CN109670610A - Fault diagnosis optimization method based on fault propagation analysis - Google Patents

Fault diagnosis optimization method based on fault propagation analysis Download PDF

Info

Publication number
CN109670610A
CN109670610A CN201811580902.1A CN201811580902A CN109670610A CN 109670610 A CN109670610 A CN 109670610A CN 201811580902 A CN201811580902 A CN 201811580902A CN 109670610 A CN109670610 A CN 109670610A
Authority
CN
China
Prior art keywords
test
fault
diagnosis
fuzzy set
expense
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.)
Pending
Application number
CN201811580902.1A
Other languages
Chinese (zh)
Inventor
闫鹏程
连光耀
孙江生
李会杰
张西山
连云峰
梁伟杰
张连武
代冬升
李雅峰
王凯
邱文浩
杨金鹏
陈然
李宝晨
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 China 32181 Army
Original Assignee
PLA China 32181 Army
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 China 32181 Army filed Critical PLA China 32181 Army
Priority to CN201811580902.1A priority Critical patent/CN109670610A/en
Publication of CN109670610A publication Critical patent/CN109670610A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/20Administration of product repair or maintenance

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Test And Diagnosis Of Digital Computers (AREA)

Abstract

The invention discloses a kind of fault diagnosis optimization methods based on fault propagation analysis, are related to method for diagnosing faults technical field.Described method includes following steps: acquisition is diagnosed the original state fuzzy set of optimization system before testing;Test is optimized to original state fuzzy set;Judge it is above-mentioned test whether to pass through, if test pass through, be inferred to possible diagnosis XjpIf test does not pass through, it is inferred to possible diagnosis Xjf;Judge the diagnosis XjpWith diagnosis XjfWhether reach insulation request, diagnosis is exported if reaching insulation request, if not up to insulation request, judge whether there is candidate fuzzy set, candidate fuzzy set is tested if there is candidate test set then repeats the above steps;Diagnosis is exported if without candidate fuzzy set.The method can optimize the method for fault diagnosis, make the time required for fault diagnosis and cost is relatively low.

