm disturbance degree, believable disturbance degree with it is calm Dangerous disturbance degree such as the following table 21.
Table 21 can detect disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree
According to the above Calculation results, we can adjust according to system requirements and test when designing testability system The selection of point set.For example, should then avoid the test point low using detectable disturbance degree when we need verification and measurement ratio higher T3, T4, T5, uses more as far as possible can detect high test point T1, T2, the T4 of disturbance degree.When we need false alarm rate lower, then keep away Exempt from test point T1, T6, the T7 low using anti-false-alarm disturbance degree, as often as possible using the high test point T2, T3 of anti-false-alarm disturbance degree, T4,T5.When we need police instruction confidence level higher, avoid selecting as far as possible the lower test point T3 of believable disturbance degree, T4, T6 etc., and preferentially select believable disturbance degree higher T1, T2, T5, T7.If in the case where considering overall risk, I Devoid of risk disturbance degree ought to be avoided relatively low several test points, for example, test point T2, T3 and T5.Preferentially select other nothings The higher several test points of risk.
Step 4: establishing test correlation matrix when test point is not known according to Figure of description 6, such as detectable to influence Spend correlation matrix, anti-false-alarm disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree Correlation matrix.Its operation is as follows:
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, tested Correlation matrix D is as shown in table 22.
Table 22 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3 U4 U5 U6 U7
T1 1 0 0 0 0 0 0
T2 1 1 0 0 0 0 0
T3 1 1 1 0 0 0 0
T4 1 1 0 1 0 0 0
T5 0 0 0 0 1 1 1
T6 0 0 0 0 0 1 1
T7 0 0 0 0 0 0 1
(2) by row vectorIt is multiplied respectively with correlation matrix D (premultiplication), respectively obtains detectable shadow Loudness tests correlation matrix DFD, anti-false-alarm disturbance degree test correlation matrix DFA, Reliability disturbance degree test correlation matrix DCN, devoid of risk disturbance degree test correlation matrix DLSuch as the following table 23,24,25,26.
Table 23 can detect disturbance degree test correlation matrix DFD
U1 U2 U3 U4 U5 U6 U7
T1 0.9999 0 0 0 0 0 0
T2 0.9988 0.9988 0 0 0 0 0
T3 0.9489 0.9489 0.9489 0 0 0 0
T4 0.9489 0.9489 0 0.9489 0 0 0
T5 0 0 0 0 0.944 0.944 0.944
T6 0 0 0 0 0 0.989 0.989
T7 0 0 0 0 0 0 0.999
The anti-false-alarm disturbance degree of table 24 tests correlation matrix DFA
U1 U2 U3 U4 U5 U6 U7
T1 0.965 8 0 0 0 0 0 0
T2 0.9939 0.9939 0 0 0 0 0
T3 0.9972 0.9972 0.9972 0 0 0 0
T4 0.9972 0.9972 0 0.9972 0 0 0
T5 0 0 0 0 0.99 0.99 0.99
T6 0 0 0 0 0 0.91 0.91
T7 0 0 0 0 0 0 0.955
The believable disturbance degree of table 25 tests correlation matrix DCN
26 devoid of risk disturbance degree of table tests correlation matrix DL
U1 U2 U3 U4 U5 U6 U7
T1 0.99815 0 0 0 0 0 0
T2 0.96911 0.96911 0 0 0 0 0
T3 0.96911 0.96911 0.96911 0 0 0 0
T4 0.97182 0.97182 0 0.97182 0 0 0
T5 0 0 0 0 0.9542 0.9542 0.9542
T6 0 0 0 0 0 0.9807 0.9807
T7 0 0 0 0 0 0 0.985614
After obtaining above-mentioned test correlation matrix, intelligent algorithm such as particle swarm algorithm can be used and carry out the excellent of test point Choosing.Such as, if it is desired to verification and measurement ratio is higher, just should test correlation matrix D in detectable disturbance degreeFDOn the basis of surveyed Pilot it is preferred.It is lower if necessary to false alarm rate, it should to test correlation matrix D in anti-false-alarm disturbance degreeFAOn the basis of carry out Test point is preferred.If comprehensively considering overall risk, correlation matrix D just is tested in devoid of risk disturbance degreeLOn the basis of surveyed Pilot it is preferred.
In conclusion protection of the invention cannot be considered as limiting the above is only present pre-ferred embodiments Range, protection scope of the present invention are limited by appended claims.It is all to be done on the basis of spirit of that invention and scheme Any modification and improvement etc. should be all included within protection scope of the present invention.

