CN109799804A - A kind of diagnosis algorithm appraisal procedure and system based on random fault injection - Google Patents
A kind of diagnosis algorithm appraisal procedure and system based on random fault injection Download PDFInfo
- Publication number
- CN109799804A CN109799804A CN201811638187.2A CN201811638187A CN109799804A CN 109799804 A CN109799804 A CN 109799804A CN 201811638187 A CN201811638187 A CN 201811638187A CN 109799804 A CN109799804 A CN 109799804A
- Authority
- CN
- China
- Prior art keywords
- fault
- diagnosis algorithm
- location model
- indicate
- formula
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The present invention relates to trailer system fault diagnosis fields, disclose a kind of diagnosis algorithm appraisal procedure and system based on random fault injection, to carry out more comprehensively more objective assessment to system diagnosis algorithm to be assessed；The method of the present invention includes direct fault location model library is established, the fault message of each of direct fault location model library direct fault location model is layered, fault message includes abort situation, fault type and fault parameter；Layered extraction is carried out to the fault message of each direct fault location model at random according to sets requirement, chooses corresponding direct fault location model for diagnosis algorithm to be assessed；Diagnosis algorithm to be assessed is tested using the direct fault location model of selection, and according to test result calculations evaluation index, and the calculated value of evaluation index is matched with the evaluation grade of setting, to obtain the evaluation grade of diagnosis algorithm to be assessed.
Description
Technical field
The present invention relates to trailer system fault diagnosis field more particularly to a kind of diagnosis algorithms based on random fault injection
Appraisal procedure and system.
Background technique
With the continuous development of modern railway transportation technology, the safety of bullet train have become high-speed railway operation with
The matter of utmost importance of development, critical system one of of the Traction Drive control system as bullet train operational safety, and high speed arrange
One of the main source of overall height hair failure often uses for the safe and reliable operation of real-time detection Traction Drive control system
Various diagnosis algorithms are measured in real time Traction Drive control system, can detect in time and diagnose Traction Drive control system
Failure, but used fault diagnosis algorithm is before the use that puts into operation, it is necessary to pass through the standard of its algorithm of experimental verification
True property.Currently, the mode tested diagnosis algorithm is that the determination of lower fault scenes (type, parameter etc.) is manually set mostly
Type test can not carry out a large amount of random tests of all standing, test result to all possible fault scenes (type, parameter etc.)
In the presence of one-sidedness and contingency largely；In addition, the assessment for test result, existing appraisal procedure is only chosen several
Evaluation index is difficult performance that is objective, comprehensive, synthetically assessing diagnosis algorithm.
Therefore, how it is more acurrate more comprehensively, objectively to system diagnosability algorithm carry out test and evaluation become one
Urgent problem.
Summary of the invention
It is an object of that present invention to provide a kind of diagnosis algorithm appraisal procedures and system based on random fault injection, with more complete
Face more objectively carries out test and evaluation to the diagnosis algorithm of Traction Drive control system.
To achieve the above object, the present invention provides a kind of diagnosis algorithm appraisal procedure based on random fault injection, packets
Include following steps:
S1: establishing direct fault location model library, by the event of each of direct fault location model library direct fault location model
Barrier information is layered, and the fault message includes abort situation, fault type and fault parameter；
S2: layered extraction is carried out the fault message of each direct fault location model at random according to sets requirement, is
Diagnosis algorithm to be assessed chooses corresponding direct fault location model；
S3: diagnosis algorithm to be assessed is tested using the direct fault location model chosen in S2, and according to test result
Evaluation index is calculated, and the calculated value of the evaluation index is matched with the evaluation grade of setting, to obtain to be assessed examine
The evaluation grade of disconnected algorithm.
Preferably, in the S3, the evaluation index of the setting include basic performance indices, Key Performance Indicator and
Integrated performance index；
The basic performance indices include: detection retardation rate, sensitivity, verification and measurement ratio, false detection rate, omission factor, abort situation
Discrimination power, fault type discrimination power, fault parameter discrimination power；
The Key Performance Indicator includes reagency, validity and sense；Wherein, the reagency, it is described effectively
Property and the sense are grouped by all basic performance indices respectively and carry out combined weighted and be calculated；
The integrated performance index is calculated by all key performance indicator weightings.
