CN103488826B - Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model - Google Patents

Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model Download PDF

Info

Publication number
CN103488826B
CN103488826B CN201310415072.8A CN201310415072A CN103488826B CN 103488826 B CN103488826 B CN 103488826B CN 201310415072 A CN201310415072 A CN 201310415072A CN 103488826 B CN103488826 B CN 103488826B
Authority
CN
China
Prior art keywords
degradation
formula
amount
distributed constant
stress
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.)
Expired - Fee Related
Application number
CN201310415072.8A
Other languages
Chinese (zh)
Other versions
CN103488826A (en
Inventor
党香俊
姜同敏
封雷
孙富强
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.)
Beihang University
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 CN201310415072.8A priority Critical patent/CN103488826B/en
Publication of CN103488826A publication Critical patent/CN103488826A/en
Application granted granted Critical
Publication of CN103488826B publication Critical patent/CN103488826B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a kind of amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model, comprise following step: step 1, select the experience acceleration model of corresponding accelerated life test; Step 2, set up the amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation; Step 3, the amount of degradation distributed constant value of utilizing gained amount of degradation distributed constant acceleration model to extrapolate under target stress. The present invention adopts and is carried out modeling in accelerated life test field by the experience acceleration model of large quantity research and application attestation, can effectively utilize correlation test information in product research process and the Test Information of Related product, improve the credibility of result of the test. The present invention, taking model inference as basis, has avoided under Small Sample Size, and while adopting merely homing method to describe distributed constant and stress relation, accuracy is difficult to the problem ensureing.