Description

Fault diagnosis optimization method based on fault propagation analysis
Technical field
The present invention relates to method for diagnosing faults technical field more particularly to a kind of fault diagnosises based on fault propagation analysis Optimization method.
Background technique
Diagnostic Strategy optimization mainly includes that test item preferably optimizes with diagnostic process, and final purpose is realized to product Accurate, the quick detection of failure be isolated, for implement maintenance support work decision support is provided.Diagnostic Strategy optimization it is main according to According to being correlative relationship between product failure and test item, this correlative relationship and interiors of products fault propagation path are tight It is close to be connected.Method for diagnosing faults in the prior art will not generally optimize the strategy of fault diagnosis, therefore cause failure Time required for diagnosing and higher cost.
Summary of the invention
The technical problem to be solved by the present invention is to how provide one kind to optimize the method for fault diagnosis, make Time required for fault diagnosis and the lower-cost fault diagnosis optimization method based on fault propagation analysis.
In order to solve the above technical problems, the technical solution used in the present invention is: a kind of event based on fault propagation analysis Barrier diagnosis optimization method, it is characterised in that include the following steps:
Acquisition is diagnosed the original state fuzzy set of optimization system before testing;
Test is optimized to original state fuzzy set;
Judge it is above-mentioned test whether to pass through, if test pass through, be inferred to possible diagnosis XjpIf test is not led to It crosses, is inferred to possible diagnosis Xjf
Judge the diagnosis XjpWith diagnosis XjfWhether insulation request is reached, it is defeated if reaching insulation request Diagnosis out judges whether there is candidate fuzzy set if not up to insulation request, if there is candidate test set then repeats Step is stated to test candidate fuzzy set;Diagnosis is exported if without candidate fuzzy set.
A further technical solution lies in: the method also includes to diagnosis before testing original state fuzzy set Policy optimization carries out the step of data describe, and specifically comprises the following steps:
The basic settings of diagnostic method;
Essential information needed for constructing diagnostic method;
Target required for Optimizing fault diagnosis.
A further technical solution lies in: the basic settings of the diagnostic method includes:
1) single fault: occur assuming that being diagnosed optimization system only one most failure during diagnosis;
2) two-value is tested: the result of each test only passes through or not by two kinds of situation;
3) reliable test result: i.e. each test result is reliably, false-alarm and missing inspection to be not present;
4) test is independent: i.e. each test operation is independent of one another in product test and decision process, the independent also spy of test Reference valence is independent, i.e., it is constant that the cost of each step is unrelated with pervious operation;
5) system mode is constant: i.e. during system diagnostics, system fault condition does not change, i.e. only fault-free State and permanent fault state, without intermittent fault and transient fault.
A further technical solution lies in: the essential information includes:
It is diagnosed fuzzy set F, the F={ f of optimization system before testing0,f1,…fi…,fm};Wherein, f0Indicate fault-free Conclusion, fiIndicate the conclusion for only having i-th of failure to occur;
Diagnosis probability distribution set P:P={ P (f0),…P(fi),…P(fm), wherein P (f0) indicate system without reason The probability of barrier, P (fi) indicate only faulty fiThe probability of generation, is solved by following formula:
In formula, λiFor failure fiConstant failure rate;λkFor failure fkConstant failure rate, m be diagnosis probability distribution Number of elements in set P;
Available test set T:T={ t in system1,…,tn, it is specified that each test is two-value output, and test knot Fruit is all that reliably, n is the number of element in test set T;
Test execution expense set C:C={ c1,…,cn, indicate that test execution time, manpower estimate, it is specified that test fee With for constant, i.e., independent of testing sequence;
Correlation matrix B:B=[bij](m+1)*nIndicate the logical relation of test and diagnostic conclusion, element definition bij= dij, wherein dijIndicate the correlation of failure and test.
A further technical solution lies in: the determination in the target of optimization required for fault diagnosis: spent by cycle tests Average time and cycle tests average test expense as optimization target.
A further technical solution lies in: judge the diagnosis whether reach insulation request method it is as follows:
It enables and executes test tjPreceding system fuzzy set to be isolated is X, tests tjHave by with not pass through two kinds output, it is corresponding Inference conclusion is denoted as XjpAnd Xjf:
If tjPass through, then has
If tjDo not pass through, then has
Then fiNecessarily belong to set XjpAnd XjfOne of, and set X cannot be belonged to simultaneouslyjpAnd Xjf, that is, meet:
Xjf∪Xjp=X
Xjf∩Xjp
If system is invalid there are multiple faults, above-mentioned inference machine.
A further technical solution lies in: the heuristic function are as follows:
k*=argmaxjh(X;tj) or k*=argminjh(X;tj)
Above formula indicates that best test No. is k in next step for fuzzy set X to be isolated*, which is candidate test set T In can make function h (X;tj) value is maximum or the smallest test of value;h(X;tj) for calculating candidate test tjMould to be isolated The cost or ability of paste collection X.
A further technical solution lies in: using test point during optimizing test to original state fuzzy set Least Diagnostic Strategy generating algorithm, based on the preferred result of test point, according to test point preferred result, with the survey selected Pilot is tested, by test result it is normal whether determine in next step test.
A further technical solution lies in: test is optimized to original state fuzzy set using AO* searching method:
AO* searching method is using a kind of heuristic function based on the estimation of minimum testing expense:
k*=argminj{cj+P(Xjp)h(Xjp)+P(Xjf)h(Xjf)}
In formula, h (Xjp) and h (Xjf) respectively indicate set XjpAnd XjfMinimum testing expense estimation;
Given fuzzy set X is L according to the huffman coding average word length that its probability calculation obtains each failure*(X), to quilt Choosing test is ranked up (0≤c according to expense size1≤…≤cn), then the testing expense valuation functions of X are
In formula,For round numbers;The valuation functions give the lower bound of fuzzy set X isolation expense to be isolated;Use comentropy Huffman coding is substituted, to obtain another assessment of fees function:
In formula, H (X) is the entropy of fuzzy set X to be isolated;
It is similar to derive based on+1 expense estimation function of entropy:
The beneficial effects of adopting the technical scheme are that the method to spent by cycle tests by putting down Two aspect of average test expense of equal time and cycle tests carries out preferably method for diagnosing faults, the method for diagnosing faults for being Time and cost is relatively low so that method for diagnosing faults can be realized accurate, quick detection to product failure be isolated, be reality It applies maintenance support work and decision support is provided.
Detailed description of the invention
The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
Fig. 1 is the main flow chart of the method for the embodiment of the present invention;
Fig. 2 is the least Diagnostic Strategy generation method flow chart of test point in the method for the embodiment of the present invention.
Specific embodiment
With reference to the attached drawing in the embodiment of the present invention, technical solution in the embodiment of the present invention carries out clear, complete Ground description, it is clear that described embodiment is only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other Embodiment shall fall within the protection scope of the present invention.
In the following description, numerous specific details are set forth in order to facilitate a full understanding of the present invention, but the present invention can be with Implemented using other than the one described here other way, those skilled in the art can be without prejudice to intension of the present invention In the case of do similar popularization, therefore the present invention is not limited by the specific embodiments disclosed below.
Diagnostic Strategy optimization mainly includes that test item preferably optimizes with diagnostic process, and final purpose is realized to product Accurate, the quick detection of failure be isolated, for implement maintenance support work decision support is provided.Diagnostic Strategy optimization it is main according to According to being correlative relationship between product failure and test item, this correlative relationship and interiors of products fault propagation path are tight It is close to be connected.
The basic assumption of Diagnostic Strategy:
1) single fault, i.e. hypothesis system only one most failure during diagnosis occur.
2) two-value is tested, the result of each test only " passing through ", " not passing through " two kinds of situations.
3) reliable test result, i.e., each test result are reliably, false-alarm and missing inspection to be not present.
4) test is independent, i.e., each test operation is independent of one another in product test and decision process.The independent also spy of test Reference valence is independent, i.e., it is constant that the cost of each step is unrelated with pervious operation.