Claims (8)

1. a kind of probabilistic testability modeling method of consideration test point, this method are applied in avionics system local circuit, It is characterized by: this method includes following four step:
Step 1: after describing system using dependency graph representation model, each component is found according to the historical summary of system The failure rate of corresponding failure rate and corresponding test point；The model is the cascaded structure of 3 components composition, is diagnosis and signal TRANSFER MODEL；Wherein each component distribution has a test point；Each component corresponds to failure rate and is shown in Table 1, accordingly tests point failure Rate is shown in Table 2；
1 unit failure rate of table
U(i) U1 U2 U3 λ 0.3 0.3 0.3
2 test point failure rate of table
T(i) T1 T2 T3 λ 0.2 0.1 0.1
Step 2: the undetectable rate UFDR of system of the dependency graph representation model is calculatedS, the second class false alarm rate FARSAnd it can not Trust rate CNRRS, it specifically includes:
1. undetectable situation:
Undetectable condition occurs for its system of the dependency graph representation model of cascaded structure: some component malfunction in system , then from the component, all signal stream downstreams test point includes that the component corresponds to test point all failures, then the failure is nothing Method detection；
For the dependency graph representation model of cascaded structure, it is assumed that part count is n, and test point number is also n, then system is not Detectable rate, that is, UFDRSAre as follows:
In formula: UFDRSThe undetectable rate of finger system, λUiRefer to the failure rate of i-th of component, λTjRefer to the failure rate of j-th of test point；
2. the second class false alarm condition:
Judge false-alarm method are as follows: when component malfunction in non-least significant end:
A, the corresponding test point of most end end pieces cannot centainly break down；
B, its corresponding test point of the component centainly breaks down, and lights continuously in signal stream from the corresponding test of the component Several test points include itself all break down, two conditions all meet；
For the dependency graph representation model of cascaded structure, it is assumed that part count is n, test point number is also n, then system False alarm rate index, that is, FARSAre as follows:
In formula: FARSFinger system false alarm rate index, λUiRefer to the failure rate of i-th of component, λTjRefer to the failure rate of j-th of test point；
3. untrustworthy situation:
Untrustworthy to refer to when the lucky failure of test point, given result, which cannot correspond to any one, has event in test point Normal instruction state when barrier；So-called normality, exactly or system unit is all referred to without failure or failure and correctly Show, state in addition to this is referred to as abnormal；It is corresponding from its in some component malfunction under single fault hypothesis Test point starts all test points positioned at signal stream downstream, includes oneself, breaks down, recognize more than or equal to 1 as long as having To be abnormal；Abnormal includes undetectable situation, the second class false alarm condition, untrustworthy situation；
In the cascaded structure correlation diagram flow model of n component composition, abnormal rate, that is, ANSR of systemSAre as follows:
In formula: ANSRSThe abnormal rate of finger system, λUiRefer to the failure rate of i-th of component, λTkRefer to the failure rate of k-th of test point；
CNRRS=ANSRS-UFDRS-FARS (4)
In formula: CNRRSThe untrustworthy rate of finger system, ANSRSThe abnormal rate of finger system, UFDRSThe undetectable rate of finger system, FARSFinger system false alarm rate index；
Step 3: corresponding undetectable disturbance degree when each test point is not known is found out in the dependency graph representation model WFDi, false-alarm disturbance degree WFA, untrustworthy disturbance degree WCNAnd system risk disturbance degree WL, and finally find out its detectable influence DegreeAnti- false-alarm disturbance degreeBelievable disturbance degreeWith devoid of risk disturbance degree