Preferably, it further comprises the steps of: and constructs at least three index groups according to the basic performance indices, be respectively as follows: reaction
Power index group, Validity Index group and sense index group；
The reagency index group includes: the detection retardation rate and the sensitivity；
The Validity Index group includes: the verification and measurement ratio, the false detection rate and the omission factor；
The sense index group includes: the abort situation discrimination power, the fault type discrimination power and the event
Hinder parameter identification rate.
Preferably, the direct fault location model in the direct fault location model library is Q, calculation formula are as follows:
In formula, i indicates i-th of abort situation, i=1 ..., m, and may wherein break down in m expression system position
Sum, l indicate the fault type sum of i-th of abort situation.
Preferably, in the S1, by the fault message of each of direct fault location model library direct fault location model
When being layered, the layering is realized by the direct fault location model f of setting, wherein the calculating of the direct fault location model f of setting
Formula are as follows:
In formula, modiIndicate abort situation,Indicate fault type,All indicate fault parameter.
Preferably, the S2 specifically includes the following steps:
S21: the Cumulative Distribution Function of abort situation is calculatedCalculation formula are as follows:
In formula,Indicate abort situation modiThe probability being drawn into,
S22: the abort situation extracted, calculation formula are determined are as follows:
In formula, n indicates n-th test,Indicate n-th test in order to determine that abort situation is uniformly extracted from [0,1]
Random number, anIndicate the abort situation that n-th test is extracted, wherein anValue range is { 1,2 ..., m },
S23: calculating abort situation is anWhen fault typeCumulative distribution functionCalculation formula are as follows:
In formula,
Randomly selecting abort situation is anWhen fault type beCalculation formula are as follows:
In formula,Indicate n-th test in order to determine random number that fault type is uniformly extracted from [0,1],It indicates
The abort situation that n-th test is extracted is anWhen fault type,Value range isWherein
S24: calculating abort situation is anFault type isV-th of fault parameterCumulative distribution functionCalculation formula are as follows:
In formula,For v-th of fault parameterProbability density function；
(0,1) is divided into N equal part, n-th of section DnAre as follows:
N-th is tested, (0,1) is carried outIt is secondary to be uniformly distributed sampling and obtainV-th of fault parameterIn section DnIn Probability Point are as follows:
Calculate Probability PointCorresponding fault parameter valueFormula are as follows:
S25: abort situation, fault type and the fault parameter in summary randomly selected choose n-th and extract to obtain
Direct fault location model fn, calculation formula are as follows:
In formula, n=1 ..., N, N indicate to choose the experiment for assessing whole direct fault location models needed for diagnosis algorithm
Total degree.
Preferably, the calculation formula of the experiment total degree N are as follows:
In formula, ε indicates tolerance, and ε ∈ (0,1), α indicate confidence level, and have:
In formula,Indicate the perfect estimation value of parameter, λ indicates the actual estimated value of parameter, and Pr () indicates general
Rate.
Preferably, the S3 specifically includes the following steps:
S31: using choose direct fault location model diagnosis algorithm to be assessed is tested, set direct fault location model as
fn, noise ωn, the direct fault location model f that is extracted using n-thn, n-th test is carried out to diagnosis algorithm, record is surveyed
Examination output result are as follows:
In formula, rn=1 expression diagnosis algorithm decision-making system M breaks down, rn=0 expression diagnosis algorithm decision-making system M is not sent out
Raw failure, φnIndicate detection delay threshold value, ξnIndicate failure weak degrees threshold value, wherein failure weak degrees are by fault parameter
SetIt determines, is denoted asδnIndicate that fault parameter recognizes threshold value,Indicate time of failure,It indicates at the latest
Failure occurs to determine time, tnIt indicates the earliest time of failure that diagnosis algorithm is diagnosed to be, indicates to calculate using diagnosis to be assessed
The abort situation that method determines Indicate the fault type determined using diagnosis algorithm to be assessed,Indicate the fault parameter set determined using diagnosis algorithm to be assessed；Wherein,
Indicate that abort situation can not recognize；Indicate that fault type can not recognize；Indicate theA fault parameterIt can not recognize；
S32: basic performance indices, Key Performance Indicator and comprehensive performance are calculated according to the test output result in S31 and referred to
Mark is obtained to the basic performance of diagnosis algorithm n times random test to be assessed, key performance and synthetic performance evaluation grade.
Preferably, in the S2, the direct fault location model of selection includes one or at least two；When the event of the selection
When barrier injection model is one, the sets requirement are as follows: the corresponding failure probability of happening of direct fault location model is 1；
When the direct fault location model of the selection is at least two, the sets requirement are as follows: the phase of direct fault location model
Answering the sum of fault rate is 1.