5) system mode is constant, that is, assume during system diagnostics, system fault condition does not change, i.e., only without Malfunction and permanent fault state, without intermittent fault and transient fault.
Essential information needed for constructing Diagnostic Strategy:
Layering Diagnostic Strategy problem can be defined by five-tuple (F, p, T, c, B), in which:
1) fuzzy set of system before testing is denoted as F={ f0,f1,…fi…,fm}.Wherein, f0Indicate fault-free conclusion, fiIndicate the conclusion for only having i-th of failure to occur.When designated analysis object and Fault Isolation precision, if object is LRU1, it is desirable that It is isolated to single SRU, then { f1,…,fmBy being subordinate to LRU1Each SRU each fault mode constitute.
2) diagnosis probability distribution set P:P={ P (f0),…P(fi),…P(fm), wherein P (f0) indicate system without The probability of failure, P (fi) indicate only faulty fiThe probability of generation, is solved by following formula:
In formula, λiFor failure fiConstant failure rate.
3) available test set T:T={ t in system1,…,tn, it is specified that each test is two-value output, and test It as a result is all reliable;
4) test execution expense set C:C={ c1,…,cn, indicate that test execution time, manpower estimate, it is specified that testing Expense is constant, i.e., independent of testing sequence;
5) correlation matrix B:B=[bij](m+1)*nIndicate the logical relation of test and diagnostic conclusion, element definition is bij=dij, wherein dijIndicate the correlation of failure and test.
Optimization aim:
Diagnostic Strategy optimization problem is a typical multi-standard problem, needs to consider following four factor:
1. average time spent by cycle tests;2. the average test expense of cycle tests;3. in test and diagnostic mistake It may be damaged caused by product in journey or other adverse consequences;It is negatively affected caused by test 4. executing, as BIT false-alarm causes User distrust and passive attitude.
Since risk and influence factor are generally difficult to determine, more only consider in practice time and expense because Element.
The least optimization object function of average test expense, is layered the optimization object function of diagnosis is defined as:
In formula, DoptIt indicates to meet isolation required precision, operational blocks which partition system faintly is not isolated and average test expense is least Diagnostic Strategy;D(i)Indicate the cycle tests that leaf node xi is isolated in diagnostic tree D, | D(i)| indicate the length of the sequence,Indicate sequence D(i)In k-th test expense.P(xi) it is leaf node xiProbability, if xiCorresponding Mr. Yu individually diagnoses Conclusion Ui, then have
General process:
Diagnostic Strategy generally uses sequential diagnostic method, this just determine the building process of Diagnostic Strategy be one " preferably The cyclic process of test → diagnostic reasoning → further preferably test ", basic procedure are as shown in Figure 1.By upper figure it is found that construct one A Diagnostic Strategy needs to solve following two critical issue:
1) how according to it is selected test and its output result reasoning obtain system diagnostics conclusion.
2) each step how test by optimum choice.
As shown in Figure 1, the embodiment of the invention discloses a kind of fault diagnosis optimization method based on fault propagation analysis, packet Include following steps:
Acquisition is diagnosed the original state fuzzy set of optimization system before testing;
Test is optimized to original state fuzzy set;
Judge it is above-mentioned test whether to pass through, if test pass through, be inferred to possible diagnosis XpIf test is not led to It crosses, is inferred to possible diagnosis Xf
Judge the diagnosis XjpWith diagnosis XjfWhether insulation request is reached, it is defeated if reaching insulation request Diagnosis out judges whether there is candidate fuzzy set if not up to insulation request, if there is candidate test set then repeats Step is stated to test candidate fuzzy set;Diagnosis is exported if without candidate fuzzy set.
Fault Isolation inference machine:
It, will be according to test after each test execution since the test in Diagnostic Strategy is that sequence executes As a result carry out the possible diagnosis of inference system, this reasoning process is known as the reasoning of system mode, specific reasoning side Method, referred to as " inference machine ".
It enables and executes test tjPreceding system fuzzy set to be isolated is X, tjHave by with not pass through two kinds output, corresponding reasoning Conclusion might as well be denoted as XjpAnd Xjf, X is derived by XjpAnd XjfMethod be exactly inference machine.
If tjPass through, then has
If tjDo not pass through, then has
If diagnosis Xif, then fiNecessarily belong to set XjpAnd XjfOne of, and set X cannot be belonged to simultaneouslyjpAnd Xjf, Meet:
Xjf∪Xjp=X
Xjf∩Xjp
If system is invalid there are multiple faults, above-mentioned inference machine.
The heuristic function for instructing test optimization to sort:
The possible diagnosis of system after Fault Isolation inference machine can only be tested according to given test and its result come reasoning, It not can determine that and tested actually using which in next step.In this regard, be usually to construct a function to use in next step to determine actually Which test, the function are known as " heuristic evaluation function ", abbreviation heuristic function.
The general type of heuristic function is as follows:
k*=argmaxjh(X;tj) or k*=argminjh(X;tj)
Above formula indicates that best test No. is k in next step for current fuzzy set X to be isolated*, which is candidate survey It can make function h (X in examination collection T;tj) value is maximum or the smallest test of value;h(X;tj) for calculating candidate test tjTo every Cost or ability from fuzzy set X.
The least Diagnostic Strategy generating algorithm of test point
So-called Diagnostic Strategy refers to testing sequence when fault detection and isolation.It is system testing/BIT detailed design The basis of analysis, while also technical support is provided for exterior diagnostic test.This Diagnostic Strategy can be not only used for product design Stage, it can also be used to fault diagnosis when service stage repairs.
The least Diagnostic Strategy of test point is first detected and is isolated afterwards based on the preferred result of test point, is clicked with test Sequencing out formulates diagnostic test strategy.Specific method is to be carried out according to test point preferred result with the test point selected Test, by test result it is normal whether determine in next step test.Diagnostic Strategy process is as shown in Figure 2.
Fault detection weight computing:
Assuming that the simplified correlation matrix of system is D=[dij]m×n, then the fault detection weight (table of j-th of test point Show and provide detection useful information how many opposite measurements) WFDIt can be calculated with following formula, i.e.,
Fault Isolation weight computing:
Assuming that the simplified correlation matrix of UUT is D=[dij]m×n, then the Fault Isolation weight of j-th of test point (mentions For the opposite measurement of Fault Isolation information) WFIIt can be calculated with following formula, i.e.,
In formula:Column matrix TjThe number that middle element is 1;
Column matrix TjThe number that middle element is 0;
Z- matrix function, z≤2P, p is to have been chosen as Fault Isolation number of test points.
Fault detection sequence:
1. with the 1st detection test point (FD TP) test macro:
If test result is normal, and " 0 " element corresponds to submatrixIt is not present, then fault-free.
If test result is normal, and " 0 " element corresponds to submatrixIn the presence of then being tested with the 2nd FD with TP
2. being tested with the 2nd FD with TP
If test result is normal, andIt is not present, then fault-free.
If test result is normal, andIn the presence of, then need with next test point test.
3. select next FD to be tested with TP, untilUntil being not present (selected FD is finished with TP).
4. such as any step testing result is abnormal, then fault isolation routine should be gone to.
Fault Isolation sequence:
1. with the 1st isolation test point (FI TP) test macro, by its result (normal or abnormal) system phase Closing property matrix is divided into two partsWith
As test result be it is normal, can determine thatFault-free, failure existIn, it need to be tested with the 2nd FI with TP
As test result be it is abnormal, then can determine thatFault-free, andThere are failures, need to test with the 2nd FI with TP
2. being again divided into two parts with the 2nd remaining faulty part (faulty submatrix) of FI TP test With
As test result be it is normal, then failure existsIn, it need to continue to test with TP with next FI
As test result be it is abnormal, then failure existsIn, it need to be continued with next FI TP pairIt is tested.
3. testing faulty submatrix with next FI TP(or), and it is divided into two, repeat above-mentioned mistake Until submatrix of the journey after dividing becomes uniline (component units or an ambiguity group of correspondence system).
4. during the test, any step isolation test, after original matrix is divided into two submatrixs, such as certain sub- square Battle array has become single row, then does not just have to test to the submatrix;It is not that the submatrix of uniline should continue to test to another.
Fault diagnosis tree:
Above-mentioned fault detection be isolated sequence analysis as a result, can be come out with the graphical representation of simple image.From first A FD is started with test point, draws Liang Ge branch by its test result " normal " and " abnormal ":
1. normal (being indicated with " 0 ") branch, continues to test with the 2nd FD with TP, then draw two analyses, wherein abnormal Branch is tested with FI with TP, is transferred to isolation branch;And wherein normal branch continues to test with FD with TP, until being finished FD TP, Decision-making system has fault-free, just draws detection ordering figure.