Undetectable disturbance degree, false-alarm disturbance degree and the untrustworthy disturbance degree of required test point see the table below 3:
3 disturbance degree index table of table
T(i) T1 T2 T3 WFDi 0.003 0.0036 0.336 WFA 0.27 0.27 0 WCN 0.006 0.018 0.025
According to analysis, it is assumed that undetectable rate loudness, false alarm rate loudness, untrustworthy rate loudness weight be respectively α, β, γ Value is α > β > γ and alpha+beta+γ=1；The value for determining α, β, γ respectively is 0.5,0.3,0.2, then calculates system risk shadow Loudness is as shown in table 4；
4 system risk disturbance degree of table
T(i) T1 T2 T3 WL 0.0177 0.0135 0.0218
It is influenced by table 3, the detectable disturbance degree of 4 available test points, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk Degree such as the following table 5；
Table 5 can detect disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree
Step 4: test correlation matrix when test point is not known, including detectable disturbance degree correlation matrix, anti-void are established Alert disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree correlation matrix；
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, it is related to obtain test Property matrix D is as shown in table 6；
Table 6 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3 T1 1 0 0 T2 1 1 0 T3 1 1 1
(2) by row vectorIt is multiplied respectively with correlation matrix D, respectively obtains detectable disturbance degree test phase Closing property matrix DFD, anti-false-alarm disturbance degree test correlation matrix DFA, Reliability disturbance degree test correlation matrix DCN, devoid of risk Disturbance degree tests correlation matrix DLSuch as the following table 7,8,9,10；
Table 7 can detect disturbance degree test correlation matrix DFD
U1 U2 U3 T1 0.997 0 0 T2 0.9964 0.9964 0 T3 0.664 0.664 0.664
The anti-false-alarm disturbance degree of table 8 tests correlation matrix DFA
U1 U2 U3 T1 0.73 0 0 T2 0.73 0.73 0 T3 1 1 1
The believable disturbance degree of table 9 tests correlation matrix DCN
U1 U2 U3 T1 0.994 0 0 T2 0.982 0.982 0 T3 0.975 0.975 0.975
10 devoid of risk disturbance degree of table tests correlation matrix DL
Detectable disturbance degree, which is obtained, by above four steps tests correlation matrix DFD, anti-false-alarm disturbance degree test correlation square Battle array DFA, Reliability disturbance degree test correlation matrix DCNCorrelation matrix D is tested with devoid of risk disturbance degreeLAfterwards, using intelligent calculation Method carries out the preferred of test point.
2. a kind of probabilistic testability modeling method of consideration test point, this method are applied in avionics system local circuit, It is characterized by: this method includes following four step:
Step 1: after describing system using dependency graph representation model, each component is found according to the historical summary of system The failure rate of corresponding failure rate and corresponding test point；The model is the parallel-connection structure of 6 components composition, is diagnosis and signal TRANSFER MODEL；Wherein each component distribution has a test point；Each component corresponds to failure rate and is shown in Table 11, accordingly tests point failure Rate is shown in Table 12；
11 unit failure rate of table
U(i) U1 U2 U3 U4 U5 U6 λ 0.5 0.4 0.5 0.3 0.5 0.5
12 test point failure rate of table
T(i) T1 T2 T3 T4 T5 T6 λ 0.2 0.2 0.1 0.1 0.1 0.1
Step 2: the undetectable rate UFDR of system of the dependency graph representation model is calculatedS；
In formula: UFDRSThe undetectable rate of finger system, λUiRefer to the failure rate of i-th of component, λTiRefer to the failure rate of i-th of test point； Step 3: corresponding undetectable disturbance degree W when each test point is not known is found out in the dependency graph representation modelFDi, and Finally find out its detectable disturbance degreeThe undetectable disturbance degree of test point see the table below 13:
13 disturbance degree index table of table
T(i) T1 T2 T3 T4 T5 T6 WFDi 0.1 0.08 0.05 0.03 0.05 0.05
By detectable disturbance degree such as the following table 14 of the available test point of table 13；
Table 14 can detect disturbance degree
Step 4: detectable disturbance degree correlation matrix D when test point is not known is establishedFD；
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, it is related to obtain test Property matrix D is as shown in Table 15；
Table 15 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3 U4 U5 U6 T1 1 0 0 0 0 0 T2 0 1 0 0 0 0 T3 0 0 1 0 0 0 T4 0 0 0 1 0 0 T5 0 0 0 0 1 0 T6 0 0 0 0 0 1
(2) by row vectorIt is multiplied with correlation matrix D, respectively obtains detectable disturbance degree test correlation matrix DFDSuch as following table 16；
Table 16 can detect disturbance degree test correlation matrix DFD
Detectable disturbance degree, which is obtained, by above four steps tests correlation matrix DFDAfterwards, test point is carried out using intelligent algorithm It is preferred.
3. a kind of probabilistic testability modeling method of consideration test point, this method are applied in avionics system local circuit, It is characterized by: this method includes following four step:
Step 1: after describing system using dependency graph representation model, each component is found according to the historical summary of system The failure rate of corresponding failure rate and corresponding test point；The model is the mixed structure of 7 components composition, is diagnosis and signal TRANSFER MODEL；Wherein each component distribution has a test point；Each component corresponds to failure rate and is shown in Table 17, accordingly tests point failure Rate is shown in Table 18；
17 unit failure rate of table
U(i) U1 U2 U3 U4 U5 U6 U7 λ 0.5 0.5 0.5 0.5 0.5 0.5 0.5
18 test point failure rate of table
T(i) T1 T2 T3 T4 T5 T6 T7 λ 0.1 0.2 0.1 0.1 0.1 0.2 0.1
Step 2: the undetectable rate UFDR of system of the dependency graph representation model is calculatedS, the second class false alarm rate FARSAnd it can not Trust rate CNRRS, it specifically includes:
1. undetectable situation:
For the mixed structure dependency graph representation model of n component composition, if certain component breaks down and all cannot be by signal stream Test point detect, then will appear undetectable situation；
For the mixed structure dependency graph representation model of n component composition, the undetectable rate UFDR of systemSTo combine specific phase Closing property illustrates model to find out；Seeking the undetectable rate UFDR of systemSWhen need comprehensive parallel-connection structure dependency graph representation model with The method for solving of cascaded structure dependency graph representation model；
2. the second class false alarm condition:
Second class false-alarm occurs first failure and but quotes second failure；
For the mixed structure dependency graph representation model of n component composition, need to find out in conjunction with specific dependency graph representation model Its system false alarm rate index, that is, FARS；Seeking system false alarm rate index FARSWhen need comprehensive parallel-connection structure dependency graph representation model With the method for solving of cascaded structure dependency graph representation model；
When there is false-alarm, following two situation centainly will appear:
A, centainly there is at least one test point normal work in the end of signal stream
B, its corresponding test point of the component centainly breaks down
3. untrustworthy situation:
For the mixed structure dependency graph representation model of n component composition, if having found out the i.e. ANSR of the abnormal rate of its systemS, system Undetectable rate UFDRSWith system false alarm rate index, that is, FARS, then its corresponding untrustworthy rate CNRR of systemSIt is as follows:
CNRRS=ANSRS-UFDRS-FARS
In formula: CNRRSThe untrustworthy rate of finger system, ANSRSThe abnormal rate of finger system, UFDRSThe undetectable rate of finger system, FARSFinger system false alarm rate index；Step 3: it is found out in the dependency graph representation model corresponding when each test point is not known Undetectable disturbance degree WFDi, false-alarm disturbance degree WFA, untrustworthy disturbance degree WCNAnd system risk disturbance degree WL, and it is final Find out its detectable disturbance degreeAnti- false-alarm disturbance degreeBelievable disturbance degreeWith devoid of risk disturbance degree
Undetectable disturbance degree, false-alarm disturbance degree and the untrustworthy disturbance degree of required test point see the table below 19:
19 disturbance degree index table of table
The value of α, β, γ are α > β > γ and alpha+beta+γ=1；The value for determining α, β, γ respectively is 0.5,0.3,0.2, then calculates System risk disturbance degree is as shown in table 20 out；
20 system risk disturbance degree of table
T(i) T1 T2 T3 T4 T5 T6 T7 WL 0.001844 0.030886 0.030886 0.02818 0.0458 0.0193 0.014386
By table 19, the detectable disturbance degree of 20 available test points, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk shadow Loudness such as the following table 21；
Table 21 can detect disturbance degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree
Step 4: test correlation matrix when test point is not known, including detectable disturbance degree correlation matrix, anti-void are established Alert disturbance degree correlation matrix, believable disturbance degree correlation matrix and system risk disturbance degree correlation matrix；
(1) do not consider in the case where testing point reliability, according to the direct corresponding relationship of failure and test point, it is related to obtain test Property matrix D is as shown in table 22；
Table 22 do not consider test point reliability in the case where failure and test homography D
U1 U2 U3 U4 U5 U6 U7 T1 1 0 0 0 0 0 0 T2 1 1 0 0 0 0 0 T3 1 1 1 0 0 0 0 T4 1 1 0 1 0 0 0 T5 0 0 0 0 1 1 1 T6 0 0 0 0 0 1 1 T7 0 0 0 0 0 0 1
(2) by row vectorIt is multiplied respectively with correlation matrix D, respectively obtains detectable disturbance degree test phase Closing property matrix DFD, anti-false-alarm disturbance degree test correlation matrix DFA, Reliability disturbance degree test correlation matrix DCN, devoid of risk Disturbance degree tests correlation matrix DLSuch as the following table 23,24,25,26；
Table 23 can detect disturbance degree test correlation matrix DFD
U1 U2 U3 U4 U5 U6 U7 T1 0.9999 0 0 0 0 0 0 T2 0.9988 0.9988 0 0 0 0 0 T3 0.9489 0.9489 0.9489 0 0 0 0 T4 0.9489 0.9489 0 0.9489 0 0 0 T5 0 0 0 0 0.944 0.944 0.944 T6 0 0 0 0 0 0.989 0.989 T7 0 0 0 0 0 0 0.999
The anti-false-alarm disturbance degree of table 24 tests correlation matrix DFA
U1 U2 U3 U4 U5 U6 U7 T1 0.965 8 0 0 0 0 0 0 T2 0.9939 0.9939 0 0 0 0 0 T3 0.9972 0.9972 0.9972 0 0 0 0 T4 0.9972 0.9972 0 0.9972 0 0 0 T5 0 0 0 0 0.99 0.99 0.99 T6 0 0 0 0 0 0.91 0.91 T7 0 0 0 0 0 0 0.955
The believable disturbance degree of table 25 tests correlation matrix DCN
U1 U2 U3 U4 U5 U6 U7 T1 0.97962 0 0 0 0 0 0 T2 0.98463 0.98463 0 0 0 0 0 T3 0.96912 0.96912 0.96912 0 0 0 0 T4 0.96912 0.96912 0 0.96912 0 0 0 T5 0 0 0 0 0.9841 0.9841 0.9841 T6 0 0 0 0 0 0.9335 0.9335 T7 0 0 0 0 0 0 0.9735
26 devoid of risk disturbance degree of table tests correlation matrix DL
U1 U2 U3 U4 U5 U6 U7 T1 0.99815 0 0 0 0 0 0 T2 0.96911 0.96911 0 0 0 0 0 T3 0.96911 0.96911 0.96911 0 0 0 0 T4 0.97182 0.97182 0 0.97182 0 0 0 T5 0 0 0 0 0.9542 0.9542 0.9542 T6 0 0 0 0 0 0.9807 0.9807 T7 0 0 0 0 0 0 0.985614
Detectable disturbance degree, which is obtained, by above four steps tests correlation matrix DFD, anti-false-alarm disturbance degree test correlation square Battle array DFA, Reliability disturbance degree test correlation matrix DCNCorrelation matrix D is tested with devoid of risk disturbance degreeLAfterwards, using intelligent calculation Method carries out the preferred of test point.