As a general technical idea, the present invention also provides a kind of diagnosis algorithms based on random fault injection to assess system
System including memory, processor and stores the computer program that can be run on a memory and on a processor, the processing
The step of device realizes the above method when executing the computer program.
The invention has the following advantages:
The present invention provide it is a kind of based on random fault injection diagnosis algorithm appraisal procedure and system, by sets requirement with
Machine carries out layered extraction to the fault message of each of direct fault location model library direct fault location model, calculates for diagnosis to be assessed
Method randomly selects corresponding direct fault location model and is tested, according to test result calculations evaluation index, and by evaluation index
Calculated value is matched with the evaluation grade of setting, can be to be evaluated to system to obtain the evaluation grade of diagnosis algorithm to be assessed
Estimate diagnosis algorithm and carries out more comprehensively more objective assessment.
Below with reference to accompanying drawings, the present invention is described in further detail.
Detailed description of the invention
The attached drawing constituted part of this application is used to provide further understanding of the present invention, schematic reality of the invention
It applies example and its explanation is used to explain the present invention, do not constitute improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is the diagnosis algorithm appraisal procedure flow chart based on random fault injection of the preferred embodiment of the present invention.
Specific embodiment
The embodiment of the present invention is described in detail below in conjunction with attached drawing, but the present invention can be defined by the claims
Implement with the multitude of different ways of covering.
Unless otherwise defined, all technical terms used hereinafter and the normally understood meaning of those skilled in the art
It is identical.The similar word of "one" or " one " etc. used in present patent application specification and claims does not indicate
Quantity limitation, but indicate that there are at least one.
Embodiment 1
Referring to Fig. 1, the present embodiment provides a kind of diagnosis algorithm appraisal procedures based on random fault injection, including following step
It is rapid:
S1: establishing direct fault location model library, and the failure of each of direct fault location model library direct fault location model is believed
Breath is layered, and fault message includes abort situation, fault type and fault parameter；
S2: layered extraction is carried out to the fault message of each direct fault location model at random according to sets requirement, is to be evaluated
Estimate diagnosis algorithm and chooses corresponding direct fault location model；
S3: diagnosis algorithm to be assessed is tested using the direct fault location model chosen in S2, and according to test result
Evaluation index is calculated, and the calculated value of evaluation index is matched with the evaluation grade of setting, is calculated with obtaining diagnosis to be assessed
The evaluation grade of method.
The above-mentioned diagnosis algorithm appraisal procedure based on random fault injection can be more comprehensively more objectively to Traction Drive
The diagnosis algorithm of control system is assessed.
Specifically, in actual experiment, firstly, building direct fault location model library, it should be noted that institute in the present embodiment
The direct fault location model library of finger is that direct fault location model library commonly used in the prior art does not repeat herein.The present embodiment
In, the direct fault location model in the direct fault location model library of building is Q, calculation formula are as follows:
In formula, i indicates i-th of abort situation, i=1 ..., m, and may wherein break down in m expression system position
Sum, l indicate the fault type sum of i-th of abort situation.
Further, the fault message of each of direct fault location model library direct fault location model is layered, this
In embodiment, fault message is divided into three layers, it is preferable that first layer is abort situation, and the second layer is fault type, and third layer is
Fault parameter.And hierarchical operations realize the layering by the direct fault location model f of setting, wherein the direct fault location mould of setting
The calculation formula of type f are as follows:
In formula, modiIndicate abort situation,Indicate fault type,All indicate fault parameter.
Then, it chooses corresponding direct fault location model from direct fault location model library, in the present embodiment, discusses and choose at least
To test diagnosis algorithm to be assessed, (in this case, diagnosis algorithm to be assessed can be used for detecting a variety of two direct fault location models
Fault type).At this point, corresponding event occurs for the direct fault location model that should meet all selections when choosing direct fault location model
Hindering the sum of probability is 1.
Specifically, abort situation, fault type, fault parameter and the model that the traction drive that the present embodiment is chosen includes
It encloses as shown in table 1.And for a kind of diagnosable failure simultaneously identification of defective position and type occur for the diagnosis algorithm that the present embodiment is chosen
But it is unable to the algorithm of identified parameters.