Description

Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model
Technical field
The present invention relates to a kind of amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model, belong to accelerated degradation testTechnical field.
Background technology
In the time that suffered stress is normal working stress, the performance degradation of high reliability long life product is process very slowly. At thisUnder the condition of kind, expect to obtain enough degraded datas, certainly will will experience the longer time and pay corresponding manpower and materials cost. ForWithin a short period of time obtain the data volume that analyzing and processing needs, in laboratory environment, adopt that to improve product suffered stress harshThe method of level, the degenerative process of expedite product, be current solution high reliability long life prediction and evaluation problem focus direction itOne. Develop into and accelerate stress degradation experiment from the degradation experiment of normal applied stress, test period and relevant support expense cost obtainArrive significantly and reduced. Shorten in product development cycle facilitates and change at this, corresponding data processing also becomesObtain more difficult. The degraded data will speed up under stress is transformed under regular service conditions, is accelerate degraded data processing mainOne of difficult point. Bibliography [1] (Zhao Jianyin, Sun Quan, Peng Baohua, Zhou Jinglun. based on the reliability of accelerated degradation test dataAnalyze, electron mass, 2005 (7), 30-33) in acceleration degeneration equation concept has been proposed, suppose that amount of degradation function exists derivative,Taking the Arrhenius in reactivity model as basis, acceleration degeneration equation and logarithmic form thereof are provided. But, existing acceleration mouldIn type, except reactivity model, also has application very extensively against experience acceleration model (bibliography [2]: Huang such as power rate modelsGraceful, Jiang Tongmin. in accelerated life test, add up acceleration model summary, equipment Environmental Engineering, 2010,7 (4), 57-62). ByBe not taking reaction rate as basis in experience acceleration models such as contrary power rates, in bibliography [1], accelerate the modeling of degeneration equationJourney is not also suitable for experience acceleration model.
For this problem, taking the experience acceleration model very ripe in accelerated life test field as basis, set up reflection hereinThe amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation. In model inference process, with widely used contraryPower rate model is example, sets up corresponding numerical relationship model, and the amount of degradation distributed constant under extrapolation normal stress is convenient to accelerate to move backChange the practical application of test.
Summary of the invention
The object of the invention is the relationship modeling technical problem in order to solve amount of degradation distributed constant and stress in accelerated degradation test,A kind of amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model is proposed.
Described method comprises the steps:
(1) select the experience acceleration model of corresponding accelerated life test;
(2) set up the amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation;
(3) utilize the amount of degradation distributed constant value under gained amount of degradation distributed constant acceleration model extrapolation target stress.
The invention has the advantages that:
(1) adopt and carried out modeling in accelerated life test field by the experience acceleration model of large quantity research and application attestation, canEffectively to utilize correlation test information in product research process and the Test Information of Related product, improve the credible of result of the testProperty;
(2) avoided under Small Sample Size, while adopting merely homing method to describe distributed constant and stress relation, accuracy is difficult toThe problem ensureing.
Brief description of the drawings
Fig. 1 is method flow diagram of the present invention.
Detailed description of the invention
Below in conjunction with drawings and Examples, the present invention is described in further detail.
The present invention is a kind of amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model, flow process as shown in Figure 1, toolBody implementation method is as follows:
Step 1, select the experience acceleration model of corresponding accelerated life test;
For experience acceleration model, based on engineer, properties of product long-term observation summed up and proposed, and in practical applicationIn obtained the confirmation of great number tested data. For example contrary power rate model description the relation between voltage or pressure and life of product,Coffin-Manson model description the relation between temperature cycles stress and life of product. In hypothesis test process, product is executedThe acceleration stress adding is voltage stress, and corresponding acceleration model is selected contrary power rate model.
Contrary power rate model representation for accelerated life test is:
ξ=Av-C(1)
In formula: ξ is certain life characteristics; A is a normal number; C is a normal number relevant with activation energy; V is stress, often getsVoltage.
Step 2, set up the amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation;
For degenerating the life-span, be commonly defined as product performance index from degeneration initial time to the time that reaches defined threshold and experienceSpan, this defined threshold is the degradation failure threshold value of product, conventionally uses DfRepresent.
ξ=f(Df)-1(2)
In formula, f (.) is the universal expression formula of performance degradation function, adopts the form of implicit function herein, so that in formulation processExpression.
In fact, the determined value on degradation failure threshold value is not strict, relevant with actual demand. For example, for critical product or portionThe degeneration life-span of part, in order to reduce the risk of product failure, improve margin of safety, it is more conservative that degradation failure threshold value can be determined;Otherwise, can be decided to be more optimistic index. Degradation failure, also referred to as " soft failure ", was lost efficacy with demblee form, i.e. " hard failure "Compare, in the certainty in " life-span ", have certain difference. When degradation failure changes of threshold, " life-span " of degradation failure also withVariation. Therefore, under theoretical case, degradation failure threshold value can get arbitrarily on the occasion of, correspondingly have:
t=f(df)-1,df≥0(3)
In formula, variable dfRepresent degradation failure threshold value DfCan in degeneration value is positive scope, change continuously, variable t represents correspondingDegeneration life-span ξ be also continually varying time value.
Contrary power rate model description the relation of life of product and suffered stress, do not change with the variation of life of product. SoIn the same moment in life-span, when t=ξ, formula (1) and formula (2) can be combined into:
f(df)-1=Av-C(4)
In formula, no longer include time dependent amount. And dfIn fact be just selected as the degeneration value of " degradation failure threshold value ", withGeneral amount of degradation has the identical domain of definition, might as well replace d with general amount of degradation Mf, to make the derivation of equation more straightSee, obtain:
f(M)-1=Av-C,M=df(5)
Formula (5) is taken the logarithm:
ln(f(M)-1)=lnA-Clnv(6)
From formula (6), can find out the functional transformation ln (f (M) of general amount of degradation M-1) and the natural logrithm ln ν of stress between beLinear relationship, coefficient C wherein and lnA(or other constant terms) can obtain by the linear regression to data. Due to allThe distributed constants such as value μ, scale parameter η are the index that embodies data general characteristic, and its meaning has embodied the degeneration of " totally "Measure, belong to the category of general amount of degradation, therefore itself and the same coincidence formula of relation (5) of stress and the model of formula (6), i.e. formula (5)With formula (6) described the relation of amount of degradation distributed constant and stress, this conclusion is applicable to formula (9) and formula below too(10)。
The present invention is taking contrary power rate model as example, and the amount of degradation distributed constant acceleration model having provided based on experience acceleration model was derivedJourney. For other experience acceleration models, suppose that its life-span acceleration model has following common version:
WhereinFor the function of stress S, be the generic representation form on experience acceleration model equation right side. With derivation phase aboveWith, in the time of t=ξ, in conjunction with formula (7) and formula (3), there are formula (4) and the corresponding universal expression form of formula (5):
Model parameter to be estimated in formula (9) can, in conjunction with concrete equation form, be carried out proper transformation with reference to formula (6), by letter simultaneouslySimple regression method just can be obtained, and does not repeat them here.
Step 3, the amount of degradation distributed constant value of utilizing gained amount of degradation distributed constant acceleration model to extrapolate under target stress;
Taking contrary power rate model as example, obtain performance degenrate function f (.) according to Degradation path, application of formula (6) adopts to be accelerated under stress notWith the data of amount of degradation distributed constant under stress, return the parameters C obtaining in i moment acceleration model formulaiAnd lnAi(or itsHis constant term). Under the normal stress in any i momentExtrapolation, can be by normal stress value ν0Substitution outer apply-official nowFormula:
ln ( f ( M i 0 ) - 1 ) = ln A i - C i ln v 0 - - - ( 10 )
Solve and obtain. Herein, the extrapolation of other experience acceleration models solves identical therewith.
Embodiment 1: in the accelerated degradation test of certain product, the acceleration stress applying is voltage stress, and the 90h moment surveysObtaining the Mean Parameters that under each stress, amount of degradation distributes is: (7.5V, 0.14750), (7.0V, 0.0652), (6.7V, 0.0316). NotAffect in the situation of method validation, the magnitude of Mean Parameters has been done relevant treatment, and unit omits. The degeneration initial value of this productBe 0, Degradation path function f (.) has the form of f (t)=at. Adopt below under the extracting method extrapolation 6.5V of institute of the present invention normal stressAverage.
Step 1, select the experience acceleration model of corresponding accelerated life test;
Because the stress that product is applied is voltage stress, the contrary power rate model in experience acceleration model is applicable to describing this product longevityThe relation of life and stress, selecting contrary power rate model is accelerated life model, its expression formula is as shown in formula (1).
Step 2, set up the amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation;
The modeling process of amount of degradation distributed constant acceleration model, referring to the content in step 2 above, no longer repeats herein.
Step 3, the amount of degradation distributed constant value of utilizing gained amount of degradation distributed constant acceleration model to extrapolate under target stress;
For the data in 90h moment, by Degradation path function f (.) substitution formula (5), have
a - 1 μ 90 = A 90 v - C 90 - - - ( 11 )
After taking the logarithm, conversion obtains
lnμ90=ln(aA90)-C90lnv(12)
Through linear regression, try to achieve
C90=-13.5003,ln(aA90)=-29.0834
By ν0=6.5 substitution formula (12) obtain the average under 6.5V normal stress