4. a kind of probabilistic testability modeling method of consideration test point according to claim 1 or 2 or 3, feature It is: refers to the diagram representation method of one of testability correlation at " dependency graph representation model " described in step 1, Dependency graph representation model is on the basis of functional block diagram, to clearly show that function is believed after UUT function and structure classifying rationally The interconnected relationship in breath stream direction and each building block, and position and the number for understanding initial testing point are marked, shown with this The correlative relationship of each building block and each test point；Refer to the functional unit in system at " component " described in step 1, Component is component, components, component, equipment, subsystem；Based on the division for carrying out function and structure to UUT, determine that component arrives Bottom is component, components, component, equipment, any in subsystem.
5. a kind of probabilistic testability modeling method of consideration test point according to claim 1 or 2 or 3, feature Be: " undetectable disturbance degree ", " false-alarm disturbance degree ", " untrustworthy disturbance degree " described in step 3 respectively refer to survey To the influence journey of system false alarm rate index when pilot is unreliable to the influence degree of the undetectable rate of system, test node failure To the influence degree of the untrustworthy rate of system when degree, test node failure；Disturbance degree illustrates test point to the shadow of test performance The degree of sound, reflects the test point in the case where existing dependency graph representation model and dependability parameter, the system of influence can not be examined The weight of survey rate, false alarm rate and untrustworthy rate；" system risk disturbance degree " described in step 3, refers to test point not , can be respectively to undetectable situation to the influence degree of system risk when reliable, false alarm condition, untrustworthy situation imparting is not With weighted value, to indicate system risk, it is assumed that undetectable rate, false alarm rate, untrustworthy rate weight be respectively α, β, γ, Then a risk function L is set, then
L=α × UFDRS+β×FARS+γ×CNRRS, alpha+beta+γ=1.
6. a kind of probabilistic testability modeling method of consideration test point according to claim 1 or 2 or 3, feature It is: " detectable disturbance degree ", " anti-false-alarm disturbance degree ", " believable influence " degree and " devoid of risk shadow described in step 3 Loudness ", respectively successively with " undetectable disturbance degree ", " false-alarm disturbance degree ", " untrustworthy disturbance degree ", " system risk influence Corresponding and the two probability of degree " and be 1；The sum that can be detected disturbance degree and undetectable disturbance degree is 1, anti-false-alarm disturbance degree Sum with false-alarm disturbance degree is 1, and the sum of untrustworthy disturbance degree and believable influence is 1, devoid of risk disturbance degree and system risk Disturbance degree and be also 1.
7. a kind of probabilistic testability modeling method of consideration test point according to claim 1 or 2 or 3, feature Be: described in step 3 " found out in the dependency graph representation model each test point it is uncertain when it is corresponding can not Disturbance degree, false-alarm disturbance degree, untrustworthy disturbance degree and system risk disturbance degree are detected, and finally finds out its detectable influence Degree, anti-false-alarm disturbance degree, believable disturbance degree and devoid of risk disturbance degree ", method for solving is as follows:
According to relevant mathematical theory, following test point disturbance degree formula is defined；