Table 1
Firstly, determining the abort situation in direct fault location model library, abort situation Mod is by all in traction drive
The position mod that may be broken down1、mod2Composition:
Mod={ mod1, mod2}；
In formula, mod1For inverter three-phase current V phase, mod2For sensor three-phase current U phase.
Determine that the fault type in direct fault location model library, fault type S are all mod1、mod2Corresponding failure
Type set S1、S2Set, S1、S2It is respectively as follows:
In formula, s11For broken bar fault type, s12For shorted-turn fault type；s21For deviation fault type, s22For gain
Fault type.
Then, in the present embodiment, direct fault location model sum Q are as follows:
Q=2+2=4；
Determine that the fault type in direct fault location model library, fault parameter Θ are all fault type s11、s12、s21、s22Point
Not contained fault parameter Θ11、Θ12、Θ21、Θ22Set, Θ11、Θ12、Θ21、Θ22It is respectively as follows:
In formula,ForV-th of fault parameter,Continuous value on the range intervals shown in table 1 respectively.Its section is
(0,257.4), (0,0.076), (0,1), (0,257.4), (0,0.076), (0,1), (0,10), (0,10).
Then, in the present embodiment in direct fault location model library three layers setting direct fault location model f are as follows:
In formula, i=1,2, j1=1,2, j2=1,2, g11=3, g12=3, g21=1, g22=1.
Then, abort situation is randomly selected；
Specifically, abort situation mod is first calculatediCumulative distribution functionCalculation formula are as follows:
In the present embodiment,For abort situation mod1The probability being drawn into；For abort situation
mod2The probability being drawn into.
It should be noted that difference variable is numbered for convenience, it is assumed that tested when previous for the 5th.(0,1) is carried out equal
Even distribution sampling obtains 0.213, determines that extracting abort situation is mod by following formula1:
In formula,
Randomly select fault type；
Calculate fault type when abort situation is 1Cumulative distribution function
In the present embodiment,For fault type s11The probability being drawn into；For fault type s12Quilt
The probability being drawn into.
(0,1) is carried out being uniformly distributed sampling and obtains 0.582, determines that extracting fault type is s by following formula12:
Randomly select fault parameter；
It is s that calculating abort situation, which is 1 fault type,12Three fault parametersCumulative distribution function
In the present embodiment,For fault parameterProbability density function；For fault parameterProbability density function；For fault parameterProbability density function.
Preferably, the present embodiment gives tolerance ε=0.1 and confidence alpha=95%, chooses overall test times N by following formula
=81:
(0,1) is divided into 81 equal parts, then the 5th section
Sampling, which is uniformly distributed, to (0,1) progress 3 times obtains [0.721,0.396,0.581], fault parameter
Respectively in section D1、D2、D3In Probability Point are as follows:
Calculate Probability PointCorresponding fault parameter value
The then direct fault location model f that the 5th extracts5Are as follows:
(mod1, s12, [24.167,0.00561,0.085])；
N=1 ..., 81 obtains whole 81 random fault injection models needed for assessment diagnosis algorithm accordingly.
Further, according to the direct fault location model f of extraction5Diagnosis algorithm is tested.
Setting traction drive M includes direct fault location model f5And noise ω5:
M(f5, ω5)；
In formula, ω5The white noise of system when representing the 5th test.
The direct fault location model f extracted in traction drive M using the 5th5, the 5th is carried out to diagnosis algorithm
Secondary test, record test output result are as follows:
In the present embodiment, r5=1, i.e. diagnosis algorithm judgement is broken down；φ5=2, postpone threshold value, ξ for detection5=0.1,
For failure weak degrees threshold value, failure weak degrees are by fault parameter set Θ12It determines,δ5=1,
Threshold value is recognized for fault parameter,For time of failure；Occur to determine the time for failure at the latest；t5=1.32,
The earliest time of failure being diagnosed to be for diagnosis algorithm；
The respectively abort situation of diagnosis algorithm judgement, fault type, fault parameter.Wherein, fault parameterIt can not distinguish
Know.
It should be noted that the detection delay threshold value in the present embodiment, failure weak degrees threshold value, fault parameter recognize threshold
Value is all set according to routine experimentation and experience, here, not repeating.