Claims (1)

1. the amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model, is characterized in that, comprises followingStep:
Step 1, select the experience acceleration model of corresponding accelerated life test;
If the acceleration stress that product is applied is voltage stress, select contrary power rate model;
For the contrary power rate model representation of accelerated life test be now:
ξ=Av-C(1)
In formula: ξ is certain life characteristics; A is a normal number; C is a normal number relevant with activation energy; V is stress;Step 2, set up the amount of degradation distributed constant acceleration model of amount of degradation distributed constant and stress relation;
Use DfThe degradation failure threshold value that represents product, the life characteristics of product is:
ξ=f(Df)-1(2)
In formula, f (.) is the universal expression formula of performance degradation function, adopts the form of implicit function herein;
Under theoretical case, degradation failure threshold value get arbitrarily on the occasion of, correspondingly have:
t=f(df)-1,df≥0(3)
In formula, variable dfRepresent degradation failure threshold value DfIn degeneration value is positive scope, change continuously, variable t represents to move back accordinglyChanging life-span ξ is also continually varying time value;
Contrary power rate model description the relation of life of product and suffered stress, do not change with the variation of life of product, thusIn the same moment in life-span, when t=ξ, formula (1) and formula (2) are combined into:
f(df)-1=Av-C(4)
With general amount of degradation M replacement df, have:
f(M)-1=Av-C,M=df(5)
Formula (5) is taken the logarithm:
ln(f(M)-1)=lnA-Clnv(6)
Formula (5) and formula (6) have been described amount of degradation distributed constant and relation stress;
Suppose that life-span acceleration model has following common version:
WhereinFor the function of stress S, in the time of t=ξ, in conjunction with formula (7) and formula (3), there are formula (4) and formula (5) correspondingUniversal expression form:
Step 3, the amount of degradation distributed constant value of utilizing gained amount of degradation distributed constant acceleration model to extrapolate under target stress;
Obtain degenrate function f (.) according to Degradation path, amount of degradation distributed constant under different stress under application of formula (6) employing acceleration stressData, return and obtain parameters C in i moment acceleration model formulaiAnd lnAi, to moving back under the normal stress in any i momentChange amountExtrapolation, by normal stress value ν0Substitution extrapolation formula now:
ln ( f ( M i 0 ) - 1 ) = lnA i - C i lnv 0 - - - ( 10 )
Solve and obtain.
CN201310415072.8A 2013-09-12 2013-09-12 Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model Expired - Fee Related CN103488826B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310415072.8A CN103488826B (en) 2013-09-12 2013-09-12 Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310415072.8A CN103488826B (en) 2013-09-12 2013-09-12 Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model