(1) undetectable disturbance degree:
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point；
First by parameter UFDRSLocal derviation is sought, obtains its position disturbance degree, then passes through the product of failure rate with itself again, Undetectable disturbance degree is obtained；
For the dependency graph representation model of n component composition, the undetectable disturbance degree of n test point can use row vector WFDSuch as following table Show:
WFD=[WFD1, WFD2, WFD3..., WFDn]
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point；
The detectable disturbance degree of n test pointAre as follows:
In formula: WFDiIndicate the undetectable disturbance degree of i-th of test point；
(2) false-alarm disturbance degree
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point；
First by parameter FARSLocal derviation is sought, obtains its position disturbance degree, then passes through the product of failure rate with itself again, Test point false-alarm disturbance degree is obtained；
For the dependency graph representation model of n component composition, the false-alarm disturbance degree of n test point can use row vector WFAIt is following to indicate:
WFA=[WFA1, WFA2, WFA3..., WFAn]
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point；
The anti-false-alarm of n test point surveys disturbance degreeAre as follows:
In formula: WFAiIndicate the false-alarm disturbance degree of i-th of test point；
(3) untrustworthy disturbance degree:
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point,
First by parameter CNRRSLocal derviation is sought, obtains its position disturbance degree, then passes through the product of failure rate with itself again, The untrustworthy disturbance degree of test point is obtained；
For the dependency graph representation model of n component composition, the untrustworthy disturbance degree row vector W of n test pointCNSuch as following table Show:
WCN=[WCN1, WCN2, WCN3..., WCNn]
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point；
The believable disturbance degree of n test pointAre as follows:
In formula: WCNiIndicate the untrustworthy disturbance degree of i-th of test point；
(4) system risk disturbance degree
Assuming that undetectable rate, false alarm rate, untrustworthy rate weight are respectively α, β, γ, a risk function L is then set；
Order L=α × UFDRS+β×FARS+γ×CNRRS, alpha+beta+γ=1,
The system risk disturbance degree for defining test point at this time is as follows:
In formula: WLiIndicate the system risk disturbance degree of i-th of test point, λTiIndicate the failure rate of i-th of test point；
For the dependency graph representation model of n component composition, the system risk disturbance degree of n test point can use row vector WLSuch as following table Show:
WL=[WL1, WL2, WL3..., WLn]
In formula: WLiIndicate the system risk disturbance degree of i-th of test point；
The devoid of risk disturbance degree of n test pointAre as follows:
In formula: WLiIndicate the system risk disturbance degree of i-th of test point；
The value of α, β, γ are α > β > γ and alpha+beta+γ=1.
8. a kind of probabilistic testability modeling method of consideration test point according to claim 1 or 2 or 3, feature It is: " detectable disturbance degree correlation matrix ", " anti-false-alarm disturbance degree correlation matrix " described in step 4, " credible Rely disturbance degree correlation matrix " and " system risk disturbance degree correlation matrix ", when the matrix is all test point uncertainty Corresponding test correlation matrix, the value not instead of Boolean in these correlation matrixes, the probability value between 0 and 1.
CN201610177270.9A 2016-03-25 2016-03-25 A kind of probabilistic testability modeling method of consideration test point Active CN105844023B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610177270.9A CN105844023B (en) 2016-03-25 2016-03-25 A kind of probabilistic testability modeling method of consideration test point