As the present embodiment preferred embodiment, in the present embodiment, the assessment for assessing diagnosis algorithm of setting refers to
Mark includes detection retardation rate (Detection Delay Rate, DDR), sensitivity (Sensity, SEN), verification and measurement ratio (Fault
Detection Rate, FDR), false detection rate (False Alarm Rate, FAR), omission factor (Miss Rate, MR), fault bit
Set discrimination power (Fault Module RecognizationRate, FMRR), fault type discrimination power (Fault
TypeRecognizationRate, FTRR), fault parameter discrimination power (Fault ParameterRecognizationRate,
FPRR).When calculating evaluation index, specific formula for calculation are as follows:
In formula, DDR ∈ [0,1], if rn=1 andIt sets up, ddrn=1, on the contrary ddrn=0.
In formula, SEN ∈ [0,1], if rn=1 andIt sets up, senn=1, on the contrary senn=0.
In formula, FDR ∈ [0,1], if rn=1 andIt sets up, fdrn=1, on the contrary fdrn=0.
In formula, FAR ∈ [0,1], if rn=1 andIt sets up, farn=1, on the contrary farn=0.
In formula, MR ∈ [0,1], ifIt sets up, mrn=1, on the contrary mrn=0.
In formula, FMRR ∈ [0,1], ifIt sets up, fmrrn=1, on the contrary fmrrn=0.
In formula, FTRR ∈ [0,1], ifIt sets up, ftrrn=1, on the contrary ftrrn=0.
In formula, FPRR ∈ [0,1], ifIt sets up, fprrn=1, on the contrary fprrn=0.
Further, in order to preferably assess diagnosis algorithm, above-mentioned evaluation index is divided into reagency index
Group, Validity Index group and sense index group, by reagency index group, Validity Index group and sense index group
It is considered as Key Performance Indicator.Specifically, it will test retardation rate and the sensitivity be divided into reagency index group；Will test rate,
False detection rate and omission factor are divided into Validity Index group；By abort situation discrimination power, fault type discrimination power and failure
Parameter identification rate is divided into sense index group.Wherein reagency index group is Ω1, Validity Index group be Ω2, sense refers to
Mark group is Ω3, weighting is combined by the basic performance indices that each index group is included, wherein combined weighted calculation formula are as follows:
In formula,For k-th of basic performance indices in the τ index group, whenIt is bigger to represent diagnosis algorithm performance more
When good,Otherwise, ForWeight,hτFor the basic of the τ index group
Performance indicator number.Ωτ∈ [0,100], τ=1,2,3, it is strong that corresponding key performance is respectively represented close to 100.
In the present embodiment, it is calculated:
In formula, Ωτ∈ [0,100], τ=1,2,3, wherein end value respectively represents that reagency is strong, validity close to 100
By force, sense is strong.
Then, it is weighted to obtain integrated performance index by all Key Performance Indicators, wherein weighted formula are as follows:
In formula, ητFor ΩτWeight,
In the present embodiment,
I=0.2 × Ω1+0.4×Ω2+0.4×Ω3=74；
It should be noted that when I ∈ [0,60), indicate that the comprehensive performance of tested diagnosis algorithm is general；When I ∈ [60,
85), indicate that the comprehensive performance of tested diagnosis algorithm is good；When I ∈ [85,100], the comprehensive of tested diagnosis algorithm is indicated
Close excellent performance.In the present embodiment, I=74 indicates that the comprehensive performance of tested diagnosis algorithm is good.
Above-mentioned diagnosis algorithm, the test result injected using n times random fault, takes three layers of evaluation index system, counts
Calculate and obtain basic performance indices, Key Performance Indicator (reagency, validity, sense) and assessment of integrated performance index etc.
Grade can carry out more comprehensively more objective assessment to system diagnosis algorithm to be assessed.
Embodiment 2
As interchangeable embodiment, direct fault location model can be by adjusting abort situation, fault type, fault parameter
Probability value realize Determinate test or random test.That is, if need to being determined property test when, can enableThen realize
To determinate fault position modiTest (tested at this point, need to only choose a direct fault location model, except choose failure mould
It is consistent in remaining method and above-described embodiment outside the number of type, do not repeat herein)；It enablesThen realize to certainty
Fault typeTest (being tested at this point, need to only choose a direct fault location model)；It enablesThen realize to determination
Property fault parameterTest (being tested at this point, need to only choose a direct fault location model).