2. abnormal (being indicated with " 1 ") branch tests with the 1st FI with TP, press as a result " 0 " and drawing two for " 1 " A branch;It is tested respectively with the 2nd FI with TP again, draws Liang Ge branch.Branch is continuously drawn in this way, it is selected until being finished FI TP just depict isolation precedence diagram until each branches end is the single component units of system or ambiguity group.
AO* Diagnostic Strategy generating algorithm:
Heuristic evaluation function
Because generating optimal cycle tests is a np complete problem, it is therefore necessary to use heuristic guidance with or The search of figure.This heuristic is by using heuristic evaluation function HEF, to avoid enumerating all possible sons of decision tree Collection.HEF is the heuristic evaluation function h (x) for being easy the optimal test cost calculated, and wherein h* (x) indicates any one Test cost of the fuzzy subset x to leaf node.
According to the similitude of cycle tests and huffman coding problem, heuristic evaluation function HEF can be derived.By suddenly The characteristic of Fu Man coding come help to derive with or graph search during the heuristic evaluation function HEF that uses.
Attribute: the conditional average Huffman encoded length w* (x) of any fuzzy subset x is conditional average coding Length provides a lower limit.All end at x in the 1 (x) of any testing algorithm.It shows themselves in that in form
Wherein in the cycle tests containing x, if having used test tjIt can identify failure si, then aij(x)=1, otherwise aij (x)=0, and
For probability P (si) corresponding optimal Huffman code length vector.
The probability of malfunction descending sort of corresponding fuzzy set x to be isolated is P (s1)≥P(s2)≥…≥P(sm), wherein si∈ X, then with probability vector P=[P (s1), P (s2) ..., P (sm)] corresponding optimal Huffman code length vector w=[1, 2 ..., m1, m, m], the average Huffman code length w* (x) of fuzzy subset x is
With or figure G in, if heuristic evaluation function h (x) has h (x) h* (x) for figure each of G node x, Then it is acceptable, and is optimal test cost.Notice that h (x) just can be infinite only when h* (x) is infinite.According to The attribute of huffman coding can derive acceptable heuristic evaluation function HEF, and formula is as shown in theorem.
Theorem: hypothesis test cost is 0≤c according to ascending sort without loss of generality1≤c2...≤cn, then under h (x) It limits as follows:
Wherein w ' (x) is the integer part of w* (x).
The basic ideas of AO* searching method:
AO* searching algorithm be with or graph search a kind of method for solving, basic ideas are as follows:
1, the point of minor matters caused by each test is successively calculated for root node, calculates using Huffman encoding and this is isolated Cost is isolated in minimum needed for a little minor matters points, then cost weighted sum obtained by calculating, as the heuristic function value of the test, than The heuristic function value of more all tests, the temporary the smallest test of selected numerical value are tested as the first step;
2, for the smallest minor matters point of cost is isolated under the test, second step test is selected according to preceding method;
3, the heuristic function value of the test is gradually fed back upwards, i.e., first the isolation cost of minor matters point above it is repaired Just, then with the heuristic function value that the correction value of the minor matters point tests first it is modified;
4, it after correcting, resequences to all candidate test below root node according to heuristic function value, temporarily selects The smallest test of fixed number value is tested as the first step, and applies preceding method selection second step test.
The heuristic function that AO* searching method uses:
AO* searching method is using a kind of heuristic function based on the estimation of minimum testing expense:
k*=argminj{cj+P(Xjp)h(Xjp)+P(Xjf)h(Xjf)}
In formula, h (Xjp) and h (Xjf) respectively indicate set XjpAnd XjfMinimum testing expense estimation.
Given fuzzy set X is L according to the huffman coding average word length that its probability calculation obtains each failure*(X), to quilt Choosing test is ranked up (0≤c according to expense size1≤…≤cn), then the testing expense valuation functions of X are
In formula,For round numbers.The valuation functions give the lower bound of state set X isolation expense.Suddenly with comentropy substitution Fu Man coding, to obtain another assessment of fees function:
In formula, H (X) is the entropy of failure fuzzy set X.
It is similar to derive based on+1 expense estimation function of entropy:
The heuristic function of three above heuristic function and greedy algorithm is different, they are not selection information and expense ratio It is worth maximum test, but selects the least test of follow-up diagnosis expense.Due to being expense approximate evaluation, selected test Only local optimum.