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610177270.9A CN105844023B (en) 2016-03-25 2016-03-25 A kind of probabilistic testability modeling method of consideration test point

Publications (2)

Publication Number Publication Date
CN105844023A CN105844023A (en) 2016-08-10
CN105844023B true CN105844023B (en) 2019-08-16

Family

ID=56583648

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610177270.9A Active CN105844023B (en) 2016-03-25 2016-03-25 A kind of probabilistic testability modeling method of consideration test point

Country Status (1)

Country Link
CN (1) CN105844023B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111626622A (en) * 2020-05-28 2020-09-04 哈尔滨工业大学 Circuit system testability index prediction method considering uncertainty
WO2023115300A1 (en) * 2021-12-21 2023-06-29 Qualcomm Incorporated Dynamic over-the-air testing

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7050935B1 (en) * 1999-03-08 2006-05-23 Bombardier Transportation Gmbh Method for assessing the reliability of technical systems
CN101980225A (en) * 2010-11-16 2011-02-23 中国人民解放军63908部队 Method for implementing testability analysis and diagnosis decision system for electronic products
CN103927259A (en) * 2014-04-18 2014-07-16 北京航空航天大学 Fault detection and isolation synthesis method based on testability modeling data
CN104063593A (en) * 2014-06-18 2014-09-24 电子科技大学 Board-level circuit testability index calculation method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7050935B1 (en) * 1999-03-08 2006-05-23 Bombardier Transportation Gmbh Method for assessing the reliability of technical systems
CN101980225A (en) * 2010-11-16 2011-02-23 中国人民解放军63908部队 Method for implementing testability analysis and diagnosis decision system for electronic products
CN103927259A (en) * 2014-04-18 2014-07-16 北京航空航天大学 Fault detection and isolation synthesis method based on testability modeling data
CN104063593A (en) * 2014-06-18 2014-09-24 电子科技大学 Board-level circuit testability index calculation method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
《基于多信号模型的综合化航电系统综合诊断算法设计》;文佳;《电讯技术》;20140315;第54卷(第3期);361-367
《基于多信号流图的分层系统测试性建模与分析》;吕晓明等;《北京航空航天大学学报》;20110930;第37卷(第9期);1151-1155
《航空机电系统测试性建模与分析新方法》;代京等;《航空学报》;20100228;第31卷(第2期);277-284

Also Published As

Publication number Publication date
CN105844023A (en) 2016-08-10

Similar Documents

Publication Publication Date Title
Housh et al. Model-based approach for cyber-physical attack detection in water distribution systems
Perelman et al. Sensor placement for fault location identification in water networks: A minimum test cover approach
Moslemi et al. A fast, decentralized covariance selection-based approach to detect cyber attacks in smart grids
Sarrate et al. Sensor placement for leak detection and location in water distribution networks
Simeu-Abazi et al. A methodology of alarm filtering using dynamic fault tree
CN105512011B (en) A kind of electronics testability modeling appraisal procedure
CN105045251B (en) The demand analysis of industrial control system functional safety and information security and fusion method
CN104461896B (en) Aerospace system key software evaluation method based on credible attribute
CN104125112B (en) Physical-information fuzzy inference based smart power grid attack detection method
CN105976264A (en) Display method and system for comprehensive drainage information monitoring
CN116029884B (en) River pollution tracing method based on river channel water pollution tracing model
CN107092654A (en) Based on Change in Mean detect alarm normally with abnormal deviation data examination method and device
CN105786678A (en) Relevance model-based testability prediction method
CN103678881B (en) Composite fault diagnosis method based on combination of artificial immunity and evidence theory
CN106209829A (en) A kind of network security management system based on warning strategies
CN105608842A (en) Nuclear reactor fuel failure online monitoring alarm device
CN103023028A (en) Rapid grid failure positioning method based on dependency graph of entities
CN105844023B (en) A kind of probabilistic testability modeling method of consideration test point
Liao et al. Identifying effective management factors across human errors–A case in elevator installation
Li et al. New automatic incident detection algorithm based on traffic data collected for journey time estimation
CN109062194A (en) Main Compressor In Fccu group method for diagnosing faults and system under the influence of a kind of closed loop
Dahan et al. Network inspection for detecting strategic attacks
CN106302412A (en) A kind of intelligent checking system for the test of information system crushing resistance and detection method
Badri et al. Empirical Analysis for Investigating the Effect of Control Flow Dependencies on Testability of Classes.
CN105574299A (en) Safety pre-evaluation method for rail transit signal system

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
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20210111

Address after: 100089 No. 1105, 11 / F, Boyan building, 238 North Fourth Ring Middle Road, Haidian District, Beijing

Patentee after: Beijing Tianhang Changying Technology Co.,Ltd.

Address before: 100191 No. 37, Haidian District, Beijing, Xueyuan Road

Patentee before: BEIHANG University