Embodiment 3
The present embodiment provides a kind of diagnosis algorithm assessment systems based on random fault injection, including memory, processor
And the computer program that can be run on a memory and on a processor is stored, the processor executes the computer program
The step of Shi Shixian above method.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of diagnosis algorithm appraisal procedure based on random fault injection, which comprises the following steps:
S1: establishing direct fault location model library, and the failure of each of direct fault location model library direct fault location model is believed
Breath is layered, and the fault message includes abort situation, fault type and fault parameter；
S2: layered extraction is carried out the fault message of each direct fault location model at random according to sets requirement, is to be evaluated
Estimate diagnosis algorithm and chooses corresponding direct fault location model；
S3: diagnosis algorithm to be assessed is tested using the direct fault location model chosen in S2, and according to test result calculations
Evaluation index, and the calculated value of the evaluation index is matched with the evaluation grade of setting, it is calculated with obtaining diagnosis to be assessed
The evaluation grade of method.
2. the diagnosis algorithm appraisal procedure according to claim 1 based on random fault injection, which is characterized in that the S3
In, the evaluation index of the setting includes basic performance indices, Key Performance Indicator and integrated performance index；
The basic performance indices include: detection retardation rate, sensitivity, verification and measurement ratio, false detection rate, omission factor, abort situation identification
Rate, fault type discrimination power, fault parameter discrimination power；
The Key Performance Indicator includes reagency, validity and sense；Wherein, the reagency, the validity,
And the sense is grouped by all basic performance indices respectively and carries out combined weighted and is calculated；
The integrated performance index is calculated by all key performance indicator weightings.
3. the diagnosis algorithm appraisal procedure according to claim 2 based on random fault injection, which is characterized in that further include
Step: according to the basic performance indices construct at least three index groups, be respectively as follows: reagency index group, Validity Index group,
And sense index group；
The reagency index group includes: the detection retardation rate and the sensitivity；
The Validity Index group includes: the verification and measurement ratio, the false detection rate and the omission factor；
The sense index group includes: the abort situation discrimination power, the fault type discrimination power and failure ginseng
Number discrimination power.
4. the diagnosis algorithm appraisal procedure according to claim 1 based on random fault injection, which is characterized in that the event
Hindering the direct fault location model in injection model library is Q, calculation formula are as follows:
In formula, i indicates i-th of abort situation, i=1 ..., m, the sum for the position that may wherein break down in m expression system, l
Indicate the fault type sum of i-th of abort situation.
5. the diagnosis algorithm appraisal procedure according to claim 1 based on random fault injection, which is characterized in that the S1
In, when the fault message of each of direct fault location model library direct fault location model is layered, pass through setting
Direct fault location model f realizes the layering, wherein the calculation formula of the direct fault location model f of setting are as follows:
In formula, modiIndicate abort situation,Indicate fault type,All indicate fault parameter.
6. the diagnosis algorithm appraisal procedure according to claim 1 based on random fault injection, which is characterized in that the S2
Specifically includes the following steps:
S21: the Cumulative Distribution Function of abort situation is calculatedCalculation formula are as follows:
In formula,Indicate abort situation modiThe probability being drawn into,
S22: the abort situation extracted, calculation formula are determined are as follows:
In formula, n indicates n-th test,Indicate n-th test in order to determine that abort situation uniformly extracts from [0,1] with
Machine number, anIndicate the abort situation that n-th test is extracted, anValue range is { 1,2 ..., m }, wherein
S23: calculating abort situation is anWhen fault typeCumulative distribution functionCalculation formula are as follows:
In formula,
Randomly selecting abort situation is anWhen fault type beCalculation formula are as follows:
In formula,Indicate n-th test in order to determine random number that fault type is uniformly extracted from [0,1],Indicate n-th
The abort situation that test is extracted is anWhen fault type,Value range isWherein,
S24: calculating abort situation is anFault type isV-th of fault parameterCumulative distribution functionMeter
Calculate formula are as follows:
In formula,For v-th of fault parameterProbability density function；
(0,1) is divided into N equal part, n-th of section DnAre as follows:
N-th is tested, (0,1) is carried outIt is secondary to be uniformly distributed sampling and obtain
V-th of fault parameterIn section DnIn Probability Point are as follows:
Calculate Probability PointCorresponding fault parameter valueFormula are as follows:
S25: abort situation, fault type and the fault parameter in summary randomly selected choose the event that n-th extracts
Hinder injection model fn, calculation formula are as follows:
In formula, n=1 ..., N, N indicate to choose always secondary for assessing the experiment of whole direct fault location models needed for diagnosis algorithm
Number.
7. the diagnosis algorithm appraisal procedure according to claim 6 based on random fault injection, which is characterized in that the reality
Test the calculation formula of total degree N are as follows:
In formula, ε indicates tolerance, and ε ∈ (0,1), α indicate confidence level, and have:
In formula,Indicate the perfect estimation value of parameter, λ indicates the actual estimated value of parameter, and Pr () indicates probability.