Claims (9)

1. a kind of fault diagnosis optimization method based on fault propagation analysis, it is characterised in that include the following steps:
Acquisition is diagnosed the original state fuzzy set of optimization system before testing;
Test is optimized to original state fuzzy set;
Judge it is above-mentioned test whether to pass through, if test pass through, be inferred to possible diagnosis XjpIf test does not pass through, It is inferred to possible diagnosis Xjf
Judge the diagnosis XjpWith diagnosis XjfWhether reach insulation request, diagnosis is exported if reaching insulation request Conclusion judges whether there is candidate fuzzy set if not up to insulation request, if there is candidate test set then repeats the above steps Candidate fuzzy set is tested;Diagnosis is exported if without candidate fuzzy set.
2. the fault diagnosis optimization method as described in claim 1 based on fault propagation analysis, it is characterised in that: to initial State fuzzy set tested before the method also includes to Diagnostic Strategy optimize carry out data describe the step of, specifically include as Lower step:
The basic settings of diagnostic method;
Essential information needed for constructing diagnostic method;
Target required for Optimizing fault diagnosis.
3. the fault diagnosis optimization method as claimed in claim 2 based on fault propagation analysis, which is characterized in that the diagnosis The basic settings of method includes:
1) single fault: occur assuming that being diagnosed optimization system only one most failure during diagnosis;
2) two-value is tested: the result of each test only passes through or not by two kinds of situation;
3) reliable test result: i.e. each test result is reliably, false-alarm and missing inspection to be not present;
4) test is independent: i.e. each test operation is independent of one another in product test and decision process, and test is independent also to refer in particular to generation Valence is independent, i.e., it is constant that the cost of each step is unrelated with pervious operation;
5) system mode is constant: i.e. during system diagnostics, system fault condition does not change, i.e. only unfaulty conditions With permanent fault state, without intermittent fault and transient fault.
4. the fault diagnosis optimization method as claimed in claim 2 based on fault propagation analysis, which is characterized in that described basic Information includes:
It is diagnosed fuzzy set F, the F={ f of optimization system before testing0,f1,…fi…,fm};Wherein, f0Indicate fault-free conclusion, fiIndicate the conclusion for only having i-th of failure to occur;
Diagnosis probability distribution set P:P={ P (f0),…P(fi),…P(fm), wherein P (f0) indicate that system is trouble-free Probability, P (fi) indicate only faulty fiThe probability of generation, is solved by following formula:
In formula, λiFor failure fiConstant failure rate;λkFor failure fkConstant failure rate, m be diagnosis probability distribution set P Middle number of elements;
Available test set T:T={ t in system1,…,tn, it is specified that each test is two-value output, and test result is all It is that reliably, n is the number of element in test set T;
Test execution expense set C:C={ c1,…,cn, indicate that test execution time, manpower estimate, it is specified that testing expense is Constant, i.e., independent of testing sequence;
Correlation matrix B:B=[bij](m+1)*nIndicate the logical relation of test and diagnostic conclusion, element definition bij=dij, Wherein dijIndicate the correlation of failure and test.
5. the fault diagnosis optimization method as claimed in claim 2 based on fault propagation analysis, which is characterized in that examined in failure Determine when the target optimized required for disconnected: the average test expense of average time spent by cycle tests and cycle tests is made For the target of optimization.
6. the fault diagnosis optimization method as claimed in claim 2 based on fault propagation analysis, which is characterized in that described in judgement The method whether diagnosis reaches insulation request is as follows:
It enables and executes test tjPreceding system fuzzy set to be isolated is X, tests tjHave by with not pass through two kinds output, corresponding reasoning Conclusion is denoted as XjpAnd Xjf:
If tjPass through, then has
If tjDo not pass through, then has
Then fiNecessarily belong to set XjpAnd XjfOne of, and set X cannot be belonged to simultaneouslyjpAnd Xjf, that is, meet:
Xjf∪Xjp=X
Xjf∩Xjp
If system is invalid there are multiple faults, above-mentioned inference machine.
7. the fault diagnosis optimization method as described in claim 1 based on fault propagation analysis, which is characterized in that the inspiration Function are as follows:
k*=argmaxjh(X;tj) or k*=argminjh(X;tj)
Above formula indicates that best test No. is k in next step for fuzzy set X to be isolated*, which can be made in candidate test set T Function h (X;tj) value is maximum or the smallest test of value;h(X;tj) for calculating candidate test tjFuzzy set X's to be isolated Cost or ability.
8. the fault diagnosis optimization method as described in claim 1 based on fault propagation analysis, which is characterized in that initial State fuzzy set uses the least Diagnostic Strategy generating algorithm of test point during optimizing test, with the preferred of test point As a result based on, according to test point preferred result, tested with the test point selected, by test result it is normal whether determine It tests in next step.
9. the fault diagnosis optimization method as described in claim 1 based on fault propagation analysis, it is characterised in that: use AO* Searching method optimizes test to original state fuzzy set:
AO* searching method is using a kind of heuristic function based on the estimation of minimum testing expense:
k*=argminj{cj+P(Xjp)h(Xjp)+P(Xjf)h(Xjf)}
In formula, h (Xjp) and h (Xjf) respectively indicate set XjpAnd XjfMinimum testing expense estimation;
Given fuzzy set X is L according to the huffman coding average word length that its probability calculation obtains each failure*(X), to selected test (0≤c is ranked up according to expense size1≤…≤cn), then the testing expense valuation functions of X are
In formula,For round numbers;The valuation functions give the lower bound of fuzzy set X isolation expense to be isolated;It is substituted with comentropy Huffman coding, to obtain another assessment of fees function:
In formula, H (X) is the entropy of fuzzy set X to be isolated;
It is similar to derive based on+1 expense estimation function of entropy:
CN201811580902.1A 2018-12-24 2018-12-24 Fault diagnosis optimization method based on fault propagation analysis Pending CN109670610A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811580902.1A CN109670610A (en) 2018-12-24 2018-12-24 Fault diagnosis optimization method based on fault propagation analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811580902.1A CN109670610A (en) 2018-12-24 2018-12-24 Fault diagnosis optimization method based on fault propagation analysis