Publications (2)

Publication Number Publication Date
CN103488826A CN103488826A (en) 2014-01-01
CN103488826B true CN103488826B (en) 2016-05-04

Family

ID=49829046

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310415072.8A Expired - Fee Related CN103488826B (en) 2013-09-12 2013-09-12 Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model

Country Status (1)

Country Link
CN (1) CN103488826B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104181457B (en) * 2014-08-15 2017-01-18 中国电子科技集团公司第二十四研究所 Method for selecting optimal semiconductor device temperature and humidity combined stress acceleration model
CN107515965B (en) * 2017-07-27 2019-07-23 北京航空航天大学 A kind of acceleration degeneration modelling evaluation method based on uncertain course
CN107704663A (en) * 2017-09-14 2018-02-16 中国电子科技集团公司第二十四研究所 A kind of semiconductor device temperature pulsating stress acceleration model method for optimizing

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101793927A (en) * 2010-01-12 2010-08-04 北京航空航天大学 Optimization design method of step-stress accelerated degradation test

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8664970B2 (en) * 2011-03-14 2014-03-04 Universal Display Corporation Method for accelerated lifetesting of large area OLED lighting panels

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101793927A (en) * 2010-01-12 2010-08-04 北京航空航天大学 Optimization design method of step-stress accelerated degradation test

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张恒.基于性能退化数据的可靠性建模与评估理论研究.《万方学术论文库》.2010,第1-75页. *
茆诗松.加速寿命试验的加速模型.《质量与可靠性》.2003,(第02期),第15-17页. *

Also Published As

Publication number Publication date
CN103488826A (en) 2014-01-01

Similar Documents

Publication Publication Date Title
Zhao et al. Optimal inspection and replacement policy based on experimental degradation data with covariates
CN101620045B (en) Method for evaluating reliability of stepping stress quickened degradation experiment based on time sequence
Nasiru et al. The new weibull-pareto distribution
Ranković et al. Modelling of dam behaviour based on neuro-fuzzy identification
Dimitrov Comparative analysis of methods for modelling the short‐term probability distribution of extreme wind turbine loads
CN102708306A (en) Prediction method for q-precentile life of intelligent meter
CN104316872A (en) Pressure relay storage life prediction method based on step-down degradation acceleration test
Venturini et al. Prediction reliability of a statistical methodology for gas turbine prognostics
CN103065052B (en) A kind of method measuring the engineering goods theoretical life-span
Huang et al. Time-dependent reliability model of deteriorating structures based on stochastic processes and Bayesian inference methods
Wang et al. A simulation method to estimate two types of time-varying failure rate of dynamic systems
Luo et al. Accelerated reliability demonstration under competing failure modes
CN103488826B (en) Amount of degradation distributed constant modeling Extrapolation method based on experience acceleration model
CN103530449A (en) Multivariable accelerating storage test optimized design method for lifetime piece on missile
Chiodo et al. Inverse Log-logistic distribution for Extreme Wind Speed modeling: Genesis, identification and Bayes estimation.
Wang et al. Degradation reliability modeling based on an independent increment process with quadratic variance
Kong et al. Remaining useful life prediction for degrading systems with random shocks considering measurement uncertainty
Sun et al. Accelerated degradation process analysis based on the nonlinear Wiener process with covariates and random effects
CN110728008A (en) Method and device for determining expected life of smart electric meter
Wang et al. Classical and Bayesian estimations of improved Weibull–Weibull distribution for complete and censored failure times data
Song et al. A sliding sequence importance resample filtering method for rolling bearings remaining useful life prediction based on two Wiener-process models
Zhao et al. Rolling bearing remaining useful life prediction based on wiener process
CN103630588A (en) Rapid response method of electrochemical gas sensor
Liu et al. Field reliability prediction based on degradation data and environmental data
Ma et al. Real time reliability analysis based on the performance degradation data and Bayesian method

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20160504

Termination date: 20200912

CF01 Termination of patent right due to non-payment of annual fee