8. the diagnosis algorithm appraisal procedure according to claim 3 based on random fault injection, which is characterized in that the S3
Specifically includes the following steps:
S31: diagnosis algorithm to be assessed is tested using the direct fault location model chosen, sets direct fault location model as fn, make an uproar
Sound is ωn, the direct fault location model f that is extracted using n-thn, n-th test, record test output are carried out to diagnosis algorithm
As a result are as follows:
In formula, rn=1 expression diagnosis algorithm decision-making system M breaks down, rnEvent does not occur for=0 expression diagnosis algorithm decision-making system M
Barrier, φnIndicate detection delay threshold value, ξnIndicate failure weak degrees threshold value, wherein failure weak degrees are by fault parameter setIt determines, is denoted asδnIndicate that fault parameter recognizes threshold value,Indicate time of failure,Indicate event at the latest
Barrier occurs to determine time, tnIt indicates the earliest time of failure that diagnosis algorithm is diagnosed to be, indicates to use diagnosis algorithm to be assessed
The abort situation of judgement Indicate the fault type determined using diagnosis algorithm to be assessed,Indicate the fault parameter set determined using diagnosis algorithm to be assessed；Wherein,
Indicate that abort situation can not recognize；Indicate that fault type can not recognize；Indicate theA failure ginseng
NumberIt can not recognize；
S32: basic performance indices, Key Performance Indicator and integrated performance index are calculated according to the test output result in S31, obtained
It obtains to the basic performance of diagnosis algorithm n times random test to be assessed, key performance and synthetic performance evaluation grade.
9. the diagnosis algorithm appraisal procedure according to claim 1 based on random fault injection, which is characterized in that the S2
In, the direct fault location model of selection includes one or at least two；When the direct fault location model of the selection is one, institute
State sets requirement are as follows: the corresponding failure probability of happening of direct fault location model is 1；
When the direct fault location model of the selection is at least two, the sets requirement are as follows: the corresponding event of direct fault location model
Hindering the sum of probability of happening is 1.
10. a kind of diagnosis algorithm assessment system based on random fault injection, including memory, processor and it is stored in storage
On device and the computer program that can run on a processor, which is characterized in that when the processor executes the computer program
The step of realizing any the method for the claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811638187.2A CN109799804B (en) | 2018-12-29 | 2018-12-29 | Diagnostic algorithm evaluation method and system based on random fault injection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811638187.2A CN109799804B (en) | 2018-12-29 | 2018-12-29 | Diagnostic algorithm evaluation method and system based on random fault injection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109799804A true CN109799804A (en) | 2019-05-24 |
| CN109799804B CN109799804B (en) | 2020-01-24 |
Family
ID=66558245
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811638187.2A Active CN109799804B (en) | 2018-12-29 | 2018-12-29 | Diagnostic algorithm evaluation method and system based on random fault injection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109799804B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111176310A (en) * | 2019-12-31 | 2020-05-19 | 北京星际荣耀空间科技有限公司 | Test method, device and system for carrier rocket attitude control system |
| CN111708649A (en) * | 2020-06-10 | 2020-09-25 | 中国工商银行股份有限公司 | Service application system attack detection method and system |
| CN112631948A (en) * | 2021-01-18 | 2021-04-09 | 北京格物仿真科技发展有限公司 | Method for evaluating intelligent algorithm grade |
| CN114625103A (en) * | 2022-02-22 | 2022-06-14 | 北京空天技术研究所 | Fault-tolerant control capability evaluation method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104390776A (en) * | 2014-12-10 | 2015-03-04 | 北京航空航天大学 | Fault detection, diagnosis and performance evaluation method for redundant aileron actuator |
| CN105044539A (en) * | 2015-06-29 | 2015-11-11 | 中南大学 | Power device IGBT open circuit fault injection method and system |
| CN105425782A (en) * | 2015-12-28 | 2016-03-23 | 中南大学 | High-speed train traction transmission control system real-time fault simulation platform |
| US9824000B1 (en) * | 2015-10-21 | 2017-11-21 | Amazon Technologies, Inc. | Testing calling code dynamically with random error injection based on user-specified configuration |