Publications (1)

Publication Number Publication Date
CN109670610A true CN109670610A (en) 2019-04-23

Family

ID=66147112

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811580902.1A Pending CN109670610A (en) 2018-12-24 2018-12-24 Fault diagnosis optimization method based on fault propagation analysis

Country Status (1)

Country Link
CN (1) CN109670610A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111426937A (en) * 2020-04-07 2020-07-17 吉林大学 Fault diagnosis method based on fault-free information test score
CN112163189A (en) * 2020-11-03 2021-01-01 北京机电工程研究所 Equipment fault diagnosis method based on test sequence probability matrix
CN112529286A (en) * 2020-12-08 2021-03-19 哈尔滨工业大学 Fault diagnosis strategy optimization method based on Rollout algorithm
CN113221496A (en) * 2021-05-06 2021-08-06 电子科技大学 Fault diagnosis method based on three-dimensional testability analysis model
CN113887452A (en) * 2021-10-09 2022-01-04 中国舰船研究设计中心 Fault diagnosis optimization method based on correlation matrix
CN113962394A (en) * 2021-09-10 2022-01-21 重庆邮电大学 Automatic generation method of correlation model based fault diagnosis inference machine

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108304661A (en) * 2018-02-05 2018-07-20 南京航空航天大学 Diagnosis prediction method based on TDP models

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108304661A (en) * 2018-02-05 2018-07-20 南京航空航天大学 Diagnosis prediction method based on TDP models

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
杨鹏: "基于相关性模型的诊断策略优化设计技术", 《中国博士学位论文全文数据库(电子期刊)工程科技Ⅱ辑》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111426937A (en) * 2020-04-07 2020-07-17 吉林大学 Fault diagnosis method based on fault-free information test score
CN111426937B (en) * 2020-04-07 2021-09-24 吉林大学 Fault diagnosis method based on fault-free information test score
CN112163189A (en) * 2020-11-03 2021-01-01 北京机电工程研究所 Equipment fault diagnosis method based on test sequence probability matrix
CN112529286A (en) * 2020-12-08 2021-03-19 哈尔滨工业大学 Fault diagnosis strategy optimization method based on Rollout algorithm
CN113221496A (en) * 2021-05-06 2021-08-06 电子科技大学 Fault diagnosis method based on three-dimensional testability analysis model
CN113221496B (en) * 2021-05-06 2022-06-14 电子科技大学 Fault diagnosis method based on three-dimensional testability analysis model
CN113962394A (en) * 2021-09-10 2022-01-21 重庆邮电大学 Automatic generation method of correlation model based fault diagnosis inference machine
CN113887452A (en) * 2021-10-09 2022-01-04 中国舰船研究设计中心 Fault diagnosis optimization method based on correlation matrix
CN113887452B (en) * 2021-10-09 2024-05-21 中国舰船研究设计中心 Fault diagnosis optimization method based on correlation matrix

Similar Documents

Publication Publication Date Title
CN109670610A (en) Fault diagnosis optimization method based on fault propagation analysis
CN109086889B (en) Terminal fault diagnosis method, device and system based on neural network
CN100554980C (en) Be used for supporting the method for the fault functional unit of recognology device
US5463768A (en) Method and system for analyzing error logs for diagnostics
Huang Evolving neural nets for fault diagnosis of power transformers
CA2387929C (en) Method and apparatus for diagnosing difficult to diagnose faults in a complex system
KR940001445B1 (en) Method & apparatus for diagnosing fault
CN108732528A (en) A kind of digitalized electrical energy meter method for diagnosing faults based on depth confidence network
CN109325293A (en) A kind of Multiple faults diagnosis approach based on correlation models
CN106932670A (en) A kind of distribution power automation terminal method for diagnosing status based on D S evidence theories
US20120016824A1 (en) Method for computer-assisted analyzing of a technical system
CN107291063A (en) Diagnostic device and diagnostic method for the operation of monitoring technology facility
CN109670611A (en) A kind of power information system method for diagnosing faults and device
CN107966648B (en) A kind of embedded failure diagnosis method based on correlation matrix
CN113626267A (en) Method for evaluating uncertainty fault diagnosis efficiency of complex electronic system
CN109870306A (en) A kind of tractor engine Fault diagnosis and forecast method
CN115037603A (en) Diagnosis evaluation method, device and system of electricity consumption information acquisition equipment
CN112882875A (en) Fault diagnosis method
Kipersztok et al. Another look at sensitivity of Bayesian networks to imprecise probabilities
CN105759782B (en) A kind of car fault diagnosis construction of strategy method based on density of infection
CN111860881B (en) Multi-incentive equipment fault maintenance and troubleshooting method and device
CN115563868A (en) Fault diagnosis method and device for oil circuit system of aviation alternating-current generator
Amati et al. An incremental approach to functional diagnosis
Zuiev et al. Questions of radioelectronic equipment diagnostics programs efficiency analysis
US11886831B2 (en) Data sorting device and method, and monitoring and diagnosis device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20190423