| US20180349573A1 (en) * | 2006-11-03 | 2018-12-06 | Bluerisc, Inc. | Securing microprocessors against information leakage and physical tampering |
-
2018
- 2018-12-29 CN CN201811638187.2A patent/CN109799804B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180349573A1 (en) * | 2006-11-03 | 2018-12-06 | Bluerisc, Inc. | Securing microprocessors against information leakage and physical tampering |
| CN104390776A (en) * | 2014-12-10 | 2015-03-04 | 北京航空航天大学 | Fault detection, diagnosis and performance evaluation method for redundant aileron actuator |
| CN105044539A (en) * | 2015-06-29 | 2015-11-11 | 中南大学 | Power device IGBT open circuit fault injection method and system |
| US9824000B1 (en) * | 2015-10-21 | 2017-11-21 | Amazon Technologies, Inc. | Testing calling code dynamically with random error injection based on user-specified configuration |
| CN105425782A (en) * | 2015-12-28 | 2016-03-23 | 中南大学 | High-speed train traction transmission control system real-time fault simulation platform |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111176310A (en) * | 2019-12-31 | 2020-05-19 | 北京星际荣耀空间科技有限公司 | Test method, device and system for carrier rocket attitude control system |
| CN111708649A (en) * | 2020-06-10 | 2020-09-25 | 中国工商银行股份有限公司 | Service application system attack detection method and system |
| CN111708649B (en) * | 2020-06-10 | 2024-04-09 | 中国工商银行股份有限公司 | Attack detection method and system for service application system |
| CN112631948A (en) * | 2021-01-18 | 2021-04-09 | 北京格物仿真科技发展有限公司 | Method for evaluating intelligent algorithm grade |
| CN114625103A (en) * | 2022-02-22 | 2022-06-14 | 北京空天技术研究所 | Fault-tolerant control capability evaluation method and system |
| CN114625103B (en) * | 2022-02-22 | 2024-01-05 | 北京空天技术研究所 | Fault-tolerant control capability assessment method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109799804B (en) | 2020-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109799804A (en) | A kind of diagnosis algorithm appraisal procedure and system based on random fault injection | |
| CN104764869B (en) | Transformer gas fault diagnosis and alarm method based on multidimensional characteristics | |
| CN106233115B (en) | Estimate the whether normal method of measured value of the physical parameter of aircraft engine | |
| CN109444682A (en) | The construction method of partial discharge of switchgear diagnostic system based on multi-information fusion | |
| CN103076394B (en) | Safety evaluation method for ocean platform based on integration of vibration identification frequencies and vibration mode | |
| CN110085006A (en) | Ship monitoring method, device, system and storage medium | |
| CN108039987A (en) | Critical infrastructures fragility assessment method based on multi-layer-coupled relational network | |
| CN104035431B (en) | The acquisition methods of kernel functional parameter and system for non-linear process monitoring | |
| CN105466693B (en) | The pre- diagnostic method of Fault of Diesel Fuel System based on gray model | |
| CN108680807A (en) | The Diagnosis Method of Transformer Faults and system of network are fought based on condition production | |
| CN106022366B (en) | A kind of rotating machinery method for diagnosing faults based on neighbour's evidence fusion | |
| CN104777831B (en) | A kind of method for diagnosing faults of fluid pressure type immersed pump system | |
| CN105512011B (en) | A kind of electronics testability modeling appraisal procedure | |
| CN106198749B (en) | A kind of data fusion method of the multiple sensor based on Metal Crack monitoring | |
| CN107356417B (en) | A kind of bolted joint damnification recognition method merging Time-Series analysis and comentropy | |
| CN103033648B (en) | A kind of wind sensor output data validity detection method | |
| CN107092654A (en) | Based on Change in Mean detect alarm normally with abnormal deviation data examination method and device | |
| CN108022058A (en) | A kind of wind energy conversion system state reliability estimation method | |
| CN108710637A (en) | Taxi exception track real-time detection method based on time-space relationship | |
| CN107168842A (en) | Adaptive sequential fault diagnosis method based on pmc model | |
| CN112836995A (en) | Fault activation water inrush risk evaluation method based on uncertain entropy weight method | |
| CN110058124A (en) | The intermittent fault detection method of Linear Discrete Time-delay Systems | |
| CN109902265A (en) | A kind of underground method for early warning based on hidden Markov model | |
| CN111879456B (en) | Building curtain wall safety detection method and system | |
| CN106599564A (en) | Failure monitoring and self-healing method for multi-agent combustion gas pipeline based on immune learning |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |